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  *               2003-2004 Linus Torvalds
   8  *
   9  *  Licensed under the Open Software License version 1.1
  10  */
  11
  12 #include <stdarg.h>
  13 #include <stdlib.h>
  14 #include <stdio.h>
  15 #include <string.h>
  16 #include <ctype.h>
  17 #include <unistd.h>
  18 #include <fcntl.h>
  19 #include <limits.h>
  20
  21 #include "lib.h"
  22 #include "allocate.h"
  23 #include "token.h"
  24 #include "parse.h"
  25 #include "symbol.h"
  26 #include "scope.h"
  27 #include "expression.h"
  28 #include "target.h"
  29
  30 #define warn_on_mixed (1)
  31
  32 static struct symbol_list **function_symbol_list;
  33 struct symbol_list *function_computed_target_list;
  34 struct statement_list *function_computed_goto_list;
  35
  36 static struct token *statement(struct token *token, struct statement **tree);
  37
  38 // Add a symbol to the list of function-local symbols
  39 static void fn_local_symbol(struct symbol *sym)
  40 {
  41         if (function_symbol_list)
  42                 add_symbol(function_symbol_list, sym);
  43 }
  44
  45 static int match_idents(struct token *token, ...)
  46 {
  47         va_list args;
  48
  49         if (token_type(token) != TOKEN_IDENT)
  50                 return 0;
  51
  52         va_start(args, token);
  53         for (;;) {
  54                 struct ident * next = va_arg(args, struct ident *);
  55                 if (!next)
  56                         return 0;
  57                 if (token->ident == next)
  58                         return 1;
  59         }
  60 }
  61
  62
  63 struct statement *alloc_statement(struct position pos, int type)
  64 {
  65         struct statement *stmt = __alloc_statement(0);
  66         stmt->type = type;
  67         stmt->pos = pos;
  68         return stmt;
  69 }
  70
  71 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list);
  72
  73 static struct symbol * indirect(struct position pos, struct ctype *ctype, int type)
  74 {
  75         struct symbol *sym = alloc_symbol(pos, type);
  76
  77         sym->ctype.base_type = ctype->base_type;
  78         sym->ctype.modifiers = ctype->modifiers & ~MOD_STORAGE;
  79
  80         ctype->base_type = sym;
  81         ctype->modifiers &= MOD_STORAGE;
  82         return sym;
  83 }
  84
  85 static struct symbol *lookup_or_create_symbol(enum namespace ns, enum type type, struct token *token)
  86 {
  87         struct symbol *sym = lookup_symbol(token->ident, ns);
  88         if (!sym) {
  89                 sym = alloc_symbol(token->pos, type);
  90                 bind_symbol(sym, token->ident, ns);
  91                 if (type == SYM_LABEL)
  92                         fn_local_symbol(sym);
  93         }
  94         return sym;
  95 }
  96
  97 /*
  98  * NOTE! NS_LABEL is not just a different namespace,
  99  * it also ends up using function scope instead of the
 100  * regular symbol scope.
 101  */
 102 struct symbol *label_symbol(struct token *token)
 103 {
 104         return lookup_or_create_symbol(NS_LABEL, SYM_LABEL, token);
 105 }
 106
 107 static struct token *struct_union_enum_specifier(enum type type,
 108         struct token *token, struct ctype *ctype,
 109         struct token *(*parse)(struct token *, struct symbol *))
 110 {
 111         struct symbol *sym;
 112         struct position *repos;
 113
 114         ctype->modifiers = 0;
 115         if (token_type(token) == TOKEN_IDENT) {
 116                 sym = lookup_symbol(token->ident, NS_STRUCT);
 117                 if (!sym ||
 118                     (sym->scope != block_scope &&
 119                      (match_op(token->next,';') || match_op(token->next,'{')))) {
 120                         // Either a new symbol, or else an out-of-scope
 121                         // symbol being redefined.
 122                         sym = alloc_symbol(token->pos, type);
 123                         bind_symbol(sym, token->ident, NS_STRUCT);
 124                 }
 125                 if (sym->type != type)
 126                         error_die(token->pos, "invalid tag applied to %s", show_typename (sym));
 127                 ctype->base_type = sym;
 128                 repos = &token->pos;
 129                 token = token->next;
 130                 if (match_op(token, '{')) {
 131                         // The following test is actually wrong for empty
 132                         // structs, but (1) they are not C99, (2) gcc does
 133                         // the same thing, and (3) it's easier.
 134                         if (sym->symbol_list)
 135                                 error_die(token->pos, "redefinition of %s", show_typename (sym));
 136                         sym->pos = *repos;
 137                         token = parse(token->next, sym);
 138                         token = expect(token, '}', "at end of struct-union-enum-specifier");
 139
 140                         // Mark the structure as needing re-examination
 141                         sym->examined = 0;
 142                 }
 143                 return token;
 144         }
 145
 146         // private struct/union/enum type
 147         if (!match_op(token, '{')) {
 148                 sparse_error(token->pos, "expected declaration");
 149                 ctype->base_type = &bad_ctype;
 150                 return token;
 151         }
 152
 153         sym = alloc_symbol(token->pos, type);
 154         token = parse(token->next, sym);
 155         ctype->base_type = sym;
 156         return expect(token, '}', "at end of specifier");
 157 }
 158
 159 static struct token *parse_struct_declaration(struct token *token, struct symbol *sym)
 160 {
 161         return struct_declaration_list(token, &sym->symbol_list);
 162 }
 163
 164 static struct token *struct_or_union_specifier(enum type type, struct token *token, struct ctype *ctype)
 165 {
 166         return struct_union_enum_specifier(type, token, ctype, parse_struct_declaration);
 167 }
 168
 169 typedef struct {
 170         int x;
 171         unsigned long long y;
 172 } Num;
 173
 174 static void upper_boundary(Num *n, Num *v)
 175 {
 176         if (n->x > v->x)
 177                 return;
 178         if (n->x < v->x) {
 179                 *n = *v;
 180                 return;
 181         }
 182         if (n->y < v->y)
 183                 n->y = v->y;
 184 }
 185
 186 static void lower_boundary(Num *n, Num *v)
 187 {
 188         if (n->x < v->x)
 189                 return;
 190         if (n->x > v->x) {
 191                 *n = *v;
 192                 return;
 193         }
 194         if (n->y > v->y)
 195                 n->y = v->y;
 196 }
 197
 198 static int type_is_ok(struct symbol *type, Num *upper, Num *lower)
 199 {
 200         int shift = type->bit_size;
 201         int is_unsigned = type->ctype.modifiers & MOD_UNSIGNED;
 202
 203         if (!is_unsigned)
 204                 shift--;
 205         if (upper->x == 0 && upper->y >> shift)
 206                 return 0;
 207         if (lower->x == 0 || (!is_unsigned && (~lower->y >> shift) == 0))
 208                 return 1;
 209         return 0;
 210 }
 211
 212 static struct symbol *bigger_enum_type(struct symbol *s1, struct symbol *s2)
 213 {
 214         if (s1->bit_size < s2->bit_size) {
 215                 s1 = s2;
 216         } else if (s1->bit_size == s2->bit_size) {
 217                 if (s2->ctype.modifiers & MOD_UNSIGNED)
 218                         s1 = s2;
 219         }
 220         if (s1->bit_size < bits_in_int)
 221                 return &int_ctype;
 222         return s1;
 223 }
 224
 225 static void cast_enum_list(struct symbol_list *list, struct symbol *base_type)
 226 {
 227         struct symbol *sym;
 228
 229         FOR_EACH_PTR(list, sym) {
 230                 struct expression *expr = sym->initializer;
 231                 struct symbol *ctype;
 232                 if (expr->type != EXPR_VALUE)
 233                         continue;
 234                 ctype = expr->ctype;
 235                 if (ctype->bit_size == base_type->bit_size)
 236                         continue;
 237                 cast_value(expr, base_type, expr, ctype);
 238         } END_FOR_EACH_PTR(sym);
 239 }
 240
 241 static struct token *parse_enum_declaration(struct token *token, struct symbol *parent)
 242 {
 243         unsigned long long lastval = 0;
 244         struct symbol *ctype = NULL, *base_type = NULL;
 245         Num upper = {-1, 0}, lower = {1, 0};
 246         struct symbol_list *entries = NULL;
 247
 248         parent->examined = 1;
 249         parent->ctype.base_type = &int_ctype;
 250         while (token_type(token) == TOKEN_IDENT) {
 251                 struct expression *expr = NULL;
 252                 struct token *next = token->next;
 253                 struct symbol *sym;
 254
 255                 sym = alloc_symbol(token->pos, SYM_NODE);
 256                 bind_symbol(sym, token->ident, NS_SYMBOL);
 257                 sym->ctype.modifiers &= ~MOD_ADDRESSABLE;
 258
 259                 if (match_op(next, '=')) {
 260                         next = constant_expression(next->next, &expr);
 261                         lastval = get_expression_value(expr);
 262                         ctype = &void_ctype;
 263                         if (expr && expr->ctype)
 264                                 ctype = expr->ctype;
 265                 } else if (!ctype) {
 266                         ctype = &int_ctype;
 267                 } else if (is_int_type(ctype)) {
 268                         lastval++;
 269                 } else {
 270                         error_die(token->pos, "can't increment the last enum member");
 271                 }
 272
 273                 if (!expr) {
 274                         expr = alloc_expression(token->pos, EXPR_VALUE);
 275                         expr->value = lastval;
 276                         expr->ctype = ctype;
 277                 }
 278
 279                 sym->initializer = expr;
 280                 sym->ctype.base_type = parent;
 281                 add_ptr_list(&entries, sym);
 282
 283                 if (base_type != &bad_ctype) {
 284                         if (ctype->type == SYM_NODE)
 285                                 ctype = ctype->ctype.base_type;
 286                         if (ctype->type == SYM_ENUM) {
 287                                 if (ctype == parent)
 288                                         ctype = base_type;
 289                                 else
 290                                         ctype = ctype->ctype.base_type;
 291                         }
 292                         /*
 293                          * base_type rules:
 294                          *  - if all enum's are of the same type, then
 295                          *    the base_type is that type (two first
 296                          *    cases)
 297                          *  - if enums are of different types, they
 298                          *    all have to be integer types, and the
 299                          *    base type is at least "int_ctype".
 300                          *  - otherwise the base_type is "bad_ctype".
 301                          */
 302                         if (!base_type) {
 303                                 base_type = ctype;
 304                         } else if (ctype == base_type) {
 305                                 /* nothing */
 306                         } else if (is_int_type(base_type) && is_int_type(ctype)) {
 307                                 base_type = bigger_enum_type(base_type, ctype);
 308                         } else
 309                                 base_type = &bad_ctype;
 310                         parent->ctype.base_type = base_type;
 311                 }
 312                 if (is_int_type(base_type)) {
 313                         Num v = {.y = lastval};
 314                         if (ctype->ctype.modifiers & MOD_UNSIGNED)
 315                                 v.x = 0;
 316                         else if ((long long)lastval >= 0)
 317                                 v.x = 0;
 318                         else
 319                                 v.x = -1;
 320                         upper_boundary(&upper, &v);
 321                         lower_boundary(&lower, &v);
 322                 }
 323                 token = next;
 324                 if (!match_op(token, ','))
 325                         break;
 326                 token = token->next;
 327         }
 328         if (!base_type) {
 329                 sparse_error(token->pos, "bad enum definition");
 330                 base_type = &bad_ctype;
 331         }
 332         else if (!is_int_type(base_type))
 333                 base_type = base_type;
 334         else if (type_is_ok(base_type, &upper, &lower))
 335                 base_type = base_type;
 336         else if (type_is_ok(&int_ctype, &upper, &lower))
 337                 base_type = &int_ctype;
 338         else if (type_is_ok(&uint_ctype, &upper, &lower))
 339                 base_type = &uint_ctype;
 340         else if (type_is_ok(&long_ctype, &upper, &lower))
 341                 base_type = &long_ctype;
 342         else if (type_is_ok(&ulong_ctype, &upper, &lower))
 343                 base_type = &ulong_ctype;
 344         else if (type_is_ok(&llong_ctype, &upper, &lower))
 345                 base_type = &llong_ctype;
 346         else if (type_is_ok(&ullong_ctype, &upper, &lower))
 347                 base_type = &ullong_ctype;
 348         else
 349                 base_type = &bad_ctype;
 350         parent->ctype.base_type = base_type;
 351         parent->ctype.modifiers |= (base_type->ctype.modifiers & MOD_UNSIGNED);
 352         parent->examined = 0;
 353
 354         cast_enum_list(entries, base_type);
 355         free_ptr_list(&entries);
 356
 357         return token;
 358 }
 359
 360 static struct token *enum_specifier(struct token *token, struct ctype *ctype)
 361 {
 362         struct token *ret = struct_union_enum_specifier(SYM_ENUM, token, ctype, parse_enum_declaration);
 363
 364         ctype = &ctype->base_type->ctype;
 365         if (!ctype->base_type)
 366                 ctype->base_type = &incomplete_ctype;
 367
 368         return ret;
 369 }
 370
 371 static struct token *typeof_specifier(struct token *token, struct ctype *ctype)
 372 {
 373         struct symbol *sym;
 374
 375         if (!match_op(token, '(')) {
 376                 sparse_error(token->pos, "expected '(' after typeof");
 377                 return token;
 378         }
 379         if (lookup_type(token->next)) {
 380                 token = typename(token->next, &sym);
 381                 *ctype = sym->ctype;
 382         } else {
 383                 struct symbol *typeof_sym = alloc_symbol(token->pos, SYM_TYPEOF);
 384                 token = parse_expression(token->next, &typeof_sym->initializer);
 385
 386                 ctype->modifiers = 0;
 387                 ctype->base_type = typeof_sym;
 388         }
 389         return expect(token, ')', "after typeof");
 390 }
 391
 392 static const char * handle_attribute(struct ctype *ctype, struct ident *attribute, struct expression *expr)
 393 {
 394         if (attribute == &packed_ident ||
 395             attribute == &__packed___ident) {
 396                 ctype->alignment = 1;
 397                 return NULL;
 398         }
 399         if (attribute == &aligned_ident ||
 400             attribute == &__aligned___ident) {
 401                 int alignment = max_alignment;
 402                 if (expr)
 403                         alignment = get_expression_value(expr);
 404                 ctype->alignment = alignment;
 405                 return NULL;
 406         }
 407         if (attribute == &nocast_ident) {
 408                 ctype->modifiers |= MOD_NOCAST;
 409                 return NULL;
 410         }
 411         if (attribute == &noderef_ident) {
 412                 ctype->modifiers |= MOD_NODEREF;
 413                 return NULL;
 414         }
 415         if (attribute == &safe_ident) {
 416                 ctype->modifiers |= MOD_SAFE;
 417                 return NULL;
 418         }
 419         if (attribute == &force_ident) {
 420                 ctype->modifiers |= MOD_FORCE;
 421                 return NULL;
 422         }
 423         if (attribute == &bitwise_ident ||
 424             attribute == &__bitwise___ident) {
 425                 if (Wbitwise)
 426                         ctype->modifiers |= MOD_BITWISE;
 427                 return NULL;
 428         }
 429         if (attribute == &address_space_ident) {
 430                 if (!expr)
 431                         return "expected address space number";
 432                 ctype->as = get_expression_value(expr);
 433                 return NULL;
 434         }
 435         if (attribute == &context_ident) {
 436                 if (expr && expr->type == EXPR_COMMA) {
 437                         int input = get_expression_value(expr->left);
 438                         int output = get_expression_value(expr->right);
 439                         ctype->in_context = input;
 440                         ctype->out_context = output;
 441                         return NULL;
 442                 }
 443                 return "expected context input/output values";
 444         }
 445         if (attribute == &mode_ident ||
 446             attribute == &__mode___ident) {
 447                 if (expr && expr->type == EXPR_SYMBOL) {
 448                         struct ident *ident = expr->symbol_name;
 449
 450                         /*
 451                          * Match against __QI__/__HI__/__SI__/__DI__
 452                          *
 453                          * FIXME! This is broken - we don't actually get
 454                          * the type information updated properly at this
 455                          * stage for some reason.
 456                          */
 457                         if (ident == &__QI___ident ||
 458                             ident == &QI_ident) {
 459                                 ctype->modifiers |= MOD_CHAR;
 460                                 return NULL;
 461                         }
 462                         if (ident == &__HI___ident ||
 463                             ident == &HI_ident) {
 464                                 ctype->modifiers |= MOD_SHORT;
 465                                 return NULL;
 466                         }
 467                         if (ident == &__SI___ident ||
 468                             ident == &SI_ident) {
 469                                 /* Nothing? */
 470                                 return NULL;
 471                         }
 472                         if (ident == &__DI___ident ||
 473                             ident == &DI_ident) {
 474                                 ctype->modifiers |= MOD_LONGLONG;
 475                                 return NULL;
 476                         }
 477                         if (ident == &__word___ident ||
 478                             ident == &word_ident) {
 479                                 ctype->modifiers |= MOD_LONG;
 480                                 return NULL;
 481                         }
 482                         return "unknown mode attribute";
 483                 }
 484                 return "expected attribute mode symbol";
 485         }
 486
 487         /* Throw away for now.. */
 488         if (attribute == &__transparent_union___ident) {
 489                 if (Wtransparent_union)
 490                     return "ignoring attribute __transparent_union__";
 491                 return NULL;
 492         }
 493         if (attribute == &nothrow_ident ||
 494             attribute == &__nothrow_ident ||
 495             attribute == &__nothrow___ident)
 496                 return NULL;
 497         if (attribute == &__malloc___ident)
 498                 return NULL;
 499         if (attribute == &nonnull_ident ||
 500             attribute == &__nonnull_ident ||
 501             attribute == &__nonnull___ident)
 502                 return NULL;
 503         if (attribute == &format_ident ||
 504             attribute == &__format___ident ||
 505             attribute == &__format_arg___ident)
 506                 return NULL;
 507         if (attribute == &section_ident ||
 508             attribute == &__section___ident)
 509                 return NULL;
 510         if (attribute == &unused_ident ||
 511             attribute == &__unused___ident)
 512                 return NULL;
 513         if (attribute == &const_ident ||
 514             attribute == &__const_ident ||
 515             attribute == &__const___ident)
 516                 return NULL;
 517         if (attribute == &noreturn_ident ||
 518             attribute == &__noreturn___ident)
 519                 return NULL;
 520         if (attribute == &sentinel_ident ||
 521             attribute == &__sentinel___ident)
 522                 return NULL;
 523         if (attribute == &regparm_ident)
 524                 return NULL;
 525         if (attribute == &weak_ident ||
 526             attribute == &__weak___ident)
 527                 return NULL;
 528         if (attribute == &alias_ident)
 529                 return NULL;
 530         if (attribute == &pure_ident ||
 531             attribute == &__pure___ident)
 532                 return NULL;
 533         if (attribute == &always_inline_ident)
 534                 return NULL;
 535         if (attribute == &syscall_linkage_ident)
 536                 return NULL;
 537         if (attribute == &visibility_ident)
 538                 return NULL;
 539         if (attribute == &deprecated_ident ||
 540             attribute == &__deprecated___ident)
 541                 return NULL;
 542         if (attribute == &noinline_ident)
 543                 return NULL;
 544         if (attribute == &__used___ident)
 545                 return NULL;
 546         if (attribute == &warn_unused_result_ident ||
 547             attribute == &__warn_unused_result___ident)
 548                 return NULL;
 549         if (attribute == &model_ident ||
 550             attribute == &__model___ident)
 551                 return NULL;
 552
 553         return "unknown attribute";
 554 }
 555
 556 static struct token *attribute_specifier(struct token *token, struct ctype *ctype)
 557 {
 558         ctype->modifiers = 0;
 559         token = expect(token, '(', "after attribute");
 560         token = expect(token, '(', "after attribute");
 561
 562         for (;;) {
 563                 const char *error_str;
 564                 struct ident *attribute_name;
 565                 struct expression *attribute_expr;
 566
 567                 if (eof_token(token))
 568                         break;
 569                 if (match_op(token, ';'))
 570                         break;
 571                 if (token_type(token) != TOKEN_IDENT)
 572                         break;
 573                 attribute_name = token->ident;
 574                 token = token->next;
 575                 attribute_expr = NULL;
 576                 if (match_op(token, '('))
 577                         token = parens_expression(token, &attribute_expr, "in attribute");
 578                 error_str = handle_attribute(ctype, attribute_name, attribute_expr);
 579                 if (error_str)
 580                         sparse_error(token->pos, "attribute '%s': %s", show_ident(attribute_name), error_str);
 581                 if (!match_op(token, ','))
 582                         break;
 583                 token = token->next;
 584         }
 585
 586         token = expect(token, ')', "after attribute");
 587         token = expect(token, ')', "after attribute");
 588         return token;
 589 }
 590
 591 struct symbol * ctype_integer(unsigned long spec)
 592 {
 593         static struct symbol *const integer_ctypes[][3] = {
 594                 { &llong_ctype, &sllong_ctype, &ullong_ctype },
 595                 { &long_ctype,  &slong_ctype,  &ulong_ctype  },
 596                 { &short_ctype, &sshort_ctype, &ushort_ctype },
 597                 { &char_ctype,  &schar_ctype,  &uchar_ctype  },
 598                 { &int_ctype,   &sint_ctype,   &uint_ctype   },
 599         };
 600         struct symbol *const (*ctype)[3];
 601         int sub;
 602
 603         ctype = integer_ctypes;
 604         if (!(spec & MOD_LONGLONG)) {
 605                 ctype++;
 606                 if (!(spec & MOD_LONG)) {
 607                         ctype++;
 608                         if (!(spec & MOD_SHORT)) {
 609                                 ctype++;
 610                                 if (!(spec & MOD_CHAR))
 611                                         ctype++;
 612                         }
 613                 }
 614         }
 615
 616         sub = ((spec & MOD_UNSIGNED)
 617                ? 2
 618                : ((spec & MOD_EXPLICITLY_SIGNED)
 619                   ? 1
 620                   : 0));
 621
 622         return ctype[0][sub];
 623 }
 624
 625 struct symbol * ctype_fp(unsigned long spec)
 626 {
 627         if (spec & MOD_LONGLONG)
 628                 return &ldouble_ctype;
 629         if (spec & MOD_LONG)
 630                 return &double_ctype;
 631         return &float_ctype;
 632 }
 633
 634 static void apply_ctype(struct position pos, struct ctype *thistype, struct ctype *ctype)
 635 {
 636         unsigned long mod = thistype->modifiers;
 637
 638         if (mod) {
 639                 unsigned long old = ctype->modifiers;
 640                 unsigned long extra = 0, dup, conflict;
 641
 642                 if (mod & old & MOD_LONG) {
 643                         extra = MOD_LONGLONG | MOD_LONG;
 644                         mod &= ~MOD_LONG;
 645                         old &= ~MOD_LONG;
 646                 }
 647                 dup = (mod & old) | (extra & old) | (extra & mod);
 648                 if (dup)
 649                         sparse_error(pos, "Just how %sdo you want this type to be?",
 650                                 modifier_string(dup));
 651
 652                 conflict = !(~mod & ~old & (MOD_LONG | MOD_SHORT));
 653                 if (conflict)
 654                         sparse_error(pos, "You cannot have both long and short modifiers.");
 655
 656                 conflict = !(~mod & ~old & (MOD_SIGNED | MOD_UNSIGNED));
 657                 if (conflict)
 658                         sparse_error(pos, "You cannot have both signed and unsigned modifiers.");
 659
 660                 // Only one storage modifier allowed, except that "inline" doesn't count.
 661                 conflict = (mod | old) & (MOD_STORAGE & ~MOD_INLINE);
 662                 conflict &= (conflict - 1);
 663                 if (conflict)
 664                         sparse_error(pos, "multiple storage classes");
 665
 666                 ctype->modifiers = old | mod | extra;
 667         }
 668
 669         /* Context mask and value */
 670         ctype->in_context += thistype->in_context;
 671         ctype->out_context += thistype->out_context;
 672
 673         /* Alignment */
 674         if (thistype->alignment & (thistype->alignment-1)) {
 675                 warning(pos, "I don't like non-power-of-2 alignments");
 676                 thistype->alignment = 0;
 677         }
 678         if (thistype->alignment > ctype->alignment)
 679                 ctype->alignment = thistype->alignment;
 680
 681         /* Address space */
 682         if (thistype->as)
 683                 ctype->as = thistype->as;
 684 }
 685
 686 static void check_modifiers(struct position *pos, struct symbol *s, unsigned long mod)
 687 {
 688         unsigned long banned, wrong;
 689         unsigned long this_mod = s->ctype.modifiers;
 690         const unsigned long BANNED_SIZE = MOD_LONG | MOD_LONGLONG | MOD_SHORT;
 691         const unsigned long BANNED_SIGN = MOD_SIGNED | MOD_UNSIGNED;
 692
 693         if (this_mod & (MOD_STRUCTOF | MOD_UNIONOF | MOD_ENUMOF))
 694                 banned = BANNED_SIZE | BANNED_SIGN;
 695         else if (this_mod & MOD_SPECIALBITS)
 696                 banned = 0;
 697         else if (s->ctype.base_type == &fp_type)
 698                 banned = BANNED_SIGN;
 699         else if (s->ctype.base_type == &int_type || !s->ctype.base_type || is_int_type (s))
 700                 banned = 0;
 701         else {
 702                 // label_type
 703                 // void_type
 704                 // bad_type
 705                 // vector_type <-- whatever that is
 706                 banned = BANNED_SIZE | BANNED_SIGN;
 707         }
 708
 709         wrong = mod & banned;
 710         if (wrong)
 711                 sparse_error(*pos, "modifier %sis invalid in this context",
 712                      modifier_string (wrong));
 713 }
 714
 715
 716 static struct token *declaration_specifiers(struct token *next, struct ctype *ctype, int qual)
 717 {
 718         struct token *token;
 719
 720         while ( (token = next) != NULL ) {
 721                 struct ctype thistype;
 722                 struct ident *ident;
 723                 struct symbol *s, *type;
 724                 unsigned long mod;
 725
 726                 next = token->next;
 727                 if (token_type(token) != TOKEN_IDENT)
 728                         break;
 729                 ident = token->ident;
 730
 731                 s = lookup_symbol(ident, NS_TYPEDEF);
 732                 if (!s)
 733                         break;
 734                 thistype = s->ctype;
 735                 mod = thistype.modifiers;
 736                 if (qual && (mod & ~(MOD_ATTRIBUTE | MOD_CONST | MOD_VOLATILE)))
 737                         break;
 738                 if (mod & MOD_SPECIALBITS) {
 739                         if (mod & MOD_STRUCTOF)
 740                                 next = struct_or_union_specifier(SYM_STRUCT, next, &thistype);
 741                         else if (mod & MOD_UNIONOF)
 742                                 next = struct_or_union_specifier(SYM_UNION, next, &thistype);
 743                         else if (mod & MOD_ENUMOF)
 744                                 next = enum_specifier(next, &thistype);
 745                         else if (mod & MOD_ATTRIBUTE)
 746                                 next = attribute_specifier(next, &thistype);
 747                         else if (mod & MOD_TYPEOF)
 748                                 next = typeof_specifier(next, &thistype);
 749                         mod = thistype.modifiers;
 750                 }
 751                 type = thistype.base_type;
 752                 if (type) {
 753                         if (qual)
 754                                 break;
 755                         if (ctype->base_type)
 756                                 break;
 757                         /* User types only mix with qualifiers */
 758                         if (mod & MOD_USERTYPE) {
 759                                 if (ctype->modifiers & MOD_SPECIFIER)
 760                                         break;
 761                         }
 762                         ctype->base_type = type;
 763                 }
 764
 765                 check_modifiers(&token->pos, s, ctype->modifiers);
 766                 apply_ctype(token->pos, &thistype, ctype);
 767         }
 768
 769         /* Turn the "virtual types" into real types with real sizes etc */
 770         if (!ctype->base_type) {
 771                 struct symbol *base = &incomplete_ctype;
 772
 773                 /*
 774                  * If we have modifiers, we'll default to an integer
 775                  * type, and "ctype_integer()" will turn this into
 776                  * a specific one.
 777                  */
 778                 if (ctype->modifiers & MOD_SPECIFIER)
 779                         base = &int_type;
 780                 ctype->base_type = base;
 781         }
 782         if (ctype->base_type == &int_type) {
 783                 ctype->base_type = ctype_integer(ctype->modifiers);
 784                 ctype->modifiers &= ~MOD_SPECIFIER;
 785         } else if (ctype->base_type == &fp_type) {
 786                 ctype->base_type = ctype_fp(ctype->modifiers);
 787                 ctype->modifiers &= ~MOD_SPECIFIER;
 788         }
 789         if (ctype->modifiers & MOD_BITWISE) {
 790                 struct symbol *type;
 791                 ctype->modifiers &= ~(MOD_BITWISE | MOD_SPECIFIER);
 792                 if (!is_int_type(ctype->base_type)) {
 793                         sparse_error(token->pos, "invalid modifier");
 794                         return token;
 795                 }
 796                 type = alloc_symbol(token->pos, SYM_BASETYPE);
 797                 *type = *ctype->base_type;
 798                 type->ctype.base_type = ctype->base_type;
 799                 type->type = SYM_RESTRICT;
 800                 type->ctype.modifiers &= ~MOD_SPECIFIER;
 801                 ctype->base_type = type;
 802         }
 803         return token;
 804 }
 805
 806 static struct token *abstract_array_declarator(struct token *token, struct symbol *sym)
 807 {
 808         struct expression *expr = NULL;
 809
 810         token = parse_expression(token, &expr);
 811         sym->array_size = expr;
 812         return token;
 813 }
 814
 815 static struct token *parameter_type_list(struct token *, struct symbol *, struct ident **p);
 816 static struct token *declarator(struct token *token, struct symbol *sym, struct ident **p);
 817
 818 static struct token *handle_attributes(struct token *token, struct ctype *ctype)
 819 {
 820         for (;;) {
 821                 if (token_type(token) != TOKEN_IDENT)
 822                         break;
 823                 if (match_idents(token, &__attribute___ident, &__attribute_ident, NULL)) {
 824                         struct ctype thistype = { 0, };
 825                         token = attribute_specifier(token->next, &thistype);
 826                         apply_ctype(token->pos, &thistype, ctype);
 827                         continue;
 828                 }
 829                 if (match_idents(token, &asm_ident, &__asm_ident, &__asm___ident)) {
 830                         struct expression *expr;
 831                         token = expect(token->next, '(', "after asm");
 832                         token = parse_expression(token->next, &expr);
 833                         token = expect(token, ')', "after asm");
 834                         continue;
 835                 }
 836                 break;
 837         }
 838         return token;
 839 }
 840
 841 static struct token *direct_declarator(struct token *token, struct symbol *decl, struct ident **p)
 842 {
 843         struct ctype *ctype = &decl->ctype;
 844
 845         if (p && token_type(token) == TOKEN_IDENT) {
 846                 *p = token->ident;
 847                 token = token->next;
 848         }
 849
 850         for (;;) {
 851                 token = handle_attributes(token, ctype);
 852
 853                 if (token_type(token) != TOKEN_SPECIAL)
 854                         return token;
 855
 856                 /*
 857                  * This can be either a parameter list or a grouping.
 858                  * For the direct (non-abstract) case, we know if must be
 859                  * a parameter list if we already saw the identifier.
 860                  * For the abstract case, we know if must be a parameter
 861                  * list if it is empty or starts with a type.
 862                  */
 863                 if (token->special == '(') {
 864                         struct symbol *sym;
 865                         struct token *next = token->next;
 866                         int fn = (p && *p) || match_op(next, ')') || lookup_type(next);
 867
 868                         if (!fn) {
 869                                 struct symbol *base_type = ctype->base_type;
 870                                 token = declarator(next, decl, p);
 871                                 token = expect(token, ')', "in nested declarator");
 872                                 while (ctype->base_type != base_type)
 873                                         ctype = &ctype->base_type->ctype;
 874                                 p = NULL;
 875                                 continue;
 876                         }
 877
 878                         sym = indirect(token->pos, ctype, SYM_FN);
 879                         token = parameter_type_list(next, sym, p);
 880                         token = expect(token, ')', "in function declarator");
 881                         continue;
 882                 }
 883                 if (token->special == '[') {
 884                         struct symbol *array = indirect(token->pos, ctype, SYM_ARRAY);
 885                         token = abstract_array_declarator(token->next, array);
 886                         token = expect(token, ']', "in abstract_array_declarator");
 887                         ctype = &array->ctype;
 888                         continue;
 889                 }
 890                 break;
 891         }
 892         return token;
 893 }
 894
 895 static struct token *pointer(struct token *token, struct ctype *ctype)
 896 {
 897         unsigned long modifiers;
 898         struct symbol *base_type;
 899
 900         modifiers = ctype->modifiers & ~(MOD_TYPEDEF | MOD_ATTRIBUTE);
 901         base_type = ctype->base_type;
 902         ctype->modifiers = modifiers;
 903
 904         while (match_op(token,'*')) {
 905                 struct symbol *ptr = alloc_symbol(token->pos, SYM_PTR);
 906                 ptr->ctype.modifiers = modifiers & ~MOD_STORAGE;
 907                 ptr->ctype.as = ctype->as;
 908                 ptr->ctype.in_context += ctype->in_context;
 909                 ptr->ctype.out_context += ctype->out_context;
 910                 ptr->ctype.base_type = base_type;
 911
 912                 base_type = ptr;
 913                 ctype->modifiers = modifiers & MOD_STORAGE;
 914                 ctype->base_type = base_type;
 915                 ctype->as = 0;
 916                 ctype->in_context = 0;
 917                 ctype->out_context = 0;
 918
 919                 token = declaration_specifiers(token->next, ctype, 1);
 920                 modifiers = ctype->modifiers;
 921         }
 922         return token;
 923 }
 924
 925 static struct token *declarator(struct token *token, struct symbol *sym, struct ident **p)
 926 {
 927         token = pointer(token, &sym->ctype);
 928         return direct_declarator(token, sym, p);
 929 }
 930
 931 static struct token *handle_bitfield(struct token *token, struct symbol *decl)
 932 {
 933         struct ctype *ctype = &decl->ctype;
 934         struct expression *expr;
 935         struct symbol *bitfield;
 936         long long width;
 937
 938         if (!is_int_type(ctype->base_type)) {
 939                 sparse_error(token->pos, "invalid bitfield specifier for type %s.",
 940                         show_typename(ctype->base_type));
 941                 // Parse this to recover gracefully.
 942                 return conditional_expression(token->next, &expr);
 943         }
 944
 945         bitfield = indirect(token->pos, ctype, SYM_BITFIELD);
 946         token = conditional_expression(token->next, &expr);
 947         width = get_expression_value(expr);
 948         bitfield->bit_size = width;
 949
 950         if (width < 0 || width > INT_MAX) {
 951                 sparse_error(token->pos, "invalid bitfield width, %lld.", width);
 952                 width = -1;
 953         } else if (decl->ident && width == 0) {
 954                 sparse_error(token->pos, "invalid named zero-width bitfield `%s'",
 955                      show_ident(decl->ident));
 956                 width = -1;
 957         } else if (decl->ident) {
 958                 struct symbol *base_type = bitfield->ctype.base_type;
 959                 int is_signed = !(base_type->ctype.modifiers & MOD_UNSIGNED);
 960                 if (Wone_bit_signed_bitfield && width == 1 && is_signed) {
 961                         // Valid values are either {-1;0} or {0}, depending on integer
 962                         // representation.  The latter makes for very efficient code...
 963                         sparse_error(token->pos, "dubious one-bit signed bitfield");
 964                 }
 965                 if (Wdefault_bitfield_sign &&
 966                     base_type->type != SYM_ENUM &&
 967                     !(base_type->ctype.modifiers & MOD_EXPLICITLY_SIGNED) &&
 968                     is_signed) {
 969                         // The sign of bitfields is unspecified by default.
 970                         sparse_error(token->pos, "dubious bitfield without explicit `signed' or `unsigned'");
 971                 }
 972         }
 973         bitfield->bit_size = width;
 974         return token;
 975 }
 976
 977 static struct token *declaration_list(struct token *token, struct symbol_list **list)
 978 {
 979         struct ctype ctype = {0, };
 980
 981         token = declaration_specifiers(token, &ctype, 0);
 982         for (;;) {
 983                 struct ident *ident = NULL;
 984                 struct symbol *decl = alloc_symbol(token->pos, SYM_NODE);
 985                 decl->ctype = ctype;
 986                 token = declarator(token, decl, &ident);
 987                 decl->ident = ident;
 988                 if (match_op(token, ':')) {
 989                         token = handle_bitfield(token, decl);
 990                         token = handle_attributes(token, &decl->ctype);
 991                 }
 992                 add_symbol(list, decl);
 993                 if (!match_op(token, ','))
 994                         break;
 995                 token = token->next;
 996         }
 997         return token;
 998 }
 999
1000 static struct token *struct_declaration_list(struct token *token, struct symbol_list **list)
1001 {
1002         while (!match_op(token, '}')) {
1003                 token = declaration_list(token, list);
1004                 if (!match_op(token, ';')) {
1005                         sparse_error(token->pos, "expected ; at end of declaration");
1006                         break;
1007                 }
1008                 token = token->next;
1009         }
1010         return token;
1011 }
1012
1013 static struct token *parameter_declaration(struct token *token, struct symbol **tree)
1014 {
1015         struct ident *ident = NULL;
1016         struct symbol *sym;
1017         struct ctype ctype = { 0, };
1018
1019         token = declaration_specifiers(token, &ctype, 0);
1020         sym = alloc_symbol(token->pos, SYM_NODE);
1021         sym->ctype = ctype;
1022         *tree = sym;
1023         token = declarator(token, sym, &ident);
1024         sym->ident = ident;
1025         return token;
1026 }
1027
1028 struct token *typename(struct token *token, struct symbol **p)
1029 {
1030         struct symbol *sym = alloc_symbol(token->pos, SYM_NODE);
1031         *p = sym;
1032         token = declaration_specifiers(token, &sym->ctype, 0);
1033         return declarator(token, sym, NULL);
1034 }
1035
1036 static struct token *expression_statement(struct token *token, struct expression **tree)
1037 {
1038         token = parse_expression(token, tree);
1039         return expect(token, ';', "at end of statement");
1040 }
1041
1042 static struct token *parse_asm_operands(struct token *token, struct statement *stmt,
1043         struct expression_list **inout)
1044 {
1045         struct expression *expr;
1046
1047         /* Allow empty operands */
1048         if (match_op(token->next, ':') || match_op(token->next, ')'))
1049                 return token->next;
1050         do {
1051                 struct ident *ident = NULL;
1052                 if (match_op(token->next, '[') &&
1053                     token_type(token->next->next) == TOKEN_IDENT &&
1054                     match_op(token->next->next->next, ']')) {
1055                         ident = token->next->next->ident;
1056                         token = token->next->next->next;
1057                 }
1058                 add_expression(inout, (struct expression *)ident); /* UGGLEE!!! */
1059                 token = primary_expression(token->next, &expr);
1060                 add_expression(inout, expr);
1061                 token = parens_expression(token, &expr, "in asm parameter");
1062                 add_expression(inout, expr);
1063         } while (match_op(token, ','));
1064         return token;
1065 }
1066
1067 static struct token *parse_asm_clobbers(struct token *token, struct statement *stmt,
1068         struct expression_list **clobbers)
1069 {
1070         struct expression *expr;
1071
1072         do {
1073                 token = primary_expression(token->next, &expr);
1074                 add_expression(clobbers, expr);
1075         } while (match_op(token, ','));
1076         return token;
1077 }
1078
1079 static struct token *parse_asm(struct token *token, struct statement *stmt)
1080 {
1081         stmt->type = STMT_ASM;
1082         if (match_idents(token, &__volatile___ident, &__volatile_ident, &volatile_ident, NULL)) {
1083                 token = token->next;
1084         }
1085         token = expect(token, '(', "after asm");
1086         token = parse_expression(token, &stmt->asm_string);
1087         if (match_op(token, ':'))
1088                 token = parse_asm_operands(token, stmt, &stmt->asm_outputs);
1089         if (match_op(token, ':'))
1090                 token = parse_asm_operands(token, stmt, &stmt->asm_inputs);
1091         if (match_op(token, ':'))
1092                 token = parse_asm_clobbers(token, stmt, &stmt->asm_clobbers);
1093         token = expect(token, ')', "after asm");
1094         return expect(token, ';', "at end of asm-statement");
1095 }
1096
1097 /* Make a statement out of an expression */
1098 static struct statement *make_statement(struct expression *expr)
1099 {
1100         struct statement *stmt;
1101
1102         if (!expr)
1103                 return NULL;
1104         stmt = alloc_statement(expr->pos, STMT_EXPRESSION);
1105         stmt->expression = expr;
1106         return stmt;
1107 }
1108
1109 /*
1110  * All iterators have two symbols associated with them:
1111  * the "continue" and "break" symbols, which are targets
1112  * for continue and break statements respectively.
1113  *
1114  * They are in a special name-space, but they follow
1115  * all the normal visibility rules, so nested iterators
1116  * automatically work right.
1117  */
1118 static void start_iterator(struct statement *stmt)
1119 {
1120         struct symbol *cont, *brk;
1121
1122         start_symbol_scope();
1123         cont = alloc_symbol(stmt->pos, SYM_NODE);
1124         bind_symbol(cont, &continue_ident, NS_ITERATOR);
1125         brk = alloc_symbol(stmt->pos, SYM_NODE);
1126         bind_symbol(brk, &break_ident, NS_ITERATOR);
1127
1128         stmt->type = STMT_ITERATOR;
1129         stmt->iterator_break = brk;
1130         stmt->iterator_continue = cont;
1131         fn_local_symbol(brk);
1132         fn_local_symbol(cont);
1133 }
1134
1135 static void end_iterator(struct statement *stmt)
1136 {
1137         end_symbol_scope();
1138 }
1139
1140 static struct statement *start_function(struct symbol *sym)
1141 {
1142         struct symbol *ret;
1143         struct statement *stmt = alloc_statement(sym->pos, STMT_COMPOUND);
1144
1145         start_function_scope();
1146         ret = alloc_symbol(sym->pos, SYM_NODE);
1147         ret->ctype = sym->ctype.base_type->ctype;
1148         ret->ctype.modifiers &= ~(MOD_STORAGE | MOD_CONST | MOD_VOLATILE | MOD_INLINE | MOD_ADDRESSABLE | MOD_NOCAST | MOD_NODEREF | MOD_ACCESSED | MOD_TOPLEVEL);
1149         ret->ctype.modifiers |= (MOD_AUTO | MOD_REGISTER);
1150         bind_symbol(ret, &return_ident, NS_ITERATOR);
1151         stmt->ret = ret;
1152         fn_local_symbol(ret);
1153
1154         // Currently parsed symbol for __func__/__FUNCTION__/__PRETTY_FUNCTION__
1155         current_fn = sym;
1156
1157         return stmt;
1158 }
1159
1160 static void end_function(struct symbol *sym)
1161 {
1162         current_fn = NULL;
1163         end_function_scope();
1164 }
1165
1166 /*
1167  * A "switch()" statement, like an iterator, has a
1168  * the "break" symbol associated with it. It works
1169  * exactly like the iterator break - it's the target
1170  * for any break-statements in scope, and means that
1171  * "break" handling doesn't even need to know whether
1172  * it's breaking out of an iterator or a switch.
1173  *
1174  * In addition, the "case" symbol is a marker for the
1175  * case/default statements to find the switch statement
1176  * that they are associated with.
1177  */
1178 static void start_switch(struct statement *stmt)
1179 {
1180         struct symbol *brk, *switch_case;
1181
1182         start_symbol_scope();
1183         brk = alloc_symbol(stmt->pos, SYM_NODE);
1184         bind_symbol(brk, &break_ident, NS_ITERATOR);
1185
1186         switch_case = alloc_symbol(stmt->pos, SYM_NODE);
1187         bind_symbol(switch_case, &case_ident, NS_ITERATOR);
1188         switch_case->stmt = stmt;
1189
1190         stmt->type = STMT_SWITCH;
1191         stmt->switch_break = brk;
1192         stmt->switch_case = switch_case;
1193
1194         fn_local_symbol(brk);
1195         fn_local_symbol(switch_case);
1196 }
1197
1198 static void end_switch(struct statement *stmt)
1199 {
1200         if (!stmt->switch_case->symbol_list)
1201                 warning(stmt->pos, "switch with no cases");
1202         end_symbol_scope();
1203 }
1204
1205 static void add_case_statement(struct statement *stmt)
1206 {
1207         struct symbol *target = lookup_symbol(&case_ident, NS_ITERATOR);
1208         struct symbol *sym;
1209
1210         if (!target) {
1211                 sparse_error(stmt->pos, "not in switch scope");
1212                 stmt->type = STMT_NONE;
1213                 return;
1214         }
1215         sym = alloc_symbol(stmt->pos, SYM_NODE);
1216         add_symbol(&target->symbol_list, sym);
1217         sym->stmt = stmt;
1218         stmt->case_label = sym;
1219         fn_local_symbol(sym);
1220 }
1221
1222 static struct token *parse_return_statement(struct token *token, struct statement *stmt)
1223 {
1224         struct symbol *target = lookup_symbol(&return_ident, NS_ITERATOR);
1225
1226         if (!target)
1227                 error_die(token->pos, "internal error: return without a function target");
1228         stmt->type = STMT_RETURN;
1229         stmt->ret_target = target;
1230         return expression_statement(token->next, &stmt->ret_value);
1231 }
1232
1233 static struct token *parse_for_statement(struct token *token, struct statement *stmt)
1234 {
1235         struct symbol_list *syms;
1236         struct expression *e1, *e2, *e3;
1237         struct statement *iterator;
1238
1239         start_iterator(stmt);
1240         token = expect(token->next, '(', "after 'for'");
1241
1242         syms = NULL;
1243         e1 = NULL;
1244         /* C99 variable declaration? */
1245         if (lookup_type(token)) {
1246                 token = external_declaration(token, &syms);
1247         } else {
1248                 token = parse_expression(token, &e1);
1249                 token = expect(token, ';', "in 'for'");
1250         }
1251         token = parse_expression(token, &e2);
1252         token = expect(token, ';', "in 'for'");
1253         token = parse_expression(token, &e3);
1254         token = expect(token, ')', "in 'for'");
1255         token = statement(token, &iterator);
1256
1257         stmt->iterator_syms = syms;
1258         stmt->iterator_pre_statement = make_statement(e1);
1259         stmt->iterator_pre_condition = e2;
1260         stmt->iterator_post_statement = make_statement(e3);
1261         stmt->iterator_post_condition = NULL;
1262         stmt->iterator_statement = iterator;
1263         end_iterator(stmt);
1264
1265         return token;
1266 }
1267
1268 static struct token *parse_while_statement(struct token *token, struct statement *stmt)
1269 {
1270         struct expression *expr;
1271         struct statement *iterator;
1272
1273         start_iterator(stmt);
1274         token = parens_expression(token->next, &expr, "after 'while'");
1275         token = statement(token, &iterator);
1276
1277         stmt->iterator_pre_condition = expr;
1278         stmt->iterator_post_condition = NULL;
1279         stmt->iterator_statement = iterator;
1280         end_iterator(stmt);
1281
1282         return token;
1283 }
1284
1285 static struct token *parse_do_statement(struct token *token, struct statement *stmt)
1286 {
1287         struct expression *expr;
1288         struct statement *iterator;
1289
1290         start_iterator(stmt);
1291         token = statement(token->next, &iterator);
1292         if (token_type(token) == TOKEN_IDENT && token->ident == &while_ident)
1293                 token = token->next;
1294         else
1295                 sparse_error(token->pos, "expected 'while' after 'do'");
1296         token = parens_expression(token, &expr, "after 'do-while'");
1297
1298         stmt->iterator_post_condition = expr;
1299         stmt->iterator_statement = iterator;
1300         end_iterator(stmt);
1301
1302         return expect(token, ';', "after statement");
1303 }
1304
1305 static struct token *statement(struct token *token, struct statement **tree)
1306 {
1307         struct statement *stmt = alloc_statement(token->pos, STMT_NONE);
1308
1309         *tree = stmt;
1310         if (token_type(token) == TOKEN_IDENT) {
1311                 if (token->ident == &if_ident) {
1312                         stmt->type = STMT_IF;
1313                         token = parens_expression(token->next, &stmt->if_conditional, "after if");
1314                         token = statement(token, &stmt->if_true);
1315                         if (token_type(token) != TOKEN_IDENT)
1316                                 return token;
1317                         if (token->ident != &else_ident)
1318                                 return token;
1319                         return statement(token->next, &stmt->if_false);
1320                 }
1321
1322                 if (token->ident == &return_ident)
1323                         return parse_return_statement(token, stmt);
1324
1325                 if (token->ident == &break_ident || token->ident == &continue_ident) {
1326                         struct symbol *target = lookup_symbol(token->ident, NS_ITERATOR);
1327                         stmt->type = STMT_GOTO;
1328                         stmt->goto_label = target;
1329                         if (!target)
1330                                 sparse_error(stmt->pos, "break/continue not in iterator scope");
1331                         return expect(token->next, ';', "at end of statement");
1332                 }
1333                 if (token->ident == &default_ident) {
1334                         token = token->next;
1335                         goto default_statement;
1336                 }
1337                 if (token->ident == &case_ident) {
1338                         token = parse_expression(token->next, &stmt->case_expression);
1339                         if (match_op(token, SPECIAL_ELLIPSIS))
1340                                 token = parse_expression(token->next, &stmt->case_to);
1341 default_statement:
1342                         stmt->type = STMT_CASE;
1343                         token = expect(token, ':', "after default/case");
1344                         add_case_statement(stmt);
1345                         return statement(token, &stmt->case_statement);
1346                 }
1347                 if (token->ident == &switch_ident) {
1348                         stmt->type = STMT_SWITCH;
1349                         start_switch(stmt);
1350                         token = parens_expression(token->next, &stmt->switch_expression, "after 'switch'");
1351                         token = statement(token, &stmt->switch_statement);
1352                         end_switch(stmt);
1353                         return token;
1354                 }
1355                 if (token->ident == &for_ident)
1356                         return parse_for_statement(token, stmt);
1357
1358                 if (token->ident == &while_ident)
1359                         return parse_while_statement(token, stmt);
1360
1361                 if (token->ident == &do_ident)
1362                         return parse_do_statement(token, stmt);
1363
1364                 if (token->ident == &goto_ident) {
1365                         stmt->type = STMT_GOTO;
1366                         token = token->next;
1367                         if (match_op(token, '*')) {
1368                                 token = parse_expression(token->next, &stmt->goto_expression);
1369                                 add_statement(&function_computed_goto_list, stmt);
1370                         } else if (token_type(token) == TOKEN_IDENT) {
1371                                 stmt->goto_label = label_symbol(token);
1372                                 token = token->next;
1373                         } else {
1374                                 sparse_error(token->pos, "Expected identifier or goto expression");
1375                         }
1376                         return expect(token, ';', "at end of statement");
1377                 }
1378                 if (match_idents(token, &asm_ident, &__asm___ident, &__asm_ident, NULL)) {
1379                         return parse_asm(token->next, stmt);
1380                 }
1381                 if (token->ident == &__context___ident) {
1382                         stmt->type = STMT_CONTEXT;
1383                         token = parse_expression(token->next, &stmt->expression);
1384                         return expect(token, ';', "at end of statement");
1385                 }
1386                 if (token->ident == &__range___ident) {
1387                         stmt->type = STMT_RANGE;
1388                         token = assignment_expression(token->next, &stmt->range_expression);
1389                         token = expect(token, ',', "after range expression");
1390                         token = assignment_expression(token, &stmt->range_low);
1391                         token = expect(token, ',', "after low range");
1392                         token = assignment_expression(token, &stmt->range_high);
1393                         return expect(token, ';', "after range statement");
1394                 }
1395                 if (match_op(token->next, ':')) {
1396                         stmt->type = STMT_LABEL;
1397                         stmt->label_identifier = label_symbol(token);
1398                         return statement(token->next->next, &stmt->label_statement);
1399                 }
1400         }
1401
1402         if (match_op(token, '{')) {
1403                 stmt->type = STMT_COMPOUND;
1404                 start_symbol_scope();
1405                 token = compound_statement(token->next, stmt);
1406                 end_symbol_scope();
1407
1408                 return expect(token, '}', "at end of compound statement");
1409         }
1410
1411         stmt->type = STMT_EXPRESSION;
1412         return expression_statement(token, &stmt->expression);
1413 }
1414
1415 static struct token * statement_list(struct token *token, struct statement_list **list)
1416 {
1417         int seen_statement = 0;
1418         for (;;) {
1419                 struct statement * stmt;
1420                 if (eof_token(token))
1421                         break;
1422                 if (match_op(token, '}'))
1423                         break;
1424                 if (lookup_type(token)) {
1425                         if (seen_statement) {
1426                                 warning(token->pos, "mixing declarations and code");
1427                                 seen_statement = 0;
1428                         }
1429                         stmt = alloc_statement(token->pos, STMT_DECLARATION);
1430                         token = external_declaration(token, &stmt->declaration);
1431                 } else {
1432                         seen_statement = warn_on_mixed;
1433                         token = statement(token, &stmt);
1434                 }
1435                 add_statement(list, stmt);
1436         }
1437         return token;
1438 }
1439
1440 static struct token *parameter_type_list(struct token *token, struct symbol *fn, struct ident **p)
1441 {
1442         struct symbol_list **list = &fn->arguments;
1443
1444         if (match_op(token, ')')) {
1445                 // No warning for "void oink ();"
1446                 // Bug or feature: warns for "void oink () __attribute__ ((noreturn));"
1447                 if (p && !match_op(token->next, ';'))
1448                         warning(token->pos, "non-ANSI function declaration of function '%s'", show_ident(*p));
1449                 return token;
1450         }
1451
1452         for (;;) {
1453                 struct symbol *sym;
1454
1455                 if (match_op(token, SPECIAL_ELLIPSIS)) {
1456                         if (!*list)
1457                                 warning(token->pos, "variadic functions must have one named argument");
1458                         fn->variadic = 1;
1459                         token = token->next;
1460                         break;
1461                 }
1462
1463                 sym = alloc_symbol(token->pos, SYM_NODE);
1464                 token = parameter_declaration(token, &sym);
1465                 if (sym->ctype.base_type == &void_ctype) {
1466                         /* Special case: (void) */
1467                         if (!*list && !sym->ident)
1468                                 break;
1469                         warning(token->pos, "void parameter");
1470                 }
1471                 add_symbol(list, sym);
1472                 if (!match_op(token, ','))
1473                         break;
1474                 token = token->next;
1475
1476         }
1477         return token;
1478 }
1479
1480 struct token *compound_statement(struct token *token, struct statement *stmt)
1481 {
1482         token = statement_list(token, &stmt->stmts);
1483         return token;
1484 }
1485
1486 static struct expression *identifier_expression(struct token *token)
1487 {
1488         struct expression *expr = alloc_expression(token->pos, EXPR_IDENTIFIER);
1489         expr->expr_ident = token->ident;
1490         return expr;
1491 }
1492
1493 static struct expression *index_expression(struct expression *from, struct expression *to)
1494 {
1495         int idx_from, idx_to;
1496         struct expression *expr = alloc_expression(from->pos, EXPR_INDEX);
1497
1498         idx_from = get_expression_value(from);
1499         idx_to = idx_from;
1500         if (to) {
1501                 idx_to = get_expression_value(to);
1502                 if (idx_to < idx_from || idx_from < 0)
1503                         warning(from->pos, "nonsense array initializer index range");
1504         }
1505         expr->idx_from = idx_from;
1506         expr->idx_to = idx_to;
1507         return expr;
1508 }
1509
1510 static struct token *single_initializer(struct expression **ep, struct token *token)
1511 {
1512         int expect_equal = 0;
1513         struct token *next = token->next;
1514         struct expression **tail = ep;
1515         int nested;
1516
1517         *ep = NULL;
1518
1519         if ((token_type(token) == TOKEN_IDENT) && match_op(next, ':')) {
1520                 struct expression *expr = identifier_expression(token);
1521                 warning(token->pos, "obsolete struct initializer, use C99 syntax");
1522                 token = initializer(&expr->ident_expression, next->next);
1523                 if (expr->ident_expression)
1524                         *ep = expr;
1525                 return token;
1526         }
1527
1528         for (tail = ep, nested = 0; ; nested++, next = token->next) {
1529                 if (match_op(token, '.') && (token_type(next) == TOKEN_IDENT)) {
1530                         struct expression *expr = identifier_expression(next);
1531                         *tail = expr;
1532                         tail = &expr->ident_expression;
1533                         expect_equal = 1;
1534                         token = next->next;
1535                 } else if (match_op(token, '[')) {
1536                         struct expression *from = NULL, *to = NULL, *expr;
1537                         token = constant_expression(token->next, &from);
1538                         if (!from) {
1539                                 sparse_error(token->pos, "Expected constant expression");
1540                                 break;
1541                         }
1542                         if (match_op(token, SPECIAL_ELLIPSIS))
1543                                 token = constant_expression(token->next, &to);
1544                         expr = index_expression(from, to);
1545                         *tail = expr;
1546                         tail = &expr->idx_expression;
1547                         token = expect(token, ']', "at end of initializer index");
1548                         if (nested)
1549                                 expect_equal = 1;
1550                 } else {
1551                         break;
1552                 }
1553         }
1554         if (nested && !expect_equal) {
1555                 if (!match_op(token, '='))
1556                         warning(token->pos, "obsolete array initializer, use C99 syntax");
1557                 else
1558                         expect_equal = 1;
1559         }
1560         if (expect_equal)
1561                 token = expect(token, '=', "at end of initializer index");
1562
1563         token = initializer(tail, token);
1564         if (!*tail)
1565                 *ep = NULL;
1566         return token;
1567 }
1568
1569 static struct token *initializer_list(struct expression_list **list, struct token *token)
1570 {
1571         struct expression *expr;
1572
1573         for (;;) {
1574                 token = single_initializer(&expr, token);
1575                 if (!expr)
1576                         break;
1577                 add_expression(list, expr);
1578                 if (!match_op(token, ','))
1579                         break;
1580                 token = token->next;
1581         }
1582         return token;
1583 }
1584
1585 struct token *initializer(struct expression **tree, struct token *token)
1586 {
1587         if (match_op(token, '{')) {
1588                 struct expression *expr = alloc_expression(token->pos, EXPR_INITIALIZER);
1589                 *tree = expr;
1590                 token = initializer_list(&expr->expr_list, token->next);
1591                 return expect(token, '}', "at end of initializer");
1592         }
1593         return assignment_expression(token, tree);
1594 }
1595
1596 static void declare_argument(struct symbol *sym, struct symbol *fn)
1597 {
1598         if (!sym->ident) {
1599                 sparse_error(sym->pos, "no identifier for function argument");
1600                 return;
1601         }
1602         bind_symbol(sym, sym->ident, NS_SYMBOL);
1603 }
1604
1605 static struct token *parse_function_body(struct token *token, struct symbol *decl,
1606         struct symbol_list **list)
1607 {
1608         struct symbol_list **old_symbol_list;
1609         struct symbol *base_type = decl->ctype.base_type;
1610         struct statement *stmt, **p;
1611         struct symbol *arg;
1612
1613         old_symbol_list = function_symbol_list;
1614         if (decl->ctype.modifiers & MOD_INLINE) {
1615                 function_symbol_list = &decl->inline_symbol_list;
1616                 p = &base_type->inline_stmt;
1617         } else {
1618                 function_symbol_list = &decl->symbol_list;
1619                 p = &base_type->stmt;
1620         }
1621         function_computed_target_list = NULL;
1622         function_computed_goto_list = NULL;
1623
1624         if (decl->ctype.modifiers & MOD_EXTERN) {
1625                 if (!(decl->ctype.modifiers & MOD_INLINE))
1626                         warning(decl->pos, "function '%s' with external linkage has definition", show_ident(decl->ident));
1627         }
1628         if (!(decl->ctype.modifiers & MOD_STATIC))
1629                 decl->ctype.modifiers |= MOD_EXTERN;
1630
1631         stmt = start_function(decl);
1632
1633         *p = stmt;
1634         FOR_EACH_PTR (base_type->arguments, arg) {
1635                 declare_argument(arg, base_type);
1636         } END_FOR_EACH_PTR(arg);
1637
1638         token = compound_statement(token->next, stmt);
1639
1640         end_function(decl);
1641         if (!(decl->ctype.modifiers & MOD_INLINE))
1642                 add_symbol(list, decl);
1643         check_declaration(decl);
1644         function_symbol_list = old_symbol_list;
1645         if (function_computed_goto_list) {
1646                 if (!function_computed_target_list)
1647                         warning(decl->pos, "function '%s' has computed goto but no targets?", show_ident(decl->ident));
1648                 else {
1649                         struct statement *stmt;
1650                         FOR_EACH_PTR(function_computed_goto_list, stmt) {
1651                                 stmt->target_list = function_computed_target_list;
1652                         } END_FOR_EACH_PTR(stmt);
1653                 }
1654         }
1655         return expect(token, '}', "at end of function");
1656 }
1657
1658 static void promote_k_r_types(struct symbol *arg)
1659 {
1660         struct symbol *base = arg->ctype.base_type;
1661         if (base && base->ctype.base_type == &int_type && (base->ctype.modifiers & (MOD_CHAR | MOD_SHORT))) {
1662                 arg->ctype.base_type = &int_ctype;
1663         }
1664 }
1665
1666 static void apply_k_r_types(struct symbol_list *argtypes, struct symbol *fn)
1667 {
1668         struct symbol_list *real_args = fn->ctype.base_type->arguments;
1669         struct symbol *arg;
1670
1671         FOR_EACH_PTR(real_args, arg) {
1672                 struct symbol *type;
1673
1674                 /* This is quadratic in the number of arguments. We _really_ don't care */
1675                 FOR_EACH_PTR(argtypes, type) {
1676                         if (type->ident == arg->ident)
1677                                 goto match;
1678                 } END_FOR_EACH_PTR(type);
1679                 sparse_error(arg->pos, "missing type declaration for parameter '%s'", show_ident(arg->ident));
1680                 continue;
1681 match:
1682                 type->used = 1;
1683                 /* "char" and "short" promote to "int" */
1684                 promote_k_r_types(type);
1685
1686                 arg->ctype = type->ctype;
1687         } END_FOR_EACH_PTR(arg);
1688
1689         FOR_EACH_PTR(argtypes, arg) {
1690                 if (!arg->used)
1691                         warning(arg->pos, "nonsensical parameter declaration '%s'", show_ident(arg->ident));
1692         } END_FOR_EACH_PTR(arg);
1693
1694 }
1695
1696 static struct token *parse_k_r_arguments(struct token *token, struct symbol *decl,
1697         struct symbol_list **list)
1698 {
1699         struct symbol_list *args = NULL;
1700
1701         warning(token->pos, "non-ANSI definition of function '%s'", show_ident(decl->ident));
1702         do {
1703                 token = declaration_list(token, &args);
1704                 if (!match_op(token, ';')) {
1705                         sparse_error(token->pos, "expected ';' at end of parameter declaration");
1706                         break;
1707                 }
1708                 token = token->next;
1709         } while (lookup_type(token));
1710
1711         apply_k_r_types(args, decl);
1712
1713         if (!match_op(token, '{')) {
1714                 sparse_error(token->pos, "expected function body");
1715                 return token;
1716         }
1717         return parse_function_body(token, decl, list);
1718 }
1719
1720
1721 struct token *external_declaration(struct token *token, struct symbol_list **list)
1722 {
1723         struct ident *ident = NULL;
1724         struct symbol *decl;
1725         struct ctype ctype = { 0, };
1726         struct symbol *base_type;
1727         int is_typedef;
1728
1729         /* Top-level inline asm? */
1730         if (match_idents(token, &asm_ident, &__asm___ident, &__asm_ident, NULL)) {
1731                 struct symbol *anon = alloc_symbol(token->pos, SYM_NODE);
1732                 struct symbol *fn = alloc_symbol(token->pos, SYM_FN);
1733                 struct statement *stmt;
1734
1735                 anon->ctype.base_type = fn;
1736                 stmt = alloc_statement(token->pos, STMT_NONE);
1737                 fn->stmt = stmt;
1738
1739                 token = parse_asm(token->next, stmt);
1740
1741                 add_symbol(list, anon);
1742                 return token;
1743         }
1744
1745         /* Parse declaration-specifiers, if any */
1746         token = declaration_specifiers(token, &ctype, 0);
1747         decl = alloc_symbol(token->pos, SYM_NODE);
1748         decl->ctype = ctype;
1749         token = declarator(token, decl, &ident);
1750
1751         /* Just a type declaration? */
1752         if (!ident)
1753                 return expect(token, ';', "end of type declaration");
1754
1755         /* type define declaration? */
1756         is_typedef = (ctype.modifiers & MOD_TYPEDEF) != 0;
1757
1758         /* Typedef's don't have meaningful storage */
1759         if (is_typedef) {
1760                 ctype.modifiers &= ~MOD_STORAGE;
1761                 decl->ctype.modifiers &= ~MOD_STORAGE;
1762                 decl->ctype.modifiers |= MOD_USERTYPE;
1763         }
1764
1765         bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
1766
1767         base_type = decl->ctype.base_type;
1768
1769         if (is_typedef) {
1770                 if (base_type && !base_type->ident)
1771                         base_type->ident = ident;
1772         } else if (base_type && base_type->type == SYM_FN) {
1773                 /* K&R argument declaration? */
1774                 if (lookup_type(token))
1775                         return parse_k_r_arguments(token, decl, list);
1776                 if (match_op(token, '{'))
1777                         return parse_function_body(token, decl, list);
1778
1779                 if (!(decl->ctype.modifiers & MOD_STATIC))
1780                         decl->ctype.modifiers |= MOD_EXTERN;
1781         } else if (base_type == &void_ctype && !(decl->ctype.modifiers & MOD_EXTERN)) {
1782                 sparse_error(token->pos, "void declaration");
1783         }
1784
1785         for (;;) {
1786                 if (!is_typedef && match_op(token, '=')) {
1787                         if (decl->ctype.modifiers & MOD_EXTERN) {
1788                                 warning(decl->pos, "symbol with external linkage has initializer");
1789                                 decl->ctype.modifiers &= ~MOD_EXTERN;
1790                         }
1791                         token = initializer(&decl->initializer, token->next);
1792                 }
1793                 if (!is_typedef) {
1794                         if (!(decl->ctype.modifiers & (MOD_EXTERN | MOD_INLINE))) {
1795                                 add_symbol(list, decl);
1796                                 fn_local_symbol(decl);
1797                         }
1798                 }
1799                 check_declaration(decl);
1800
1801                 if (!match_op(token, ','))
1802                         break;
1803
1804                 token = token->next;
1805                 ident = NULL;
1806                 decl = alloc_symbol(token->pos, SYM_NODE);
1807                 decl->ctype = ctype;
1808                 token = declaration_specifiers(token, &decl->ctype, 1);
1809                 token = declarator(token, decl, &ident);
1810                 if (!ident) {
1811                         sparse_error(token->pos, "expected identifier name in type definition");
1812                         return token;
1813                 }
1814
1815                 bind_symbol(decl, ident, is_typedef ? NS_TYPEDEF: NS_SYMBOL);
1816
1817                 /* Function declarations are automatically extern unless specifically static */
1818                 base_type = decl->ctype.base_type;
1819                 if (!is_typedef && base_type && base_type->type == SYM_FN) {
1820                         if (!(decl->ctype.modifiers & MOD_STATIC))
1821                                 decl->ctype.modifiers |= MOD_EXTERN;
1822                 }
1823         }
1824         return expect(token, ';', "at end of declaration");
1825 }