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 #include <stdarg.h>
  12 #include <stdlib.h>
  13 #include <stdio.h>
  14 #include <string.h>
  15 #include <ctype.h>
  16 #include <unistd.h>
  17 #include <fcntl.h>
  18 #include <errno.h>
  19 #include <limits.h>
  20
  21 #include "lib.h"
  22 #include "token.h"
  23 #include "parse.h"
  24 #include "symbol.h"
  25 #include "scope.h"
  26 #include "expression.h"
  27 #include "target.h"
  28
  29 static int match_oplist(int op, ...)
  30 {
  31         va_list args;
  32
  33         va_start(args, op);
  34         for (;;) {
  35                 int nextop = va_arg(args, int);
  36                 if (!nextop)
  37                         return 0;
  38                 if (op == nextop)
  39                         return 1;
  40         }
  41 }
  42
  43 static struct token *comma_expression(struct token *, struct expression **);
  44
  45 struct token *parens_expression(struct token *token, struct expression **expr, const char *where)
  46 {
  47         token = expect(token, '(', where);
  48         if (match_op(token, '{')) {
  49                 struct expression *e = alloc_expression(token->pos, EXPR_STATEMENT);
  50                 struct statement *stmt = alloc_statement(token->pos, STMT_COMPOUND);
  51                 *expr = e;
  52                 e->statement = stmt;
  53                 start_symbol_scope();
  54                 token = compound_statement(token->next, stmt);
  55                 end_symbol_scope();
  56                 token = expect(token, '}', "at end of statement expression");
  57         } else
  58                 token = parse_expression(token, expr);
  59         return expect(token, ')', where);
  60 }
  61
  62 static struct token *string_expression(struct token *token, struct expression *expr)
  63 {
  64         struct string *string = token->string;
  65         struct token *next = token->next;
  66
  67         if (token_type(next) == TOKEN_STRING) {
  68                 int totlen = string->length;
  69                 char *data;
  70
  71                 do {
  72                         totlen += next->string->length-1;
  73                         next = next->next;
  74                 } while (token_type(next) == TOKEN_STRING);
  75
  76                 string = __alloc_string(totlen);
  77                 string->length = totlen;
  78                 data = string->data;
  79                 next = token;
  80                 do {
  81                         struct string *s = next->string;
  82                         int len = s->length;
  83
  84                         next = next->next;
  85                         memcpy(data, s->data, len);
  86                         data += len-1;
  87                 } while (token_type(next) == TOKEN_STRING);
  88         }
  89         expr->string = string;
  90         return next;
  91 }
  92
  93 static void get_fp_value(struct expression *expr, struct token *token)
  94 {
  95         static int fp_warned;
  96
  97         expr->ctype = &double_ctype;
  98         expr->value = 0;
  99         if (!fp_warned) {
 100                 warn(token->pos, "FP values not yet implemented");
 101                 fp_warned = 1;
 102         }
 103 }
 104
 105 #ifndef ULLONG_MAX
 106 #define ULLONG_MAX (~0ULL)
 107 #endif
 108
 109 static void get_number_value(struct expression *expr, struct token *token)
 110 {
 111         const char *str = token->number;
 112         unsigned long long value;
 113         char *end;
 114         unsigned long modifiers = 0;
 115         int overflow = 0, do_warn = 0;
 116         int try_unsigned = 1;
 117         int bits;
 118
 119         errno = 0;
 120         value = strtoull(str, &end, 0);
 121         if (end == str)
 122                 goto Enoint;
 123         if (value == ULLONG_MAX && errno == ERANGE)
 124                 overflow = 1;
 125         while (1) {
 126                 unsigned long added;
 127                 char c = *end++;
 128                 if (!c) {
 129                         break;
 130                 } else if (c == 'u' || c == 'U') {
 131                         added = MOD_UNSIGNED;
 132                 } else if (c == 'l' || c == 'L') {
 133                         added = MOD_LONG;
 134                         if (*end == c) {
 135                                 added |= MOD_LONGLONG;
 136                                 end++;
 137                         }
 138                 } else
 139                         goto Enoint;
 140                 if (modifiers & added)
 141                         goto Enoint;
 142                 modifiers |= added;
 143         }
 144         if (overflow)
 145                 goto Eoverflow;
 146         /* OK, it's a valid integer */
 147         /* decimals can be unsigned only if directly specified as such */
 148         if (str[0] != '0' && !(modifiers & MOD_UNSIGNED))
 149                 try_unsigned = 0;
 150         if (!(modifiers & MOD_LONG)) {
 151                 bits = bits_in_int - 1;
 152                 if (!(value & (~1ULL << bits))) {
 153                         if (!(value & (1ULL << bits))) {
 154                                 goto got_it;
 155                         } else if (try_unsigned) {
 156                                 modifiers |= MOD_UNSIGNED;
 157                                 goto got_it;
 158                         }
 159                 }
 160                 modifiers |= MOD_LONG;
 161                 do_warn = 1;
 162         }
 163         if (!(modifiers & MOD_LONGLONG)) {
 164                 bits = bits_in_long - 1;
 165                 if (!(value & (~1ULL << bits))) {
 166                         if (!(value & (1ULL << bits))) {
 167                                 goto got_it;
 168                         } else if (try_unsigned) {
 169                                 modifiers |= MOD_UNSIGNED;
 170                                 goto got_it;
 171                         }
 172                         do_warn |= 2;
 173                 }
 174                 modifiers |= MOD_LONGLONG;
 175                 do_warn |= 1;
 176         }
 177         bits = bits_in_longlong - 1;
 178         if (value & (~1ULL << bits))
 179                 goto Eoverflow;
 180         if (!(value & (1ULL << bits)))
 181                 goto got_it;
 182         if (!try_unsigned)
 183                 warn(expr->pos, "decimal constant %s is too big for long long",
 184                         show_token(token));
 185         modifiers |= MOD_UNSIGNED;
 186 got_it:
 187         if (do_warn)
 188                 warn(expr->pos, "constant %s is so big it is%s%s%s",
 189                         show_token(token),
 190                         (modifiers & MOD_UNSIGNED) ? " unsigned":"",
 191                         (modifiers & MOD_LONG) ? " long":"",
 192                         (modifiers & MOD_LONGLONG) ? " long":"");
 193         if (do_warn & 2)
 194                 warn(expr->pos,
 195                         "decimal constant %s is between LONG_MAX and ULONG_MAX."
 196                         " For C99 that means long long, C90 compilers are very "
 197                         "likely to produce unsigned long (and a warning) here",
 198                         show_token(token));
 199         expr->type = EXPR_VALUE;
 200         expr->ctype = ctype_integer(modifiers);
 201         expr->value = value;
 202         return;
 203 Eoverflow:
 204         error(expr->pos, "constant %s is too big even for unsigned long long",
 205                         show_token(token));
 206 Enoint:
 207         get_fp_value(expr, token);
 208 }
 209
 210 struct token *primary_expression(struct token *token, struct expression **tree)
 211 {
 212         struct expression *expr = NULL;
 213
 214         switch (token_type(token)) {
 215         case TOKEN_CHAR:
 216                 expr = alloc_expression(token->pos, EXPR_VALUE);
 217                 expr->ctype = &int_ctype;
 218                 expr->value = (unsigned char) token->character;
 219                 token = token->next;
 220                 break;
 221
 222         case TOKEN_NUMBER:
 223                 expr = alloc_expression(token->pos, EXPR_VALUE);
 224                 get_number_value(expr, token);
 225                 token = token->next;
 226                 break;
 227
 228         case TOKEN_IDENT: {
 229                 struct symbol *sym = lookup_symbol(token->ident, NS_SYMBOL | NS_TYPEDEF);
 230                 struct token *next = token->next;
 231
 232                 expr = alloc_expression(token->pos, EXPR_SYMBOL);
 233
 234                 /*
 235                  * We support types as real first-class citizens, with type
 236                  * comparisons etc:
 237                  *
 238                  *      if (typeof(a) == int) ..
 239                  */
 240                 if (sym && sym->namespace == NS_TYPEDEF) {
 241                         warn(token->pos, "typename in expression");
 242                         sym = NULL;
 243                 }
 244                 expr->symbol_name = token->ident;
 245                 expr->symbol = sym;
 246                 token = next;
 247                 break;
 248         }
 249
 250         case TOKEN_STRING: {
 251                 expr = alloc_expression(token->pos, EXPR_STRING);
 252                 token = string_expression(token, expr);
 253                 break;
 254         }
 255
 256         case TOKEN_SPECIAL:
 257                 if (token->special == '(') {
 258                         expr = alloc_expression(token->pos, EXPR_PREOP);
 259                         expr->op = '(';
 260                         token = parens_expression(token, &expr->unop, "in expression");
 261                         break;
 262                 }
 263                 if (token->special == '[' && lookup_type(token->next)) {
 264                         expr = alloc_expression(token->pos, EXPR_TYPE);
 265                         token = typename(token->next, &expr->symbol);
 266                         token = expect(token, ']', "in type expression");
 267                         break;
 268                 }
 269
 270         default:
 271                 ;
 272         }
 273         *tree = expr;
 274         return token;
 275 }
 276
 277 static struct token *expression_list(struct token *token, struct expression_list **list)
 278 {
 279         while (!match_op(token, ')')) {
 280                 struct expression *expr = NULL;
 281                 token = assignment_expression(token, &expr);
 282                 if (!expr)
 283                         break;
 284                 add_expression(list, expr);
 285                 if (!match_op(token, ','))
 286                         break;
 287                 token = token->next;
 288         }
 289         return token;
 290 }
 291
 292 /*
 293  * extend to deal with the ambiguous C grammar for parsing
 294  * a cast expressions followed by an initializer.
 295  */
 296 static struct token *postfix_expression(struct token *token, struct expression **tree, struct expression *cast_init_expr)
 297 {
 298         struct expression *expr = cast_init_expr;
 299
 300         if (!expr)
 301                 token = primary_expression(token, &expr);
 302
 303         while (expr && token_type(token) == TOKEN_SPECIAL) {
 304                 switch (token->special) {
 305                 case '[': {                     /* Array dereference */
 306                         struct expression *deref = alloc_expression(token->pos, EXPR_PREOP);
 307                         struct expression *add = alloc_expression(token->pos, EXPR_BINOP);
 308
 309                         deref->op = '*';
 310                         deref->unop = add;
 311
 312                         add->op = '+';
 313                         add->left = expr;
 314                         token = parse_expression(token->next, &add->right);
 315                         token = expect(token, ']', "at end of array dereference");
 316                         expr = deref;
 317                         continue;
 318                 }
 319                 case SPECIAL_INCREMENT:         /* Post-increment */
 320                 case SPECIAL_DECREMENT: {       /* Post-decrement */
 321                         struct expression *post = alloc_expression(token->pos, EXPR_POSTOP);
 322                         post->op = token->special;
 323                         post->unop = expr;
 324                         expr = post;
 325                         token = token->next;
 326                         continue;
 327                 }
 328                 case SPECIAL_DEREFERENCE: {     /* Structure pointer member dereference */
 329                         /* "x->y" is just shorthand for "(*x).y" */
 330                         struct expression *inner = alloc_expression(token->pos, EXPR_PREOP);
 331                         inner->op = '*';
 332                         inner->unop = expr;
 333                         expr = inner;
 334                 }
 335                 /* Fallthrough!! */
 336                 case '.': {                     /* Structure member dereference */
 337                         struct expression *deref = alloc_expression(token->pos, EXPR_DEREF);
 338                         deref->op = '.';
 339                         deref->deref = expr;
 340                         token = token->next;
 341                         if (token_type(token) != TOKEN_IDENT) {
 342                                 warn(token->pos, "Expected member name");
 343                                 break;
 344                         }
 345                         deref->member = token->ident;
 346                         token = token->next;
 347                         expr = deref;
 348                         continue;
 349                 }
 350
 351                 case '(': {                     /* Function call */
 352                         struct expression *call = alloc_expression(token->pos, EXPR_CALL);
 353                         call->op = '(';
 354                         call->fn = expr;
 355                         token = expression_list(token->next, &call->args);
 356                         token = expect(token, ')', "in function call");
 357                         expr = call;
 358                         continue;
 359                 }
 360
 361                 default:
 362                         break;
 363                 }
 364                 break;
 365         }
 366         *tree = expr;
 367         return token;
 368 }
 369
 370 static struct token *cast_expression(struct token *token, struct expression **tree);
 371 static struct token *unary_expression(struct token *token, struct expression **tree)
 372 {
 373         if (token_type(token) == TOKEN_IDENT &&
 374             (token->ident == &sizeof_ident ||
 375              token->ident == &__alignof___ident)) {
 376                 struct expression *sizeof_ex = alloc_expression(token->pos, EXPR_SIZEOF);
 377                 *tree = sizeof_ex;
 378                 tree = &sizeof_ex->unop;
 379                 token = token->next;
 380                 if (!match_op(token, '(') || !lookup_type(token->next))
 381                         return unary_expression(token, &sizeof_ex->cast_expression);
 382                 token = typename(token->next, &sizeof_ex->cast_type);
 383                 return expect(token, ')', "at end of sizeof type-name");
 384         }
 385
 386         if (token_type(token) == TOKEN_SPECIAL) {
 387                 if (match_oplist(token->special,
 388                     SPECIAL_INCREMENT, SPECIAL_DECREMENT,
 389                     '&', '*', '+', '-', '~', '!', 0)) {
 390                         struct expression *unop;
 391                         struct expression *unary;
 392                         struct token *next;
 393
 394                         next = cast_expression(token->next, &unop);
 395                         if (!unop) {
 396                                 warn(token->pos, "Syntax error in unary expression");
 397                                 return next;
 398                         }
 399                         unary = alloc_expression(token->pos, EXPR_PREOP);
 400                         unary->op = token->special;
 401                         unary->unop = unop;
 402                         *tree = unary;
 403                         return next;
 404                 }
 405
 406                 /* Gcc extension: &&label gives the address of a label */
 407                 if (match_op(token, SPECIAL_LOGICAL_AND) &&
 408                     token_type(token->next) == TOKEN_IDENT) {
 409                         struct expression *label = alloc_expression(token->pos, EXPR_LABEL);
 410                         label->label_symbol = label_symbol(token->next);
 411                         *tree = label;
 412                         return token->next->next;
 413                 }
 414
 415         }
 416
 417         return postfix_expression(token, tree, NULL);
 418 }
 419
 420 /*
 421  * Ambiguity: a '(' can be either a cast-expression or
 422  * a primary-expression depending on whether it is followed
 423  * by a type or not.
 424  *
 425  * additional ambiguity: a "cast expression" followed by
 426  * an initializer is really a postfix-expression.
 427  */
 428 static struct token *cast_expression(struct token *token, struct expression **tree)
 429 {
 430         if (match_op(token, '(')) {
 431                 struct token *next = token->next;
 432                 if (lookup_type(next)) {
 433                         struct expression *cast = alloc_expression(next->pos, EXPR_CAST);
 434                         struct symbol *sym;
 435
 436                         token = typename(next, &sym);
 437                         cast->cast_type = sym;
 438                         token = expect(token, ')', "at end of cast operator");
 439                         if (match_op(token, '{')) {
 440                                 token = initializer(&cast->cast_expression, token);
 441                                 return postfix_expression(token, tree, cast);
 442                         }
 443                         *tree = cast;
 444                         token = cast_expression(token, &cast->cast_expression);
 445                         return token;
 446                 }
 447         }
 448         return unary_expression(token, tree);
 449 }
 450
 451 /*
 452  * Generic left-to-right binop parsing
 453  *
 454  * This _really_ needs to be inlined, because that makes the inner
 455  * function call statically deterministic rather than a totally
 456  * unpredictable indirect call. But gcc-3 is so "clever" that it
 457  * doesn't do so by default even when you tell it to inline it.
 458  *
 459  * Making it a macro avoids the inlining problem, and also means
 460  * that we can pass in the op-comparison as an expression rather
 461  * than create a data structure for it.
 462  */
 463
 464 #define LR_BINOP_EXPRESSION(token, tree, type, inner, compare)          \
 465         struct expression *left = NULL;                                 \
 466         struct token * next = inner(token, &left);                      \
 467                                                                         \
 468         if (left) {                                                     \
 469                 while (token_type(next) == TOKEN_SPECIAL) {             \
 470                         struct expression *top, *right = NULL;          \
 471                         int op = next->special;                         \
 472                                                                         \
 473                         if (!(compare))                                 \
 474                                 goto out;                               \
 475                         top = alloc_expression(next->pos, type);        \
 476                         next = inner(next->next, &right);               \
 477                         if (!right) {                                   \
 478                                 warn(next->pos, "No right hand side of '%s'-expression", show_special(op));     \
 479                                 break;                                  \
 480                         }                                               \
 481                         top->op = op;                                   \
 482                         top->left = left;                               \
 483                         top->right = right;                             \
 484                         left = top;                                     \
 485                 }                                                       \
 486         }                                                               \
 487 out:                                                                    \
 488         *tree = left;                                                   \
 489         return next;                                                    \
 490
 491
 492 static struct token *multiplicative_expression(struct token *token, struct expression **tree)
 493 {
 494         LR_BINOP_EXPRESSION(
 495                 token, tree, EXPR_BINOP, cast_expression,
 496                 (op == '*') || (op == '/') || (op == '%')
 497         );
 498 }
 499
 500 static struct token *additive_expression(struct token *token, struct expression **tree)
 501 {
 502         LR_BINOP_EXPRESSION(
 503                 token, tree, EXPR_BINOP, multiplicative_expression,
 504                 (op == '+') || (op == '-')
 505         );
 506 }
 507
 508 static struct token *shift_expression(struct token *token, struct expression **tree)
 509 {
 510         LR_BINOP_EXPRESSION(
 511                 token, tree, EXPR_BINOP, additive_expression,
 512                 (op == SPECIAL_LEFTSHIFT) || (op == SPECIAL_RIGHTSHIFT)
 513         );
 514 }
 515
 516 static struct token *relational_expression(struct token *token, struct expression **tree)
 517 {
 518         LR_BINOP_EXPRESSION(
 519                 token, tree, EXPR_COMPARE, shift_expression,
 520                 (op == '<') || (op == '>') ||
 521                 (op == SPECIAL_LTE) || (op == SPECIAL_GTE)
 522         );
 523 }
 524
 525 static struct token *equality_expression(struct token *token, struct expression **tree)
 526 {
 527         LR_BINOP_EXPRESSION(
 528                 token, tree, EXPR_COMPARE, relational_expression,
 529                 (op == SPECIAL_EQUAL) || (op == SPECIAL_NOTEQUAL)
 530         );
 531 }
 532
 533 static struct token *bitwise_and_expression(struct token *token, struct expression **tree)
 534 {
 535         LR_BINOP_EXPRESSION(
 536                 token, tree, EXPR_BINOP, equality_expression,
 537                 (op == '&')
 538         );
 539 }
 540
 541 static struct token *bitwise_xor_expression(struct token *token, struct expression **tree)
 542 {
 543         LR_BINOP_EXPRESSION(
 544                 token, tree, EXPR_BINOP, bitwise_and_expression,
 545                 (op == '^')
 546         );
 547 }
 548
 549 static struct token *bitwise_or_expression(struct token *token, struct expression **tree)
 550 {
 551         LR_BINOP_EXPRESSION(
 552                 token, tree, EXPR_BINOP, bitwise_xor_expression,
 553                 (op == '|')
 554         );
 555 }
 556
 557 static struct token *logical_and_expression(struct token *token, struct expression **tree)
 558 {
 559         LR_BINOP_EXPRESSION(
 560                 token, tree, EXPR_LOGICAL, bitwise_or_expression,
 561                 (op == SPECIAL_LOGICAL_AND)
 562         );
 563 }
 564
 565 static struct token *logical_or_expression(struct token *token, struct expression **tree)
 566 {
 567         LR_BINOP_EXPRESSION(
 568                 token, tree, EXPR_LOGICAL, logical_and_expression,
 569                 (op == SPECIAL_LOGICAL_OR)
 570         );
 571 }
 572
 573 struct token *conditional_expression(struct token *token, struct expression **tree)
 574 {
 575         token = logical_or_expression(token, tree);
 576         if (match_op(token, '?')) {
 577                 struct expression *expr = alloc_expression(token->pos, EXPR_CONDITIONAL);
 578                 expr->op = token->special;
 579                 expr->left = *tree;
 580                 *tree = expr;
 581                 token = parse_expression(token->next, &expr->cond_true);
 582                 token = expect(token, ':', "in conditional expression");
 583                 token = conditional_expression(token, &expr->cond_false);
 584         }
 585         return token;
 586 }
 587
 588 struct token *assignment_expression(struct token *token, struct expression **tree)
 589 {
 590         token = conditional_expression(token, tree);
 591         if (token_type(token) == TOKEN_SPECIAL) {
 592                 static const int assignments[] = {
 593                         '=',
 594                         SPECIAL_ADD_ASSIGN, SPECIAL_SUB_ASSIGN,
 595                         SPECIAL_MUL_ASSIGN, SPECIAL_DIV_ASSIGN,
 596                         SPECIAL_MOD_ASSIGN, SPECIAL_SHL_ASSIGN,
 597                         SPECIAL_SHR_ASSIGN, SPECIAL_AND_ASSIGN,
 598                         SPECIAL_OR_ASSIGN,  SPECIAL_XOR_ASSIGN };
 599                 int i, op = token->special;
 600                 for (i = 0; i < sizeof(assignments)/sizeof(int); i++)
 601                         if (assignments[i] == op) {
 602                                 struct expression * expr = alloc_expression(token->pos, EXPR_ASSIGNMENT);
 603                                 expr->left = *tree;
 604                                 expr->op = op;
 605                                 *tree = expr;
 606                                 return assignment_expression(token->next, &expr->right);
 607                         }
 608         }
 609         return token;
 610 }
 611
 612 static struct token *comma_expression(struct token *token, struct expression **tree)
 613 {
 614         LR_BINOP_EXPRESSION(
 615                 token, tree, EXPR_COMMA, assignment_expression,
 616                 (op == ',')
 617         );
 618 }
 619
 620 struct token *parse_expression(struct token *token, struct expression **tree)
 621 {
 622         return comma_expression(token,tree);
 623 }
 624
 625