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 static int match_idents(struct token *token, ...)
  37 {
  38         va_list args;
  39
  40         if (token_type(token) != TOKEN_IDENT)
  41                 return 0;
  42
  43         va_start(args, token);
  44         for (;;) {
  45                 struct ident * next = va_arg(args, struct ident *);
  46                 if (!next)
  47                         return 0;
  48                 if (token->ident == next)
  49                         return 1;
  50         }
  51 }
  52
  53
  54 struct statement *alloc_statement(struct position pos, int type)
  55 {
  56         struct statement *stmt = __alloc_statement(0);
  57         stmt->type = type;
  58         stmt->pos = pos;
  59         return stmt;
  60 }
  61
  62 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list);
  63
  64 static struct symbol * indirect(struct position pos, struct ctype *ctype, int type)
  65 {
  66         struct symbol *sym = alloc_symbol(pos, type);
  67
  68         sym->ctype.base_type = ctype->base_type;
  69         sym->ctype.modifiers = ctype->modifiers & ~MOD_STORAGE;
  70
  71         ctype->base_type = sym;
  72         ctype->modifiers &= MOD_STORAGE;
  73         return sym;
  74 }
  75
  76 static struct symbol *lookup_or_create_symbol(enum namespace ns, enum type type, struct token *token)
  77 {
  78         struct symbol *sym = lookup_symbol(token->ident, ns);
  79         if (!sym) {
  80                 sym = alloc_symbol(token->pos, type);
  81                 sym->ident = token->ident;
  82                 bind_symbol(sym, token->ident, ns);
  83                 if (type == SYM_LABEL)
  84                         fn_local_symbol(sym);
  85         }
  86         return sym;
  87 }
  88
  89 /*
  90  * NOTE! NS_LABEL is not just a different namespace,
  91  * it also ends up using function scope instead of the
  92  * regular symbol scope.
  93  */
  94 struct symbol *label_symbol(struct token *token)
  95 {
  96         return lookup_or_create_symbol(NS_LABEL, SYM_LABEL, token);
  97 }
  98
  99 struct token *struct_union_enum_specifier(enum namespace ns, enum type type,
 100         struct token *token, struct ctype *ctype,
 101         struct token *(*parse)(struct token *, struct symbol *))
 102 {
 103         struct symbol *sym;
 104
 105         ctype->modifiers = 0;
 106         if (token_type(token) == TOKEN_IDENT) {
 107                 sym = lookup_or_create_symbol(ns, type, token);
 108                 token = token->next;
 109                 ctype->base_type = sym;
 110                 if (match_op(token, '{')) {
 111                         token = parse(token->next, sym);
 112                         token = expect(token, '}', "at end of struct-union-enum-specifier");
 113                 }
 114                 return token;
 115         }
 116
 117         // private struct/union/enum type
 118         if (!match_op(token, '{')) {
 119                 warn(token->pos, "expected declaration");
 120                 ctype->base_type = &bad_type;
 121                 return token;
 122         }
 123
 124         sym = alloc_symbol(token->pos, type);
 125         token = parse(token->next, sym);
 126         ctype->base_type = sym;
 127         return expect(token, '}', "at end of specifier");
 128 }
 129
 130 static struct token *parse_struct_declaration(struct token *token, struct symbol *sym)
 131 {
 132         return struct_declaration_list(token, &sym->symbol_list);
 133 }
 134
 135 struct token *struct_or_union_specifier(enum type type, struct token *token, struct ctype *ctype)
 136 {
 137         return struct_union_enum_specifier(NS_STRUCT, type, token, ctype, parse_struct_declaration);
 138 }
 139
 140 static struct token *parse_enum_declaration(struct token *token, struct symbol *parent)
 141 {
 142         int nextval = 0;
 143         while (token_type(token) == TOKEN_IDENT) {
 144                 struct token *next = token->next;
 145                 struct symbol *sym;
 146
 147                 sym = alloc_symbol(token->pos, SYM_ENUM);
 148                 bind_symbol(sym, token->ident, NS_SYMBOL);
 149                 sym->ctype.base_type = parent;
 150                 parent->ctype.base_type = &int_ctype;
 151
 152                 if (match_op(next, '=')) {
 153                         struct expression *expr;
 154                         next = constant_expression(next->next, &expr);
 155                         nextval = get_expression_value(expr);
 156                 }
 157                 sym->value = nextval;
 158
 159                 token = next;
 160                 if (!match_op(token, ','))
 161                         break;
 162                 token = token->next;
 163                 nextval = nextval + 1;
 164         }
 165         return token;
 166 }
 167
 168 struct token *enum_specifier(struct token *token, struct ctype *ctype)
 169 {
 170         return struct_union_enum_specifier(NS_ENUM, SYM_ENUM, token, ctype, parse_enum_declaration);
 171 }
 172
 173 struct token *typeof_specifier(struct token *token, struct ctype *ctype)
 174 {
 175         struct symbol *sym;
 176
 177         if (!match_op(token, '(')) {
 178                 warn(token->pos, "expected '(' after typeof");
 179                 return token;
 180         }
 181         if (lookup_type(token->next)) {
 182                 token = typename(token->next, &sym);
 183                 *ctype = sym->ctype;
 184         } else {
 185                 struct symbol *typeof_sym = alloc_symbol(token->pos, SYM_TYPEOF);
 186                 token = parse_expression(token->next, &typeof_sym->initializer);
 187
 188                 ctype->modifiers = 0;
 189                 ctype->base_type = typeof_sym;
 190         }
 191         return expect(token, ')', "after typeof");
 192 }
 193
 194 static const char * handle_attribute(struct ctype *ctype, struct ident *attribute, struct expression *expr)
 195 {
 196         if (match_string_ident(attribute, "packed") ||
 197             match_string_ident(attribute, "__packed__")) {
 198                 ctype->alignment = 1;
 199                 return NULL;
 200         }
 201         if (match_string_ident(attribute, "aligned") ||
 202             match_string_ident(attribute, "__aligned__")) {
 203                 int alignment = max_alignment;
 204                 if (expr)
 205                         alignment = get_expression_value(expr);
 206                 ctype->alignment = alignment;
 207                 return NULL;
 208         }
 209         if (match_string_ident(attribute, "nocast")) {
 210                 ctype->modifiers |= MOD_NOCAST;
 211                 return NULL;
 212         }
 213         if (match_string_ident(attribute, "noderef")) {
 214                 ctype->modifiers |= MOD_NODEREF;
 215                 return NULL;
 216         }
 217         if (match_string_ident(attribute, "safe")) {
 218                 ctype->modifiers |= MOD_SAFE;
 219                 return NULL;
 220         }
 221         if (match_string_ident(attribute, "force")) {
 222                 ctype->modifiers |= MOD_FORCE;
 223                 return NULL;
 224         }
 225         if (match_string_ident(attribute, "address_space")) {
 226                 if (!expr)
 227                         return "expected address space number";
 228                 ctype->as = get_expression_value(expr);
 229                 return NULL;
 230         }
 231         if (match_string_ident(attribute, "context")) {
 232                 if (expr && expr->type == EXPR_COMMA) {
 233                         int mask = get_expression_value(expr->left);
 234                         int value = get_expression_value(expr->right);
 235                         if (value & ~mask)
 236                                 return "nonsense attribute types";
 237                         ctype->contextmask |= mask;
 238                         ctype->context |= value;
 239                         return NULL;
 240                 }
 241                 return "expected context mask and value";
 242         }
 243         if (match_string_ident(attribute, "mode") ||
 244             match_string_ident(attribute, "__mode__")) {
 245                 if (expr && expr->type == EXPR_SYMBOL) {
 246                         struct ident *ident = expr->symbol_name;
 247
 248                         /*
 249                          * Match against __QI__/__HI__/__SI__/__DI__
 250                          *
 251                          * FIXME! This is broken - we don't actually get
 252                          * the type information updated properly at this
 253                          * stage for some reason.
 254                          */
 255                         if (match_string_ident(ident, "__QI__") ||
 256                             match_string_ident(ident, "QI")) {
 257                                 ctype->modifiers |= MOD_SHORT;
 258                                 ctype->base_type = ctype_integer(ctype->modifiers);
 259                                 return NULL;
 260                         }
 261                         if (match_string_ident(ident, "__HI__") ||
 262                             match_string_ident(ident, "HI")) {
 263                                 ctype->modifiers |= MOD_SHORT;
 264                                 ctype->base_type = ctype_integer(ctype->modifiers);
 265                                 return NULL;
 266                         }
 267                         if (match_string_ident(ident, "__SI__") ||
 268                             match_string_ident(ident, "SI")) {
 269                                 /* Nothing? */
 270                                 return NULL;
 271                         }
 272                         if (match_string_ident(ident, "__DI__") ||
 273                             match_string_ident(ident, "DI")) {
 274                                 ctype->modifiers |= MOD_LONGLONG;
 275                                 ctype->base_type = ctype_integer(ctype->modifiers);
 276                                 return NULL;
 277                         }
 278                         return "unknown mode attribute";
 279                 }
 280                 return "expected attribute mode symbol";
 281         }
 282
 283         /* Throw away for now.. */
 284         if (match_string_ident(attribute, "format") ||
 285             match_string_ident(attribute, "__format__"))
 286                 return NULL;
 287         if (match_string_ident(attribute, "section") ||
 288             match_string_ident(attribute, "__section__"))
 289                 return NULL;
 290         if (match_string_ident(attribute, "unused") ||
 291             match_string_ident(attribute, "__unused__"))
 292                 return NULL;
 293         if (match_string_ident(attribute, "const") ||
 294             match_string_ident(attribute, "__const__"))
 295                 return NULL;
 296         if (match_string_ident(attribute, "noreturn"))
 297                 return NULL;
 298         if (match_string_ident(attribute, "regparm"))
 299                 return NULL;
 300         if (match_string_ident(attribute, "weak"))
 301                 return NULL;
 302         if (match_string_ident(attribute, "alias"))
 303                 return NULL;
 304         if (match_string_ident(attribute, "pure"))
 305                 return NULL;
 306         if (match_string_ident(attribute, "always_inline"))
 307                 return NULL;
 308         if (match_string_ident(attribute, "syscall_linkage"))
 309                 return NULL;
 310
 311         return "unknown attribute";
 312 }
 313
 314 struct token *attribute_specifier(struct token *token, struct ctype *ctype)
 315 {
 316         ctype->modifiers = 0;
 317         token = expect(token, '(', "after attribute");
 318         token = expect(token, '(', "after attribute");
 319
 320         for (;;) {
 321                 const char *error;
 322                 struct ident *attribute_name;
 323                 struct expression *attribute_expr;
 324
 325                 if (eof_token(token))
 326                         break;
 327                 if (match_op(token, ';'))
 328                         break;
 329                 if (token_type(token) != TOKEN_IDENT)
 330                         break;
 331                 attribute_name = token->ident;
 332                 token = token->next;
 333                 attribute_expr = NULL;
 334                 if (match_op(token, '('))
 335                         token = parens_expression(token, &attribute_expr, "in attribute");
 336                 error = handle_attribute(ctype, attribute_name, attribute_expr);
 337                 if (error)
 338                         warn(token->pos, "attribute '%s': %s", show_ident(attribute_name), error);
 339                 if (!match_op(token, ','))
 340                         break;
 341                 token = token->next;
 342         }
 343
 344         token = expect(token, ')', "after attribute");
 345         token = expect(token, ')', "after attribute");
 346         return token;
 347 }
 348
 349 #define MOD_SPECIALBITS (MOD_STRUCTOF | MOD_UNIONOF | MOD_ENUMOF | MOD_ATTRIBUTE | MOD_TYPEOF)
 350 #define MOD_SPECIFIER (MOD_CHAR | MOD_SHORT | MOD_LONG | MOD_LONGLONG | MOD_SIGNED | MOD_UNSIGNED)
 351
 352 struct symbol * ctype_integer(unsigned int spec)
 353 {
 354         static struct symbol *const integer_ctypes[][2] = {
 355                 { &llong_ctype, &ullong_ctype },
 356                 { &long_ctype,  &ulong_ctype  },
 357                 { &short_ctype, &ushort_ctype },
 358                 { &char_ctype,  &uchar_ctype  },
 359                 { &int_ctype,   &uint_ctype   },
 360         };
 361         struct symbol *const (*ctype)[2];
 362
 363         ctype = integer_ctypes;
 364         if (!(spec & MOD_LONGLONG)) {
 365                 ctype++;
 366                 if (!(spec & MOD_LONG)) {
 367                         ctype++;
 368                         if (!(spec & MOD_SHORT)) {
 369                                 ctype++;
 370                                 if (!(spec & MOD_CHAR))
 371                                         ctype++;
 372                         }
 373                 }
 374         }
 375         return ctype[0][(spec & MOD_UNSIGNED) != 0];
 376 }
 377
 378 struct symbol * ctype_fp(unsigned int spec)
 379 {
 380         if (spec & MOD_LONGLONG)
 381                 return &ldouble_ctype;
 382         if (spec & MOD_LONG)
 383                 return &double_ctype;
 384         return &float_ctype;
 385 }
 386
 387 static void apply_ctype(struct position pos, struct ctype *thistype, struct ctype *ctype)
 388 {
 389         unsigned long mod = thistype->modifiers;
 390
 391         if (mod) {
 392                 unsigned long old = ctype->modifiers;
 393                 unsigned long extra = 0, dup;
 394
 395                 if (mod & old & MOD_LONG) {
 396                         extra = MOD_LONGLONG | MOD_LONG;
 397                         mod &= ~MOD_LONG;
 398                         old &= ~MOD_LONG;
 399                 }
 400                 dup = (mod & old) | (extra & old) | (extra & mod);
 401                 if (dup)
 402                         warn(pos, "Just how %sdo you want this type to be?",
 403                                 modifier_string(dup));
 404                 ctype->modifiers = old | mod | extra;
 405         }
 406
 407         /* Context mask and value */
 408         if ((ctype->context ^ thistype->context) & (ctype->contextmask & thistype->contextmask)) {
 409                 warn(pos, "inconsistend attribute types");
 410                 thistype->context = 0;
 411                 thistype->contextmask = 0;
 412         }
 413         ctype->context |= thistype->context;
 414         ctype->contextmask |= thistype->contextmask;
 415
 416         /* Alignment */
 417         if (thistype->alignment & (thistype->alignment-1)) {
 418                 warn(pos, "I don't like non-power-of-2 alignments");
 419                 thistype->alignment = 0;
 420         }
 421         if (thistype->alignment > ctype->alignment)
 422                 ctype->alignment = thistype->alignment;
 423
 424         /* Address space */
 425         ctype->as = thistype->as;
 426 }
 427
 428
 429 static struct token *declaration_specifiers(struct token *next, struct ctype *ctype, int qual)
 430 {
 431         struct token *token;
 432
 433         while ( (token = next) != NULL ) {
 434                 struct ctype thistype;
 435                 struct ident *ident;
 436                 struct symbol *s, *type;
 437                 unsigned long mod;
 438
 439                 next = token->next;
 440                 if (token_type(token) != TOKEN_IDENT)
 441                         break;
 442                 ident = token->ident;
 443
 444                 s = lookup_symbol(ident, NS_TYPEDEF);
 445                 if (!s)
 446                         break;
 447                 thistype = s->ctype;
 448                 mod = thistype.modifiers;
 449                 if (qual && (mod & ~(MOD_ATTRIBUTE | MOD_CONST | MOD_VOLATILE)))
 450                         break;
 451                 if (mod & MOD_SPECIALBITS) {
 452                         if (mod & MOD_STRUCTOF)
 453                                 next = struct_or_union_specifier(SYM_STRUCT, next, &thistype);
 454                         else if (mod & MOD_UNIONOF)
 455                                 next = struct_or_union_specifier(SYM_UNION, next, &thistype);
 456                         else if (mod & MOD_ENUMOF)
 457                                 next = enum_specifier(next, &thistype);
 458                         else if (mod & MOD_ATTRIBUTE)
 459                                 next = attribute_specifier(next, &thistype);
 460                         else if (mod & MOD_TYPEOF)
 461                                 next = typeof_specifier(next, &thistype);
 462                         mod = thistype.modifiers;
 463                 }
 464                 type = thistype.base_type;
 465                 if (type) {
 466                         if (qual)
 467                                 break;
 468                         if (ctype->base_type)
 469                                 break;
 470                         /* User types only mix with qualifiers */
 471                         if (mod & MOD_USERTYPE) {
 472                                 if (ctype->modifiers & MOD_SPECIFIER)
 473                                         break;
 474                         }
 475                         ctype->base_type = type;
 476                 }
 477
 478                 apply_ctype(token->pos, &thistype, ctype);
 479         }
 480
 481         /* Turn the "virtual types" into real types with real sizes etc */
 482         if (!ctype->base_type) {
 483                 struct symbol *base = &incomplete_ctype;
 484
 485                 /*
 486                  * If we have modifiers, we'll default to an integer
 487                  * type, and "ctype_integer()" will turn this into
 488                  * a specific one.
 489                  */
 490                 if (ctype->modifiers & MOD_SPECIFIER)
 491                         base = &int_type;
 492                 ctype->base_type = base;
 493         }
 494
 495         if (ctype->base_type == &int_type) {
 496                 ctype->base_type = ctype_integer(ctype->modifiers & MOD_SPECIFIER);
 497                 ctype->modifiers &= ~MOD_SPECIFIER;
 498                 return token;
 499         }
 500         if (ctype->base_type == &fp_type) {
 501                 ctype->base_type = ctype_fp(ctype->modifiers & MOD_SPECIFIER);
 502                 ctype->modifiers &= ~MOD_SPECIFIER;
 503                 return token;
 504         }
 505         return token;
 506 }
 507
 508 static struct token *abstract_array_declarator(struct token *token, struct symbol *sym)
 509 {
 510         struct expression *expr = NULL;
 511
 512         token = parse_expression(token, &expr);
 513         sym->array_size = expr;
 514         return token;
 515 }
 516
 517 static struct token *parameter_type_list(struct token *, struct symbol *);
 518 static struct token *declarator(struct token *token, struct symbol **tree, struct ident **p);
 519
 520 static struct token *direct_declarator(struct token *token, struct symbol **tree, struct ident **p)
 521 {
 522         struct ctype *ctype = &(*tree)->ctype;
 523
 524         if (p && token_type(token) == TOKEN_IDENT) {
 525                 *p = token->ident;
 526                 token = token->next;
 527         }
 528
 529         for (;;) {
 530                 if (match_idents(token, &__attribute___ident, &__attribute_ident, NULL)) {
 531                         struct ctype thistype = { 0, };
 532                         token = attribute_specifier(token->next, &thistype);
 533                         apply_ctype(token->pos, &thistype, ctype);
 534                         continue;
 535                 }
 536                 if (token_type(token) != TOKEN_SPECIAL)
 537                         return token;
 538
 539                 /*
 540                  * This can be either a parameter list or a grouping.
 541                  * For the direct (non-abstract) case, we know if must be
 542                  * a paramter list if we already saw the identifier.
 543                  * For the abstract case, we know if must be a parameter
 544                  * list if it is empty or starts with a type.
 545                  */
 546                 if (token->special == '(') {
 547                         struct symbol *sym;
 548                         struct token *next = token->next;
 549                         int fn = (p && *p) || match_op(next, ')') || lookup_type(next);
 550
 551                         if (!fn) {
 552                                 struct symbol *base_type = ctype->base_type;
 553                                 token = declarator(next, tree, p);
 554                                 token = expect(token, ')', "in nested declarator");
 555                                 while (ctype->base_type != base_type)
 556                                         ctype = &ctype->base_type->ctype;
 557                                 p = NULL;
 558                                 continue;
 559                         }
 560
 561                         sym = indirect(token->pos, ctype, SYM_FN);
 562                         token = parameter_type_list(next, sym);
 563                         token = expect(token, ')', "in function declarator");
 564                         continue;
 565                 }
 566                 if (token->special == '[') {
 567                         struct symbol *array = indirect(token->pos, ctype, SYM_ARRAY);
 568                         token = abstract_array_declarator(token->next, array);
 569                         token = expect(token, ']', "in abstract_array_declarator");
 570                         ctype = &array->ctype;
 571                         continue;
 572                 }
 573                 if (token->special == ':') {
 574                         struct symbol *bitfield;
 575                         struct expression *expr;
 576                         bitfield = indirect(token->pos, ctype, SYM_BITFIELD);
 577                         token = conditional_expression(token->next, &expr);
 578                         bitfield->fieldwidth = get_expression_value(expr);
 579                         continue;
 580                 }
 581                 break;
 582         }
 583         if (p) {
 584                 (*tree)->ident = *p;
 585         }
 586         return token;
 587 }
 588
 589 static struct token *pointer(struct token *token, struct ctype *ctype)
 590 {
 591         unsigned long modifiers;
 592         struct symbol *base_type;
 593
 594         modifiers = ctype->modifiers & ~(MOD_TYPEDEF | MOD_ATTRIBUTE);
 595         base_type = ctype->base_type;
 596         ctype->modifiers = modifiers;
 597
 598         while (match_op(token,'*')) {
 599                 struct symbol *ptr = alloc_symbol(token->pos, SYM_PTR);
 600                 ptr->ctype.modifiers = modifiers & ~MOD_STORAGE;
 601                 ptr->ctype.as = ctype->as;
 602                 ptr->ctype.context = ctype->context;
 603                 ptr->ctype.contextmask = ctype->contextmask;
 604                 ptr->ctype.base_type = base_type;
 605
 606                 base_type = ptr;
 607                 ctype->modifiers = modifiers & MOD_STORAGE;
 608                 ctype->base_type = base_type;
 609                 ctype->as = 0;
 610                 ctype->context = 0;
 611                 ctype->contextmask = 0;
 612
 613                 token = declaration_specifiers(token->next, ctype, 1);
 614                 modifiers = ctype->modifiers;
 615         }
 616         return token;
 617 }
 618
 619 static struct token *declarator(struct token *token, struct symbol **tree, struct ident **p)
 620 {
 621         token = pointer(token, &(*tree)->ctype);
 622         return direct_declarator(token, tree, p);
 623 }
 624
 625 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list)
 626 {
 627         while (!match_op(token, '}')) {
 628                 struct ctype ctype = {0, };
 629
 630                 token = declaration_specifiers(token, &ctype, 0);
 631                 for (;;) {
 632                         struct ident *ident = NULL;
 633                         struct symbol *decl = alloc_symbol(token->pos, SYM_NODE);
 634                         decl->ctype = ctype;
 635                         token = declarator(token, &decl, &ident);
 636                         if (match_op(token, ':')) {
 637                                 struct expression *expr;
 638                                 token = parse_expression(token->next, &expr);
 639                         }
 640                         add_symbol(list, decl);
 641                         if (!match_op(token, ','))
 642                                 break;
 643                         token = token->next;
 644                 }
 645                 if (!match_op(token, ';'))
 646                         break;
 647                 token = token->next;
 648         }
 649         return token;
 650 }
 651
 652 static struct token *parameter_declaration(struct token *token, struct symbol **tree)
 653 {
 654         struct ident *ident = NULL;
 655         struct symbol *sym;
 656         struct ctype ctype = { 0, };
 657
 658         token = declaration_specifiers(token, &ctype, 0);
 659         sym = alloc_symbol(token->pos, SYM_NODE);
 660         sym->ctype = ctype;
 661         *tree = sym;
 662         token = declarator(token, tree, &ident);
 663         return token;
 664 }
 665
 666 struct token *typename(struct token *token, struct symbol **p)
 667 {
 668         struct symbol *sym = alloc_symbol(token->pos, SYM_NODE);
 669         *p = sym;
 670         token = declaration_specifiers(token, &sym->ctype, 0);
 671         return declarator(token, &sym, NULL);
 672 }
 673
 674 struct token *expression_statement(struct token *token, struct expression **tree)
 675 {
 676         token = parse_expression(token, tree);
 677         return expect(token, ';', "at end of statement");
 678 }
 679
 680 static struct token *parse_asm_operands(struct token *token, struct statement *stmt)
 681 {
 682         struct expression *expr;
 683
 684         /* Allow empty operands */
 685         if (match_op(token->next, ':') || match_op(token->next, ')'))
 686                 return token->next;
 687         do {
 688                 token = primary_expression(token->next, &expr);
 689                 token = parens_expression(token, &expr, "in asm parameter");
 690         } while (match_op(token, ','));
 691         return token;
 692 }
 693
 694 static struct token *parse_asm_clobbers(struct token *token, struct statement *stmt)
 695 {
 696         struct expression *expr;
 697
 698         do {
 699                 token = primary_expression(token->next, &expr);
 700         } while (match_op(token, ','));
 701         return token;
 702 }
 703
 704 static struct token *parse_asm(struct token *token, struct statement *stmt)
 705 {
 706         struct expression *expr;
 707
 708         stmt->type = STMT_ASM;
 709         if (match_idents(token, &__volatile___ident, &volatile_ident)) {
 710                 token = token->next;
 711         }
 712         token = expect(token, '(', "after asm");
 713         token = parse_expression(token->next, &expr);
 714         if (match_op(token, ':'))
 715                 token = parse_asm_operands(token, stmt);
 716         if (match_op(token, ':'))
 717                 token = parse_asm_operands(token, stmt);
 718         if (match_op(token, ':'))
 719                 token = parse_asm_clobbers(token, stmt);
 720         token = expect(token, ')', "after asm");
 721         return expect(token, ';', "at end of asm-statement");
 722 }
 723
 724 /* Make a statement out of an expression */
 725 static struct statement *make_statement(struct expression *expr)
 726 {
 727         struct statement *stmt;
 728
 729         if (!expr)
 730                 return NULL;
 731         stmt = alloc_statement(expr->pos, STMT_EXPRESSION);
 732         stmt->expression = expr;
 733         return stmt;
 734 }
 735
 736 /*
 737  * All iterators have two symbols associated with them:
 738  * the "continue" and "break" symbols, which are targets
 739  * for continue and break statements respectively.
 740  *
 741  * They are in a special name-space, but they follow
 742  * all the normal visibility rules, so nested iterators
 743  * automatically work right.
 744  */
 745 static void start_iterator(struct statement *stmt)
 746 {
 747         struct symbol *cont, *brk;
 748
 749         start_symbol_scope();
 750         cont = alloc_symbol(stmt->pos, SYM_NODE);
 751         cont->ident = &continue_ident;
 752         bind_symbol(cont, &continue_ident, NS_ITERATOR);
 753         brk = alloc_symbol(stmt->pos, SYM_NODE);
 754         brk->ident = &break_ident;
 755         bind_symbol(brk, &break_ident, NS_ITERATOR);
 756
 757         stmt->type = STMT_ITERATOR;
 758         stmt->iterator_break = brk;
 759         stmt->iterator_continue = cont;
 760         fn_local_symbol(brk);
 761         fn_local_symbol(cont);
 762 }
 763
 764 static void end_iterator(struct statement *stmt)
 765 {
 766         end_symbol_scope();
 767 }
 768
 769 static struct statement *start_function(struct symbol *sym)
 770 {
 771         struct symbol *ret;
 772         struct statement *stmt = alloc_statement(sym->pos, STMT_COMPOUND);
 773
 774         start_function_scope();
 775         ret = alloc_symbol(sym->pos, SYM_NODE);
 776         ret->ident = &return_ident;
 777         ret->ctype = sym->ctype.base_type->ctype;
 778         ret->ctype.modifiers &= ~(MOD_STORAGE | MOD_CONST | MOD_VOLATILE | MOD_INLINE | MOD_ADDRESSABLE | MOD_NOCAST | MOD_NODEREF | MOD_ACCESSED | MOD_TOPLEVEL);
 779         ret->ctype.modifiers |= (MOD_AUTO | MOD_REGISTER);
 780         bind_symbol(ret, &return_ident, NS_ITERATOR);
 781         stmt->ret = ret;
 782
 783         fn_local_symbol(ret);
 784         return stmt;
 785 }
 786
 787 static void end_function(struct symbol *sym)
 788 {
 789         end_function_scope();
 790 }
 791
 792 /*
 793  * A "switch()" statement, like an iterator, has a
 794  * the "break" symbol associated with it. It works
 795  * exactly like the iterator break - it's the target
 796  * for any break-statements in scope, and means that
 797  * "break" handling doesn't even need to know whether
 798  * it's breaking out of an iterator or a switch.
 799  *
 800  * In addition, the "case" symbol is a marker for the
 801  * case/default statements to find the switch statement
 802  * that they are associated with.
 803  */
 804 static void start_switch(struct statement *stmt)
 805 {
 806         struct symbol *brk, *switch_case;
 807
 808         start_symbol_scope();
 809         brk = alloc_symbol(stmt->pos, SYM_NODE);
 810         brk->ident = &break_ident;
 811         bind_symbol(brk, &break_ident, NS_ITERATOR);
 812
 813         switch_case = alloc_symbol(stmt->pos, SYM_NODE);
 814         switch_case->ident = &case_ident;
 815         bind_symbol(switch_case, &case_ident, NS_ITERATOR);
 816         switch_case->stmt = stmt;
 817
 818         stmt->type = STMT_SWITCH;
 819         stmt->switch_break = brk;
 820         stmt->switch_case = switch_case;
 821
 822         fn_local_symbol(brk);
 823         fn_local_symbol(switch_case);
 824 }
 825
 826 static void end_switch(struct statement *stmt)
 827 {
 828         if (!stmt->switch_case->symbol_list)
 829                 warn(stmt->pos, "switch with no cases");
 830         end_symbol_scope();
 831 }
 832
 833 static void add_case_statement(struct statement *stmt)
 834 {
 835         struct symbol *target = lookup_symbol(&case_ident, NS_ITERATOR);
 836         struct symbol *sym;
 837
 838         if (!target) {
 839                 warn(stmt->pos, "not in switch scope");
 840                 return;
 841         }
 842         sym = alloc_symbol(stmt->pos, SYM_NODE);
 843         add_symbol(&target->symbol_list, sym);
 844         sym->stmt = stmt;
 845         stmt->case_label = sym;
 846         fn_local_symbol(sym);
 847 }
 848
 849 static struct token *parse_return_statement(struct token *token, struct statement *stmt)
 850 {
 851         struct symbol *target = lookup_symbol(&return_ident, NS_ITERATOR);
 852
 853         if (!target)
 854                 error(token->pos, "internal error: return without a function target");
 855         stmt->type = STMT_RETURN;
 856         stmt->ret_target = target;
 857         return expression_statement(token->next, &stmt->ret_value);
 858 }
 859
 860 static struct token *parse_for_statement(struct token *token, struct statement *stmt)
 861 {
 862         struct symbol_list *syms;
 863         struct expression *e1, *e2, *e3;
 864         struct statement *iterator;
 865
 866         start_iterator(stmt);
 867         token = expect(token->next, '(', "after 'for'");
 868
 869         syms = NULL;
 870         e1 = NULL;
 871         /* C99 variable declaration? */
 872         if (lookup_type(token)) {
 873                 token = external_declaration(token, &syms);
 874         } else {
 875                 token = parse_expression(token, &e1);
 876                 token = expect(token, ';', "in 'for'");
 877         }
 878         token = parse_expression(token, &e2);
 879         token = expect(token, ';', "in 'for'");
 880         token = parse_expression(token, &e3);
 881         token = expect(token, ')', "in 'for'");
 882         token = statement(token, &iterator);
 883
 884         stmt->iterator_syms = syms;
 885         stmt->iterator_pre_statement = make_statement(e1);
 886         stmt->iterator_pre_condition = e2;
 887         stmt->iterator_post_statement = make_statement(e3);
 888         stmt->iterator_post_condition = e2;
 889         stmt->iterator_statement = iterator;
 890         end_iterator(stmt);
 891
 892         return token;
 893 }
 894
 895 struct token *parse_while_statement(struct token *token, struct statement *stmt)
 896 {
 897         struct expression *expr;
 898         struct statement *iterator;
 899
 900         start_iterator(stmt);
 901         token = parens_expression(token->next, &expr, "after 'while'");
 902         token = statement(token, &iterator);
 903
 904         stmt->iterator_pre_condition = expr;
 905         stmt->iterator_post_condition = expr;
 906         stmt->iterator_statement = iterator;
 907         end_iterator(stmt);
 908
 909         return token;
 910 }
 911
 912 struct token *parse_do_statement(struct token *token, struct statement *stmt)
 913 {
 914         struct expression *expr;
 915         struct statement *iterator;
 916
 917         start_iterator(stmt);
 918         token = statement(token->next, &iterator);
 919         if (token_type(token) == TOKEN_IDENT && token->ident == &while_ident)
 920                 token = token->next;
 921         else
 922                 warn(token->pos, "expected 'while' after 'do'");
 923         token = parens_expression(token, &expr, "after 'do-while'");
 924
 925         stmt->iterator_post_condition = expr;
 926         stmt->iterator_statement = iterator;
 927         end_iterator(stmt);
 928
 929         return expect(token, ';', "after statement");
 930 }
 931
 932 static struct token *statement(struct token *token, struct statement **tree)
 933 {
 934         struct statement *stmt = alloc_statement(token->pos, STMT_NONE);
 935
 936         *tree = stmt;
 937         if (token_type(token) == TOKEN_IDENT) {
 938                 if (token->ident == &if_ident) {
 939                         stmt->type = STMT_IF;
 940                         token = parens_expression(token->next, &stmt->if_conditional, "after if");
 941                         token = statement(token, &stmt->if_true);
 942                         if (token_type(token) != TOKEN_IDENT)
 943                                 return token;
 944                         if (token->ident != &else_ident)
 945                                 return token;
 946                         return statement(token->next, &stmt->if_false);
 947                 }
 948
 949                 if (token->ident == &return_ident)
 950                         return parse_return_statement(token, stmt);
 951
 952                 if (token->ident == &break_ident || token->ident == &continue_ident) {
 953                         struct symbol *target = lookup_symbol(token->ident, NS_ITERATOR);
 954                         stmt->type = STMT_GOTO;
 955                         stmt->goto_label = target;
 956                         if (!target)
 957                                 warn(stmt->pos, "break/continue not in iterator scope");
 958                         return expect(token->next, ';', "at end of statement");
 959                 }
 960                 if (token->ident == &default_ident) {
 961                         token = token->next;
 962                         goto default_statement;
 963                 }
 964                 if (token->ident == &case_ident) {
 965                         token = parse_expression(token->next, &stmt->case_expression);
 966                         if (match_op(token, SPECIAL_ELLIPSIS))
 967                                 token = parse_expression(token->next, &stmt->case_to);
 968 default_statement:
 969                         stmt->type = STMT_CASE;
 970                         token = expect(token, ':', "after default/case");
 971                         add_case_statement(stmt);
 972                         return statement(token, &stmt->case_statement);
 973                 }
 974                 if (token->ident == &switch_ident) {
 975                         stmt->type = STMT_SWITCH;
 976                         start_switch(stmt);
 977                         token = parens_expression(token->next, &stmt->switch_expression, "after 'switch'");
 978                         token = statement(token, &stmt->switch_statement);
 979                         end_switch(stmt);
 980                         return token;
 981                 }
 982                 if (token->ident == &for_ident)
 983                         return parse_for_statement(token, stmt);
 984
 985                 if (token->ident == &while_ident)
 986                         return parse_while_statement(token, stmt);
 987
 988                 if (token->ident == &do_ident)
 989                         return parse_do_statement(token, stmt);
 990
 991                 if (token->ident == &goto_ident) {
 992                         stmt->type = STMT_GOTO;
 993                         token = token->next;
 994                         if (match_op(token, '*')) {
 995                                 token = parse_expression(token->next, &stmt->goto_expression);
 996                         } else if (token_type(token) == TOKEN_IDENT) {
 997                                 stmt->goto_label = label_symbol(token);
 998                                 token = token->next;
 999                         } else {
1000                                 warn(token->pos, "Expected identifier or goto expression");
1001                         }
1002                         return expect(token, ';', "at end of statement");
1003                 }
1004                 if (match_idents(token, &asm_ident, &__asm___ident, &__asm_ident, NULL)) {
1005                         return parse_asm(token->next, stmt);
1006                 }
1007                 if (match_op(token->next, ':')) {
1008                         stmt->type = STMT_LABEL;
1009                         stmt->label_identifier = label_symbol(token);
1010                         return statement(token->next->next, &stmt->label_statement);
1011                 }
1012         }
1013
1014         if (match_op(token, '{')) {
1015                 stmt->type = STMT_COMPOUND;
1016                 start_symbol_scope();
1017                 token = compound_statement(token->next, stmt);
1018                 end_symbol_scope();
1019
1020                 return expect(token, '}', "at end of compound statement");
1021         }
1022
1023         stmt->type = STMT_EXPRESSION;
1024         return expression_statement(token, &stmt->expression);
1025 }
1026
1027 struct token * statement_list(struct token *token, struct statement_list **list)
1028 {
1029         for (;;) {
1030                 struct statement * stmt;
1031                 if (eof_token(token))
1032                         break;
1033                 if (match_op(token, '}'))
1034                         break;
1035                 token = statement(token, &stmt);
1036                 add_statement(list, stmt);
1037         }
1038         return token;
1039 }
1040
1041 static struct token *parameter_type_list(struct token *token, struct symbol *fn)
1042 {
1043         struct symbol_list **list = &fn->arguments;
1044         for (;;) {
1045                 struct symbol *sym = alloc_symbol(token->pos, SYM_NODE);
1046
1047                 if (match_op(token, SPECIAL_ELLIPSIS)) {
1048                         fn->variadic = 1;
1049                         token = token->next;
1050                         break;
1051                 }
1052
1053                 if (!lookup_type(token)) {
1054                         warn(token->pos, "non-ANSI parameter list");
1055                         break;
1056                 }
1057                 token = parameter_declaration(token, &sym);
1058                 /* Special case: (void) */
1059                 if (!*list && !sym->ident && sym->ctype.base_type == &void_ctype)
1060                         break;
1061                 add_symbol(list, sym);
1062                 if (!match_op(token, ','))
1063                         break;
1064                 token = token->next;
1065
1066         }
1067         return token;
1068 }
1069
1070 struct token *compound_statement(struct token *token, struct statement *stmt)
1071 {
1072         while (!eof_token(token)) {
1073                 if (!lookup_type(token))
1074                         break;
1075                 token = external_declaration(token, &stmt->syms);
1076         }
1077         token = statement_list(token, &stmt->stmts);
1078         return token;
1079 }
1080
1081 static struct expression *identifier_expression(struct token *token)
1082 {
1083         struct expression *expr = alloc_expression(token->pos, EXPR_IDENTIFIER);
1084         expr->expr_ident = token->ident;
1085         return expr;
1086 }
1087
1088 static struct expression *index_expression(struct expression *from, struct expression *to)
1089 {
1090         int idx_from, idx_to;
1091         struct expression *expr = alloc_expression(from->pos, EXPR_INDEX);
1092
1093         idx_from = get_expression_value(from);
1094         idx_to = idx_from;
1095         if (to) {
1096                 idx_to = get_expression_value(to);
1097                 if (idx_to < idx_from || idx_from < 0)
1098                         warn(from->pos, "nonsense array initializer index range");
1099         }
1100         expr->idx_from = idx_from;
1101         expr->idx_to = idx_to;
1102         return expr;
1103 }
1104
1105 static struct token *initializer_list(struct expression_list **list, struct token *token)
1106 {
1107         for (;;) {
1108                 struct token *next = token->next;
1109                 struct expression *expr;
1110
1111                 if (match_op(token, '.') && (token_type(next) == TOKEN_IDENT) && match_op(next->next, '=')) {
1112                         add_expression(list, identifier_expression(next));
1113                         token = next->next->next;
1114                 } else if ((token_type(token) == TOKEN_IDENT) && match_op(next, ':')) {
1115                         add_expression(list, identifier_expression(token));
1116                         token = next->next;
1117                 } else if (match_op(token, '[')) {
1118                         struct expression *from = NULL, *to = NULL;
1119                         token = constant_expression(token->next, &from);
1120                         if (match_op(token, SPECIAL_ELLIPSIS))
1121                                 token = constant_expression(token->next, &to);
1122                         add_expression(list, index_expression(from, to));
1123                         token = expect(token, ']', "at end of initializer index");
1124                         token = expect(token, '=', "at end of initializer index");
1125                 }
1126
1127                 expr = NULL;
1128                 token = initializer(&expr, token);
1129                 if (!expr)
1130                         break;
1131                 add_expression(list, expr);
1132                 if (!match_op(token, ','))
1133                         break;
1134                 token = token->next;
1135         }
1136         return token;
1137 }
1138
1139 struct token *initializer(struct expression **tree, struct token *token)
1140 {
1141         if (match_op(token, '{')) {
1142                 struct expression *expr = alloc_expression(token->pos, EXPR_INITIALIZER);
1143                 *tree = expr;
1144                 token = initializer_list(&expr->expr_list, token->next);
1145                 return expect(token, '}', "at end of initializer");
1146         }
1147         return assignment_expression(token, tree);
1148 }
1149
1150 static void declare_argument(struct symbol *sym, struct symbol *fn)
1151 {
1152         if (!sym->ident) {
1153                 warn(sym->pos, "no identifier for function argument");
1154                 return;
1155         }
1156         bind_symbol(sym, sym->ident, NS_SYMBOL);
1157 }
1158
1159 static struct token *parse_function_body(struct token *token, struct symbol *decl,
1160         struct symbol_list **list)
1161 {
1162         struct symbol *base_type = decl->ctype.base_type;
1163         struct statement *stmt;
1164         struct symbol *arg;
1165
1166         function_symbol_list = &decl->symbol_list;
1167         if (decl->ctype.modifiers & MOD_EXTERN) {
1168                 if (!(decl->ctype.modifiers & MOD_INLINE))
1169                         warn(decl->pos, "function with external linkage has definition");
1170         }
1171         if (!(decl->ctype.modifiers & MOD_STATIC))
1172                 decl->ctype.modifiers |= MOD_EXTERN;
1173
1174         stmt = start_function(decl);
1175
1176         base_type->stmt = stmt;
1177         FOR_EACH_PTR (base_type->arguments, arg) {
1178                 declare_argument(arg, base_type);
1179         } END_FOR_EACH_PTR;
1180
1181         token = compound_statement(token->next, stmt);
1182
1183         end_function(decl);
1184         if (!(decl->ctype.modifiers & MOD_INLINE))
1185                 add_symbol(list, decl);
1186         check_declaration(decl);
1187         function_symbol_list = NULL;
1188         return expect(token, '}', "at end of function");
1189 }
1190
1191 static struct token *external_declaration(struct token *token, struct symbol_list **list)
1192 {
1193         struct ident *ident = NULL;
1194         struct symbol *decl;
1195         struct ctype ctype = { 0, };
1196         struct symbol *base_type;
1197         int is_typedef;
1198
1199         /* Top-level inline asm? */
1200         if (match_idents(token, &asm_ident, &__asm___ident, &__asm_ident)) {
1201                 struct symbol *anon = alloc_symbol(token->pos, SYM_NODE);
1202                 struct symbol *fn = alloc_symbol(token->pos, SYM_FN);
1203                 struct statement *stmt;
1204
1205                 anon->ctype.base_type = fn;
1206                 function_symbol_list = &anon->symbol_list;
1207                 stmt = start_function(anon);
1208                 token = parse_asm(token->next, stmt);
1209                 end_function(anon);
1210                 function_symbol_list = NULL;
1211                 add_symbol(list, anon);
1212                 return token;
1213         }
1214
1215         /* Parse declaration-specifiers, if any */
1216         token = declaration_specifiers(token, &ctype, 0);
1217         decl = alloc_symbol(token->pos, SYM_NODE);
1218         decl->ctype = ctype;
1219         token = declarator(token, &decl, &ident);
1220
1221         /* Just a type declaration? */
1222         if (!ident)
1223                 return expect(token, ';', "end of type declaration");
1224
1225         decl->ident = ident;
1226
1227         /* type define declaration? */
1228         is_typedef = (ctype.modifiers & MOD_TYPEDEF) != 0;
1229
1230         /* Typedef's don't have meaningful storage */
1231         if (is_typedef) {
1232                 ctype.modifiers &= ~MOD_STORAGE;
1233                 decl->ctype.modifiers &= ~MOD_STORAGE;
1234                 decl->ctype.modifiers |= MOD_USERTYPE;
1235         }
1236
1237         bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
1238
1239         base_type = decl->ctype.base_type;
1240         if (!is_typedef && base_type && base_type->type == SYM_FN) {
1241                 if (match_op(token, '{'))
1242                         return parse_function_body(token, decl, list);
1243
1244                 if (!(decl->ctype.modifiers & MOD_STATIC))
1245                         decl->ctype.modifiers |= MOD_EXTERN;
1246         }
1247
1248         for (;;) {
1249                 if (token_type(token) == TOKEN_IDENT) {
1250                         if (token->ident ==  &asm_ident || token->ident == &__asm_ident || token->ident == &__asm___ident) {
1251                                 struct expression *expr;
1252
1253                                 token = expect(token->next, '(', "after asm");
1254                                 token = parse_expression(token->next, &expr);
1255                                 token = expect(token, ')', "after asm");
1256                         }
1257                 }
1258                 if (!is_typedef && match_op(token, '=')) {
1259                         if (decl->ctype.modifiers & MOD_EXTERN) {
1260                                 warn(decl->pos, "symbol with external linkage has initializer");
1261                                 decl->ctype.modifiers &= ~MOD_EXTERN;
1262                         }
1263                         token = initializer(&decl->initializer, token->next);
1264                 }
1265                 if (!is_typedef) {
1266                         if (!(decl->ctype.modifiers & (MOD_EXTERN | MOD_INLINE))) {
1267                                 add_symbol(list, decl);
1268                                 if (function_symbol_list)
1269                                         fn_local_symbol(decl);
1270                         }
1271                 }
1272                 check_declaration(decl);
1273
1274                 if (!match_op(token, ','))
1275                         break;
1276
1277                 token = token->next;
1278                 ident = NULL;
1279                 decl = alloc_symbol(token->pos, SYM_NODE);
1280                 decl->ctype = ctype;
1281                 token = declaration_specifiers(token, &decl->ctype, 1);
1282                 token = declarator(token, &decl, &ident);
1283                 if (!ident) {
1284                         warn(token->pos, "expected identifier name in type definition");
1285                         return token;
1286                 }
1287
1288                 bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
1289
1290                 /* Function declarations are automatically extern unless specifically static */
1291                 base_type = decl->ctype.base_type;
1292                 if (!is_typedef && base_type && base_type->type == SYM_FN) {
1293                         if (!(decl->ctype.modifiers & MOD_STATIC))
1294                                 decl->ctype.modifiers |= MOD_EXTERN;
1295                 }
1296         }
1297         return expect(token, ';', "at end of declaration");
1298 }
1299
1300 void translation_unit(struct token *token, struct symbol_list **list)
1301 {
1302         while (!eof_token(token))
1303                 token = external_declaration(token, list);
1304         // They aren't needed any more
1305         clear_token_alloc();
1306 }