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
  19 #include "lib.h"
  20 #include "token.h"
  21 #include "parse.h"
  22 #include "symbol.h"
  23 #include "scope.h"
  24 #include "expression.h"
  25 #include "target.h"
  26
  27 static int match_oplist(int op, ...)
  28 {
  29         va_list args;
  30
  31         va_start(args, op);
  32         for (;;) {
  33                 int nextop = va_arg(args, int);
  34                 if (!nextop)
  35                         return 0;
  36                 if (op == nextop)
  37                         return 1;
  38         }
  39 }
  40
  41 static struct token *comma_expression(struct token *, struct expression **);
  42
  43 struct token *parens_expression(struct token *token, struct expression **expr, const char *where)
  44 {
  45         token = expect(token, '(', where);
  46         if (match_op(token, '{')) {
  47                 struct expression *e = alloc_expression(token->pos, EXPR_STATEMENT);
  48                 struct statement *stmt = alloc_statement(token->pos, STMT_COMPOUND);
  49                 *expr = e;
  50                 e->statement = stmt;
  51                 start_symbol_scope();
  52                 token = compound_statement(token->next, stmt);
  53                 end_symbol_scope();
  54                 token = expect(token, '}', "at end of statement expression");
  55         } else
  56                 token = parse_expression(token, expr);
  57         return expect(token, ')', where);
  58 }
  59
  60 static struct token *string_expression(struct token *token, struct expression *expr)
  61 {
  62         struct string *string = token->string;
  63         struct token *next = token->next;
  64
  65         if (token_type(next) == TOKEN_STRING) {
  66                 int totlen = string->length;
  67                 char *data;
  68
  69                 do {
  70                         totlen += next->string->length-1;
  71                         next = next->next;
  72                 } while (token_type(next) == TOKEN_STRING);
  73
  74                 string = __alloc_string(totlen);
  75                 string->length = totlen;
  76                 data = string->data;
  77                 next = token;
  78                 do {
  79                         struct string *s = next->string;
  80                         int len = s->length;
  81
  82                         next = next->next;
  83                         memcpy(data, s->data, len);
  84                         data += len-1;
  85                 } while (token_type(next) == TOKEN_STRING);
  86         }
  87         expr->string = string;
  88         return next;
  89 }
  90
  91 static void get_int_value(struct expression *expr, const char *str)
  92 {
  93         unsigned long long value = 0;
  94         unsigned int base = 10, digit, bits;
  95         unsigned long modifiers, extramod;
  96
  97         switch (str[0]) {
  98         case 'x':
  99                 base = 18;      // the -= 2 for the octal case will
 100                 str++;          // skip the 'x'
 101         /* fallthrough */
 102         case 'o':
 103                 str++;          // skip the 'o' or 'x/X'
 104                 base -= 2;      // the fall-through will make this 8
 105         }
 106         while ((digit = hexval(*str)) < base) {
 107                 value = value * base + digit;
 108                 str++;
 109         }
 110         modifiers = 0;
 111         for (;;) {
 112                 char c = *str++;
 113                 if (c == 'u' || c == 'U') {
 114                         modifiers |= MOD_UNSIGNED;
 115                         continue;
 116                 }
 117                 if (c == 'l' || c == 'L') {
 118                         if (modifiers & MOD_LONG)
 119                                 modifiers |= MOD_LONGLONG;
 120                         modifiers |= MOD_LONG;
 121                         continue;
 122                 }
 123                 break;
 124         }
 125
 126         bits = BITS_IN_LONGLONG;
 127         extramod = 0;
 128         if (!(modifiers & MOD_LONGLONG)) {
 129                 if (value & (~1ULL << (BITS_IN_LONG-1))) {
 130                         extramod = MOD_LONGLONG | MOD_LONG;
 131                 } else {
 132                         bits = BITS_IN_LONG;
 133                         if (!(modifiers & MOD_LONG)) {
 134                                 if (value & (~1ULL << (BITS_IN_INT-1))) {
 135                                         extramod = MOD_LONG;
 136                                 } else
 137                                         bits = BITS_IN_INT;
 138                         }
 139                 }
 140         }
 141         if (!(modifiers & MOD_UNSIGNED)) {
 142                 if (value & (1ULL << (bits-1))) {
 143                         extramod |= MOD_UNSIGNED;
 144                 }
 145         }
 146         if (extramod) {
 147                 /*
 148                  * Special case: "int" gets promoted directly to "long"
 149                  * for normal decimal numbers..
 150                  */
 151                 modifiers |= extramod;
 152                 if (base == 10 && modifiers == MOD_UNSIGNED) {
 153                         modifiers = MOD_LONG;
 154                         if (BITS_IN_LONG == BITS_IN_INT)
 155                                 modifiers = MOD_LONG | MOD_UNSIGNED;
 156                 }
 157
 158                 /* Hex or octal constants don't complain about missing signedness */
 159                 if (base == 10 || extramod != MOD_UNSIGNED)
 160                         warn(expr->pos, "value is so big it is%s%s%s",
 161                                 (modifiers & MOD_UNSIGNED) ? " unsigned":"",
 162                                 (modifiers & MOD_LONG) ? " long":"",
 163                                 (modifiers & MOD_LONGLONG) ? " long":"");
 164         }
 165
 166         expr->type = EXPR_VALUE;
 167         expr->ctype = ctype_integer(modifiers);
 168         expr->value = value;
 169 }
 170
 171 struct token *primary_expression(struct token *token, struct expression **tree)
 172 {
 173         struct expression *expr = NULL;
 174
 175         switch (token_type(token)) {
 176         static int fp_warned;
 177         case TOKEN_FP:
 178                 expr = alloc_expression(token->pos, EXPR_VALUE);
 179                 expr->ctype = &double_ctype;
 180                 expr->value = 0;
 181                 if (!fp_warned) {
 182                         warn(token->pos, "FP values not yet implemented");
 183                         fp_warned = 1;
 184                 }
 185                 token = token->next;
 186                 break;
 187
 188         case TOKEN_CHAR:
 189                 expr = alloc_expression(token->pos, EXPR_VALUE);
 190                 expr->ctype = &int_ctype;
 191                 expr->value = (unsigned char) token->character;
 192                 token = token->next;
 193                 break;
 194
 195         case TOKEN_INTEGER:
 196                 expr = alloc_expression(token->pos, EXPR_VALUE);
 197                 get_int_value(expr, token->integer);
 198                 token = token->next;
 199                 break;
 200
 201         case TOKEN_IDENT: {
 202                 struct symbol *sym = lookup_symbol(token->ident, NS_SYMBOL | NS_TYPEDEF);
 203                 struct token *next = token->next;
 204
 205                 expr = alloc_expression(token->pos, EXPR_SYMBOL);
 206
 207                 /*
 208                  * We support types as real first-class citizens, with type
 209                  * comparisons etc:
 210                  *
 211                  *      if (typeof(a) == int) ..
 212                  */
 213                 if (sym && sym->namespace == NS_TYPEDEF) {
 214                         warn(token->pos, "typename in expression");
 215                         sym = NULL;
 216                 }
 217                 expr->symbol_name = token->ident;
 218                 expr->symbol = sym;
 219                 token = next;
 220                 break;
 221         }
 222
 223         case TOKEN_STRING: {
 224                 expr = alloc_expression(token->pos, EXPR_STRING);
 225                 token = string_expression(token, expr);
 226                 break;
 227         }
 228
 229         case TOKEN_SPECIAL:
 230                 if (token->special == '(') {
 231                         expr = alloc_expression(token->pos, EXPR_PREOP);
 232                         expr->op = '(';
 233                         token = parens_expression(token, &expr->unop, "in expression");
 234                         break;
 235                 }
 236                 if (token->special == '[' && lookup_type(token->next)) {
 237                         expr = alloc_expression(token->pos, EXPR_TYPE);
 238                         token = typename(token->next, &expr->symbol);
 239                         token = expect(token, ']', "in type expression");
 240                         break;
 241                 }
 242
 243         default:
 244                 ;
 245         }
 246         *tree = expr;
 247         return token;
 248 }
 249
 250 static struct token *expression_list(struct token *token, struct expression_list **list)
 251 {
 252         while (!match_op(token, ')')) {
 253                 struct expression *expr = NULL;
 254                 token = assignment_expression(token, &expr);
 255                 if (!expr)
 256                         break;
 257                 add_expression(list, expr);
 258                 if (!match_op(token, ','))
 259                         break;
 260                 token = token->next;
 261         }
 262         return token;
 263 }
 264
 265 /*
 266  * extend to deal with the ambiguous C grammar for parsing
 267  * a cast expressions followed by an initializer.
 268  */
 269 static struct token *postfix_expression(struct token *token, struct expression **tree, struct expression *cast_init_expr)
 270 {
 271         struct expression *expr = cast_init_expr;
 272
 273         if (!expr)
 274                 token = primary_expression(token, &expr);
 275
 276         while (expr && token_type(token) == TOKEN_SPECIAL) {
 277                 switch (token->special) {
 278                 case '[': {                     /* Array dereference */
 279                         struct expression *deref = alloc_expression(token->pos, EXPR_PREOP);
 280                         struct expression *add = alloc_expression(token->pos, EXPR_BINOP);
 281
 282                         deref->op = '*';
 283                         deref->unop = add;
 284
 285                         add->op = '+';
 286                         add->left = expr;
 287                         token = parse_expression(token->next, &add->right);
 288                         token = expect(token, ']', "at end of array dereference");
 289                         expr = deref;
 290                         continue;
 291                 }
 292                 case SPECIAL_INCREMENT:         /* Post-increment */
 293                 case SPECIAL_DECREMENT: {       /* Post-decrement */
 294                         struct expression *post = alloc_expression(token->pos, EXPR_POSTOP);
 295                         post->op = token->special;
 296                         post->unop = expr;
 297                         expr = post;
 298                         token = token->next;
 299                         continue;
 300                 }
 301                 case '.':                       /* Structure member dereference */
 302                 case SPECIAL_DEREFERENCE: {     /* Structure pointer member dereference */
 303                         struct expression *deref = alloc_expression(token->pos, EXPR_DEREF);
 304                         deref->op = token->special;
 305                         deref->deref = expr;
 306                         token = token->next;
 307                         if (token_type(token) != TOKEN_IDENT) {
 308                                 warn(token->pos, "Expected member name");
 309                                 break;
 310                         }
 311                         deref->member = token->ident;
 312                         token = token->next;
 313                         expr = deref;
 314                         continue;
 315                 }
 316
 317                 case '(': {                     /* Function call */
 318                         struct expression *call = alloc_expression(token->pos, EXPR_CALL);
 319                         call->op = '(';
 320                         call->fn = expr;
 321                         token = expression_list(token->next, &call->args);
 322                         token = expect(token, ')', "in function call");
 323                         expr = call;
 324                         continue;
 325                 }
 326
 327                 default:
 328                         break;
 329                 }
 330                 break;
 331         }
 332         *tree = expr;
 333         return token;
 334 }
 335
 336 static struct token *cast_expression(struct token *token, struct expression **tree);
 337 static struct token *unary_expression(struct token *token, struct expression **tree)
 338 {
 339         if (token_type(token) == TOKEN_IDENT &&
 340             (token->ident == &sizeof_ident ||
 341              token->ident == &__alignof___ident)) {
 342                 struct expression *sizeof_ex = alloc_expression(token->pos, EXPR_SIZEOF);
 343                 *tree = sizeof_ex;
 344                 tree = &sizeof_ex->unop;
 345                 token = token->next;
 346                 if (!match_op(token, '(') || !lookup_type(token->next))
 347                         return unary_expression(token, &sizeof_ex->cast_expression);
 348                 token = typename(token->next, &sizeof_ex->cast_type);
 349                 return expect(token, ')', "at end of sizeof type-name");
 350         }
 351
 352         if (token_type(token) == TOKEN_SPECIAL) {
 353                 if (match_oplist(token->special,
 354                     SPECIAL_INCREMENT, SPECIAL_DECREMENT,
 355                     '&', '*', '+', '-', '~', '!', 0)) {
 356                         struct expression *unary = alloc_expression(token->pos, EXPR_PREOP);
 357                         unary->op = token->special;
 358                         *tree = unary;
 359                         return cast_expression(token->next, &unary->unop);
 360                 }
 361
 362                 /* Gcc extension: &&label gives the address of a label */
 363                 if (match_op(token, SPECIAL_LOGICAL_AND) &&
 364                     token_type(token->next) == TOKEN_IDENT) {
 365                         struct expression *label = alloc_expression(token->pos, EXPR_LABEL);
 366                         label->label_symbol = label_symbol(token->next);
 367                         *tree = label;
 368                         return token->next->next;
 369                 }
 370
 371         }
 372
 373         return postfix_expression(token, tree, NULL);
 374 }
 375
 376 /*
 377  * Ambiguity: a '(' can be either a cast-expression or
 378  * a primary-expression depending on whether it is followed
 379  * by a type or not.
 380  *
 381  * additional ambiguity: a "cast expression" followed by
 382  * an initializer is really a postfix-expression.
 383  */
 384 static struct token *cast_expression(struct token *token, struct expression **tree)
 385 {
 386         if (match_op(token, '(')) {
 387                 struct token *next = token->next;
 388                 if (lookup_type(next)) {
 389                         struct expression *cast = alloc_expression(next->pos, EXPR_CAST);
 390                         struct symbol *sym;
 391
 392                         token = typename(next, &sym);
 393                         cast->cast_type = sym->ctype.base_type;
 394                         token = expect(token, ')', "at end of cast operator");
 395                         if (match_op(token, '{')) {
 396                                 token = initializer(&cast->cast_expression, token);
 397                                 return postfix_expression(token, tree, cast);
 398                         }
 399                         *tree = cast;
 400                         token = cast_expression(token, &cast->cast_expression);
 401                         return token;
 402                 }
 403         }
 404         return unary_expression(token, tree);
 405 }
 406
 407 /* Generic left-to-right binop parsing */
 408 static struct token *lr_binop_expression(struct token *token, struct expression **tree,
 409         enum expression_type type, struct token *(*inner)(struct token *, struct expression **), ...)
 410 {
 411         struct expression *left = NULL;
 412         struct token * next = inner(token, &left);
 413
 414         if (left) {
 415                 while (token_type(next) == TOKEN_SPECIAL) {
 416                         struct expression *top, *right = NULL;
 417                         int op = next->special;
 418                         va_list args;
 419
 420                         va_start(args, inner);
 421                         for (;;) {
 422                                 int nextop = va_arg(args, int);
 423                                 if (!nextop)
 424                                         goto out;
 425                                 if (op == nextop)
 426                                         break;
 427                         }
 428                         va_end(args);
 429                         top = alloc_expression(next->pos, type);
 430                         next = inner(next->next, &right);
 431                         if (!right) {
 432                                 warn(next->pos, "No right hand side of '%s'-expression", show_special(op));
 433                                 break;
 434                         }
 435                         top->op = op;
 436                         top->left = left;
 437                         top->right = right;
 438                         left = top;
 439                 }
 440         }
 441 out:
 442         *tree = left;
 443         return next;
 444 }
 445
 446 static struct token *multiplicative_expression(struct token *token, struct expression **tree)
 447 {
 448         return lr_binop_expression(token, tree, EXPR_BINOP, cast_expression, '*', '/', '%', 0);
 449 }
 450
 451 static struct token *additive_expression(struct token *token, struct expression **tree)
 452 {
 453         return lr_binop_expression(token, tree, EXPR_BINOP, multiplicative_expression, '+', '-', 0);
 454 }
 455
 456 static struct token *shift_expression(struct token *token, struct expression **tree)
 457 {
 458         return lr_binop_expression(token, tree, EXPR_BINOP, additive_expression, SPECIAL_LEFTSHIFT, SPECIAL_RIGHTSHIFT, 0);
 459 }
 460
 461 static struct token *relational_expression(struct token *token, struct expression **tree)
 462 {
 463         return lr_binop_expression(token, tree, EXPR_COMPARE, shift_expression, '<', '>', SPECIAL_LTE, SPECIAL_GTE, 0);
 464 }
 465
 466 static struct token *equality_expression(struct token *token, struct expression **tree)
 467 {
 468         return lr_binop_expression(token, tree, EXPR_COMPARE, relational_expression, SPECIAL_EQUAL, SPECIAL_NOTEQUAL, 0);
 469 }
 470
 471 static struct token *bitwise_and_expression(struct token *token, struct expression **tree)
 472 {
 473         return lr_binop_expression(token, tree, EXPR_BINOP, equality_expression, '&', 0);
 474 }
 475
 476 static struct token *bitwise_xor_expression(struct token *token, struct expression **tree)
 477 {
 478         return lr_binop_expression(token, tree, EXPR_BINOP, bitwise_and_expression, '^', 0);
 479 }
 480
 481 static struct token *bitwise_or_expression(struct token *token, struct expression **tree)
 482 {
 483         return lr_binop_expression(token, tree, EXPR_BINOP, bitwise_xor_expression, '|', 0);
 484 }
 485
 486 static struct token *logical_and_expression(struct token *token, struct expression **tree)
 487 {
 488         return lr_binop_expression(token, tree, EXPR_LOGICAL, bitwise_or_expression, SPECIAL_LOGICAL_AND, 0);
 489 }
 490
 491 static struct token *logical_or_expression(struct token *token, struct expression **tree)
 492 {
 493         return lr_binop_expression(token, tree, EXPR_LOGICAL, logical_and_expression, SPECIAL_LOGICAL_OR, 0);
 494 }
 495
 496 struct token *conditional_expression(struct token *token, struct expression **tree)
 497 {
 498         token = logical_or_expression(token, tree);
 499         if (match_op(token, '?')) {
 500                 struct expression *expr = alloc_expression(token->pos, EXPR_CONDITIONAL);
 501                 expr->op = token->special;
 502                 expr->left = *tree;
 503                 *tree = expr;
 504                 token = parse_expression(token->next, &expr->cond_true);
 505                 token = expect(token, ':', "in conditional expression");
 506                 token = conditional_expression(token, &expr->cond_false);
 507         }
 508         return token;
 509 }
 510
 511 struct token *assignment_expression(struct token *token, struct expression **tree)
 512 {
 513         token = conditional_expression(token, tree);
 514         if (token_type(token) == TOKEN_SPECIAL) {
 515                 static const int assignments[] = {
 516                         '=',
 517                         SPECIAL_ADD_ASSIGN, SPECIAL_SUB_ASSIGN,
 518                         SPECIAL_MUL_ASSIGN, SPECIAL_DIV_ASSIGN,
 519                         SPECIAL_MOD_ASSIGN, SPECIAL_SHL_ASSIGN,
 520                         SPECIAL_SHR_ASSIGN, SPECIAL_AND_ASSIGN,
 521                         SPECIAL_OR_ASSIGN,  SPECIAL_XOR_ASSIGN };
 522                 int i, op = token->special;
 523                 for (i = 0; i < sizeof(assignments)/sizeof(int); i++)
 524                         if (assignments[i] == op) {
 525                                 struct expression * expr = alloc_expression(token->pos, EXPR_ASSIGNMENT);
 526                                 expr->left = *tree;
 527                                 expr->op = op;
 528                                 *tree = expr;
 529                                 return assignment_expression(token->next, &expr->right);
 530                         }
 531         }
 532         return token;
 533 }
 534
 535 static struct token *comma_expression(struct token *token, struct expression **tree)
 536 {
 537         return lr_binop_expression(token, tree, EXPR_COMMA, assignment_expression, ',', 0);
 538 }
 539
 540 struct token *parse_expression(struct token *token, struct expression **tree)
 541 {
 542         return comma_expression(token,tree);
 543 }
 544
 545