expression.c

   1 /*
   2  * sparse/expression.c
   3  *
   4  * Copyright (C) 2003 Transmeta Corp.
   5  *               2003 Linus Torvalds
   6  *
   7  *  Licensed under the Open Software License version 1.1
   8  *
   9  * This is the expression parsing part of parsing C.
  10  */
  11 #define _ISOC99_SOURCE
  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 <errno.h>
  20 #include <limits.h>
  21
  22 #include "lib.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 static int match_oplist(int op, ...)
  31 {
  32         va_list args;
  33
  34         va_start(args, op);
  35         for (;;) {
  36                 int nextop = va_arg(args, int);
  37                 if (!nextop)
  38                         return 0;
  39                 if (op == nextop)
  40                         return 1;
  41         }
  42 }
  43
  44 static struct token *comma_expression(struct token *, struct expression **);
  45
  46 struct token *parens_expression(struct token *token, struct expression **expr, const char *where)
  47 {
  48         token = expect(token, '(', where);
  49         if (match_op(token, '{')) {
  50                 struct expression *e = alloc_expression(token->pos, EXPR_STATEMENT);
  51                 struct statement *stmt = alloc_statement(token->pos, STMT_COMPOUND);
  52                 *expr = e;
  53                 e->statement = stmt;
  54                 start_symbol_scope();
  55                 token = compound_statement(token->next, stmt);
  56                 end_symbol_scope();
  57                 token = expect(token, '}', "at end of statement expression");
  58         } else
  59                 token = parse_expression(token, expr);
  60         return expect(token, ')', where);
  61 }
  62
  63 static struct token *string_expression(struct token *token, struct expression *expr)
  64 {
  65         struct string *string = token->string;
  66         struct token *next = token->next;
  67
  68         if (token_type(next) == TOKEN_STRING) {
  69                 int totlen = string->length-1;
  70                 char *data;
  71
  72                 do {
  73                         totlen += next->string->length-1;
  74                         next = next->next;
  75                 } while (token_type(next) == TOKEN_STRING);
  76
  77                 if (totlen > MAX_STRING) {
  78                         warn(token->pos, "trying to concatenate %d-character string (%d bytes max)", totlen, MAX_STRING);
  79                         totlen = MAX_STRING;
  80                 }
  81
  82                 string = __alloc_string(totlen+1);
  83                 string->length = totlen+1;
  84                 data = string->data;
  85                 next = token;
  86                 do {
  87                         struct string *s = next->string;
  88                         int len = s->length-1;
  89
  90                         if (len > totlen)
  91                                 len = totlen;
  92                         totlen -= len;
  93
  94                         next = next->next;
  95                         memcpy(data, s->data, len);
  96                         data += len;
  97                 } while (token_type(next) == TOKEN_STRING);
  98                 *data = '\0';
  99         }
 100         expr->string = string;
 101         return next;
 102 }
 103
 104 #ifndef ULLONG_MAX
 105 #define ULLONG_MAX (~0ULL)
 106 #endif
 107
 108 static void get_number_value(struct expression *expr, struct token *token)
 109 {
 110         const char *str = token->number;
 111         unsigned long long value;
 112         char *end;
 113         unsigned long modifiers = 0;
 114         int overflow = 0, do_warn = 0;
 115         int try_unsigned = 1;
 116         int bits;
 117
 118         errno = 0;
 119         value = strtoull(str, &end, 0);
 120         if (end == str)
 121                 goto Float;
 122         if (value == ULLONG_MAX && errno == ERANGE)
 123                 overflow = 1;
 124         while (1) {
 125                 unsigned long added;
 126                 char c = *end++;
 127                 if (!c) {
 128                         break;
 129                 } else if (c == 'u' || c == 'U') {
 130                         added = MOD_UNSIGNED;
 131                 } else if (c == 'l' || c == 'L') {
 132                         added = MOD_LONG;
 133                         if (*end == c) {
 134                                 added |= MOD_LONGLONG;
 135                                 end++;
 136                         }
 137                 } else
 138                         goto Float;
 139                 if (modifiers & added)
 140                         goto Enoint;
 141                 modifiers |= added;
 142         }
 143         if (overflow)
 144                 goto Eoverflow;
 145         /* OK, it's a valid integer */
 146         /* decimals can be unsigned only if directly specified as such */
 147         if (str[0] != '0' && !(modifiers & MOD_UNSIGNED))
 148                 try_unsigned = 0;
 149         if (!(modifiers & MOD_LONG)) {
 150                 bits = bits_in_int - 1;
 151                 if (!(value & (~1ULL << bits))) {
 152                         if (!(value & (1ULL << bits))) {
 153                                 goto got_it;
 154                         } else if (try_unsigned) {
 155                                 modifiers |= MOD_UNSIGNED;
 156                                 goto got_it;
 157                         }
 158                 }
 159                 modifiers |= MOD_LONG;
 160                 do_warn = 1;
 161         }
 162         if (!(modifiers & MOD_LONGLONG)) {
 163                 bits = bits_in_long - 1;
 164                 if (!(value & (~1ULL << bits))) {
 165                         if (!(value & (1ULL << bits))) {
 166                                 goto got_it;
 167                         } else if (try_unsigned) {
 168                                 modifiers |= MOD_UNSIGNED;
 169                                 goto got_it;
 170                         }
 171                         do_warn |= 2;
 172                 }
 173                 modifiers |= MOD_LONGLONG;
 174                 do_warn |= 1;
 175         }
 176         bits = bits_in_longlong - 1;
 177         if (value & (~1ULL << bits))
 178                 goto Eoverflow;
 179         if (!(value & (1ULL << bits)))
 180                 goto got_it;
 181         if (!try_unsigned)
 182                 warn(expr->pos, "decimal constant %s is too big for long long",
 183                         show_token(token));
 184         modifiers |= MOD_UNSIGNED;
 185 got_it:
 186         if (do_warn)
 187                 warn(expr->pos, "constant %s is so big it is%s%s%s",
 188                         show_token(token),
 189                         (modifiers & MOD_UNSIGNED) ? " unsigned":"",
 190                         (modifiers & MOD_LONG) ? " long":"",
 191                         (modifiers & MOD_LONGLONG) ? " long":"");
 192         if (do_warn & 2)
 193                 warn(expr->pos,
 194                         "decimal constant %s is between LONG_MAX and ULONG_MAX."
 195                         " For C99 that means long long, C90 compilers are very "
 196                         "likely to produce unsigned long (and a warning) here",
 197                         show_token(token));
 198         expr->type = EXPR_VALUE;
 199         expr->ctype = ctype_integer(modifiers);
 200         expr->value = value;
 201         return;
 202 Eoverflow:
 203         error(expr->pos, "constant %s is too big even for unsigned long long",
 204                         show_token(token));
 205         return;
 206 Float:
 207         expr->fvalue = strtold(str, &end);
 208         if (str == end)
 209                 goto Enoint;
 210
 211         if (*end && end[1])
 212                 goto Enoint;
 213
 214         if (*end == 'f' || *end == 'F')
 215                 expr->ctype = &float_ctype;
 216         else if (*end == 'l' || *end == 'L')
 217                 expr->ctype = &ldouble_ctype;
 218         else if (!*end)
 219                 expr->ctype = &double_ctype;
 220         else
 221                 goto Enoint;
 222
 223         expr->type = EXPR_FVALUE;
 224         return;
 225
 226 Enoint:
 227         error(expr->pos, "constant %s is not a valid number", show_token(token));
 228 }
 229
 230 struct token *primary_expression(struct token *token, struct expression **tree)
 231 {
 232         struct expression *expr = NULL;
 233
 234         switch (token_type(token)) {
 235         case TOKEN_CHAR:
 236                 expr = alloc_expression(token->pos, EXPR_VALUE);
 237                 expr->ctype = &int_ctype;
 238                 expr->value = (unsigned char) token->character;
 239                 token = token->next;
 240                 break;
 241
 242         case TOKEN_NUMBER:
 243                 expr = alloc_expression(token->pos, EXPR_VALUE);
 244                 get_number_value(expr, token);
 245                 token = token->next;
 246                 break;
 247
 248         case TOKEN_IDENT: {
 249                 struct symbol *sym = lookup_symbol(token->ident, NS_SYMBOL | NS_TYPEDEF);
 250                 struct token *next = token->next;
 251
 252                 expr = alloc_expression(token->pos, EXPR_SYMBOL);
 253
 254                 /*
 255                  * We support types as real first-class citizens, with type
 256                  * comparisons etc:
 257                  *
 258                  *      if (typeof(a) == int) ..
 259                  */
 260                 if (sym && sym->namespace == NS_TYPEDEF) {
 261                         warn(token->pos, "typename in expression");
 262                         sym = NULL;
 263                 }
 264                 expr->symbol_name = token->ident;
 265                 expr->symbol = sym;
 266                 token = next;
 267                 break;
 268         }
 269
 270         case TOKEN_STRING: {
 271                 expr = alloc_expression(token->pos, EXPR_STRING);
 272                 token = string_expression(token, expr);
 273                 break;
 274         }
 275
 276         case TOKEN_SPECIAL:
 277                 if (token->special == '(') {
 278                         expr = alloc_expression(token->pos, EXPR_PREOP);
 279                         expr->op = '(';
 280                         token = parens_expression(token, &expr->unop, "in expression");
 281                         break;
 282                 }
 283                 if (token->special == '[' && lookup_type(token->next)) {
 284                         expr = alloc_expression(token->pos, EXPR_TYPE);
 285                         token = typename(token->next, &expr->symbol);
 286                         token = expect(token, ']', "in type expression");
 287                         break;
 288                 }
 289
 290         default:
 291                 ;
 292         }
 293         *tree = expr;
 294         return token;
 295 }
 296
 297 static struct token *expression_list(struct token *token, struct expression_list **list)
 298 {
 299         while (!match_op(token, ')')) {
 300                 struct expression *expr = NULL;
 301                 token = assignment_expression(token, &expr);
 302                 if (!expr)
 303                         break;
 304                 add_expression(list, expr);
 305                 if (!match_op(token, ','))
 306                         break;
 307                 token = token->next;
 308         }
 309         return token;
 310 }
 311
 312 /*
 313  * extend to deal with the ambiguous C grammar for parsing
 314  * a cast expressions followed by an initializer.
 315  */
 316 static struct token *postfix_expression(struct token *token, struct expression **tree, struct expression *cast_init_expr)
 317 {
 318         struct expression *expr = cast_init_expr;
 319
 320         if (!expr)
 321                 token = primary_expression(token, &expr);
 322
 323         while (expr && token_type(token) == TOKEN_SPECIAL) {
 324                 switch (token->special) {
 325                 case '[': {                     /* Array dereference */
 326                         struct expression *deref = alloc_expression(token->pos, EXPR_PREOP);
 327                         struct expression *add = alloc_expression(token->pos, EXPR_BINOP);
 328
 329                         deref->op = '*';
 330                         deref->unop = add;
 331
 332                         add->op = '+';
 333                         add->left = expr;
 334                         token = parse_expression(token->next, &add->right);
 335                         token = expect(token, ']', "at end of array dereference");
 336                         expr = deref;
 337                         continue;
 338                 }
 339                 case SPECIAL_INCREMENT:         /* Post-increment */
 340                 case SPECIAL_DECREMENT: {       /* Post-decrement */
 341                         struct expression *post = alloc_expression(token->pos, EXPR_POSTOP);
 342                         post->op = token->special;
 343                         post->unop = expr;
 344                         expr = post;
 345                         token = token->next;
 346                         continue;
 347                 }
 348                 case SPECIAL_DEREFERENCE: {     /* Structure pointer member dereference */
 349                         /* "x->y" is just shorthand for "(*x).y" */
 350                         struct expression *inner = alloc_expression(token->pos, EXPR_PREOP);
 351                         inner->op = '*';
 352                         inner->unop = expr;
 353                         expr = inner;
 354                 }
 355                 /* Fallthrough!! */
 356                 case '.': {                     /* Structure member dereference */
 357                         struct expression *deref = alloc_expression(token->pos, EXPR_DEREF);
 358                         deref->op = '.';
 359                         deref->deref = expr;
 360                         token = token->next;
 361                         if (token_type(token) != TOKEN_IDENT) {
 362                                 warn(token->pos, "Expected member name");
 363                                 break;
 364                         }
 365                         deref->member = token->ident;
 366                         token = token->next;
 367                         expr = deref;
 368                         continue;
 369                 }
 370
 371                 case '(': {                     /* Function call */
 372                         struct expression *call = alloc_expression(token->pos, EXPR_CALL);
 373                         call->op = '(';
 374                         call->fn = expr;
 375                         token = expression_list(token->next, &call->args);
 376                         token = expect(token, ')', "in function call");
 377                         expr = call;
 378                         continue;
 379                 }
 380
 381                 default:
 382                         break;
 383                 }
 384                 break;
 385         }
 386         *tree = expr;
 387         return token;
 388 }
 389
 390 static struct token *cast_expression(struct token *token, struct expression **tree);
 391 static struct token *unary_expression(struct token *token, struct expression **tree)
 392 {
 393         if (token_type(token) == TOKEN_IDENT) {
 394                 if (token->ident == &sizeof_ident) {
 395                         struct expression *sizeof_ex
 396                                 = alloc_expression(token->pos, EXPR_SIZEOF);
 397                         *tree = sizeof_ex;
 398                         tree = &sizeof_ex->unop;
 399                         token = token->next;
 400                         if (!match_op(token, '(') || !lookup_type(token->next))
 401                                 return unary_expression(token, &sizeof_ex->cast_expression);
 402                         token = typename(token->next, &sizeof_ex->cast_type);
 403
 404                         if (!match_op(token, ')'))
 405                                 return expect(token, ')', "at end of sizeof type-name");
 406
 407                         token = token->next;
 408                         /*
 409                          * C99 ambiguity: the typename might have been the beginning
 410                          * of a typed initializer expression..
 411                          */
 412                         if (match_op(token, '{'))
 413                                 token = initializer(&sizeof_ex->cast_expression, token);
 414                         return token;
 415                 } else if (token->ident == &__alignof___ident) {
 416                         struct expression *alignof_ex
 417                                 = alloc_expression(token->pos, EXPR_ALIGNOF);
 418                         *tree = alignof_ex;
 419                         tree = &alignof_ex->unop;
 420                         token = token->next;
 421                         if (!match_op(token, '(') || !lookup_type(token->next))
 422                                 return unary_expression(token, &alignof_ex->cast_expression);
 423                         token = typename(token->next, &alignof_ex->cast_type);
 424                         return expect(token, ')', "at end of alignof type-name");
 425                 }
 426         }
 427
 428         if (token_type(token) == TOKEN_SPECIAL) {
 429                 if (match_oplist(token->special,
 430                     SPECIAL_INCREMENT, SPECIAL_DECREMENT,
 431                     '&', '*', '+', '-', '~', '!', 0)) {
 432                         struct expression *unop;
 433                         struct expression *unary;
 434                         struct token *next;
 435
 436                         next = cast_expression(token->next, &unop);
 437                         if (!unop) {
 438                                 warn(token->pos, "Syntax error in unary expression");
 439                                 return next;
 440                         }
 441                         unary = alloc_expression(token->pos, EXPR_PREOP);
 442                         unary->op = token->special;
 443                         unary->unop = unop;
 444                         *tree = unary;
 445                         return next;
 446                 }
 447
 448                 /* Gcc extension: &&label gives the address of a label */
 449                 if (match_op(token, SPECIAL_LOGICAL_AND) &&
 450                     token_type(token->next) == TOKEN_IDENT) {
 451                         struct expression *label = alloc_expression(token->pos, EXPR_LABEL);
 452                         label->label_symbol = label_symbol(token->next);
 453                         *tree = label;
 454                         return token->next->next;
 455                 }
 456
 457         }
 458
 459         return postfix_expression(token, tree, NULL);
 460 }
 461
 462 /*
 463  * Ambiguity: a '(' can be either a cast-expression or
 464  * a primary-expression depending on whether it is followed
 465  * by a type or not.
 466  *
 467  * additional ambiguity: a "cast expression" followed by
 468  * an initializer is really a postfix-expression.
 469  */
 470 static struct token *cast_expression(struct token *token, struct expression **tree)
 471 {
 472         if (match_op(token, '(')) {
 473                 struct token *next = token->next;
 474                 if (lookup_type(next)) {
 475                         struct expression *cast = alloc_expression(next->pos, EXPR_CAST);
 476                         struct symbol *sym;
 477
 478                         token = typename(next, &sym);
 479                         cast->cast_type = sym;
 480                         token = expect(token, ')', "at end of cast operator");
 481                         if (match_op(token, '{')) {
 482                                 token = initializer(&cast->cast_expression, token);
 483                                 return postfix_expression(token, tree, cast);
 484                         }
 485                         *tree = cast;
 486                         token = cast_expression(token, &cast->cast_expression);
 487                         return token;
 488                 }
 489         }
 490         return unary_expression(token, tree);
 491 }
 492
 493 /*
 494  * Generic left-to-right binop parsing
 495  *
 496  * This _really_ needs to be inlined, because that makes the inner
 497  * function call statically deterministic rather than a totally
 498  * unpredictable indirect call. But gcc-3 is so "clever" that it
 499  * doesn't do so by default even when you tell it to inline it.
 500  *
 501  * Making it a macro avoids the inlining problem, and also means
 502  * that we can pass in the op-comparison as an expression rather
 503  * than create a data structure for it.
 504  */
 505
 506 #define LR_BINOP_EXPRESSION(token, tree, type, inner, compare)          \
 507         struct expression *left = NULL;                                 \
 508         struct token * next = inner(token, &left);                      \
 509                                                                         \
 510         if (left) {                                                     \
 511                 while (token_type(next) == TOKEN_SPECIAL) {             \
 512                         struct expression *top, *right = NULL;          \
 513                         int op = next->special;                         \
 514                                                                         \
 515                         if (!(compare))                                 \
 516                                 goto out;                               \
 517                         top = alloc_expression(next->pos, type);        \
 518                         next = inner(next->next, &right);               \
 519                         if (!right) {                                   \
 520                                 warn(next->pos, "No right hand side of '%s'-expression", show_special(op));     \
 521                                 break;                                  \
 522                         }                                               \
 523                         top->op = op;                                   \
 524                         top->left = left;                               \
 525                         top->right = right;                             \
 526                         left = top;                                     \
 527                 }                                                       \
 528         }                                                               \
 529 out:                                                                    \
 530         *tree = left;                                                   \
 531         return next;                                                    \
 532
 533
 534 static struct token *multiplicative_expression(struct token *token, struct expression **tree)
 535 {
 536         LR_BINOP_EXPRESSION(
 537                 token, tree, EXPR_BINOP, cast_expression,
 538                 (op == '*') || (op == '/') || (op == '%')
 539         );
 540 }
 541
 542 static struct token *additive_expression(struct token *token, struct expression **tree)
 543 {
 544         LR_BINOP_EXPRESSION(
 545                 token, tree, EXPR_BINOP, multiplicative_expression,
 546                 (op == '+') || (op == '-')
 547         );
 548 }
 549
 550 static struct token *shift_expression(struct token *token, struct expression **tree)
 551 {
 552         LR_BINOP_EXPRESSION(
 553                 token, tree, EXPR_BINOP, additive_expression,
 554                 (op == SPECIAL_LEFTSHIFT) || (op == SPECIAL_RIGHTSHIFT)
 555         );
 556 }
 557
 558 static struct token *relational_expression(struct token *token, struct expression **tree)
 559 {
 560         LR_BINOP_EXPRESSION(
 561                 token, tree, EXPR_COMPARE, shift_expression,
 562                 (op == '<') || (op == '>') ||
 563                 (op == SPECIAL_LTE) || (op == SPECIAL_GTE)
 564         );
 565 }
 566
 567 static struct token *equality_expression(struct token *token, struct expression **tree)
 568 {
 569         LR_BINOP_EXPRESSION(
 570                 token, tree, EXPR_COMPARE, relational_expression,
 571                 (op == SPECIAL_EQUAL) || (op == SPECIAL_NOTEQUAL)
 572         );
 573 }
 574
 575 static struct token *bitwise_and_expression(struct token *token, struct expression **tree)
 576 {
 577         LR_BINOP_EXPRESSION(
 578                 token, tree, EXPR_BINOP, equality_expression,
 579                 (op == '&')
 580         );
 581 }
 582
 583 static struct token *bitwise_xor_expression(struct token *token, struct expression **tree)
 584 {
 585         LR_BINOP_EXPRESSION(
 586                 token, tree, EXPR_BINOP, bitwise_and_expression,
 587                 (op == '^')
 588         );
 589 }
 590
 591 static struct token *bitwise_or_expression(struct token *token, struct expression **tree)
 592 {
 593         LR_BINOP_EXPRESSION(
 594                 token, tree, EXPR_BINOP, bitwise_xor_expression,
 595                 (op == '|')
 596         );
 597 }
 598
 599 static struct token *logical_and_expression(struct token *token, struct expression **tree)
 600 {
 601         LR_BINOP_EXPRESSION(
 602                 token, tree, EXPR_LOGICAL, bitwise_or_expression,
 603                 (op == SPECIAL_LOGICAL_AND)
 604         );
 605 }
 606
 607 static struct token *logical_or_expression(struct token *token, struct expression **tree)
 608 {
 609         LR_BINOP_EXPRESSION(
 610                 token, tree, EXPR_LOGICAL, logical_and_expression,
 611                 (op == SPECIAL_LOGICAL_OR)
 612         );
 613 }
 614
 615 struct token *conditional_expression(struct token *token, struct expression **tree)
 616 {
 617         token = logical_or_expression(token, tree);
 618         if (match_op(token, '?')) {
 619                 struct expression *expr = alloc_expression(token->pos, EXPR_CONDITIONAL);
 620                 expr->op = token->special;
 621                 expr->left = *tree;
 622                 *tree = expr;
 623                 token = parse_expression(token->next, &expr->cond_true);
 624                 token = expect(token, ':', "in conditional expression");
 625                 token = conditional_expression(token, &expr->cond_false);
 626         }
 627         return token;
 628 }
 629
 630 struct token *assignment_expression(struct token *token, struct expression **tree)
 631 {
 632         token = conditional_expression(token, tree);
 633         if (token_type(token) == TOKEN_SPECIAL) {
 634                 static const int assignments[] = {
 635                         '=',
 636                         SPECIAL_ADD_ASSIGN, SPECIAL_SUB_ASSIGN,
 637                         SPECIAL_MUL_ASSIGN, SPECIAL_DIV_ASSIGN,
 638                         SPECIAL_MOD_ASSIGN, SPECIAL_SHL_ASSIGN,
 639                         SPECIAL_SHR_ASSIGN, SPECIAL_AND_ASSIGN,
 640                         SPECIAL_OR_ASSIGN,  SPECIAL_XOR_ASSIGN };
 641                 int i, op = token->special;
 642                 for (i = 0; i < sizeof(assignments)/sizeof(int); i++)
 643                         if (assignments[i] == op) {
 644                                 struct expression * expr = alloc_expression(token->pos, EXPR_ASSIGNMENT);
 645                                 expr->left = *tree;
 646                                 expr->op = op;
 647                                 *tree = expr;
 648                                 return assignment_expression(token->next, &expr->right);
 649                         }
 650         }
 651         return token;
 652 }
 653
 654 static struct token *comma_expression(struct token *token, struct expression **tree)
 655 {
 656         LR_BINOP_EXPRESSION(
 657                 token, tree, EXPR_COMMA, assignment_expression,
 658                 (op == ',')
 659         );
 660 }
 661
 662 struct token *parse_expression(struct token *token, struct expression **tree)
 663 {
 664         return comma_expression(token,tree);
 665 }
 666
 667