expression.c

   1 /*
   2  * sparse/expression.c
   3  *
   4  * Copyright (C) 2003 Transmeta Corp.
   5  *
   6  *  Licensed under the Open Software License version 1.1
   7  *
   8  * This is the expression parsing part of parsing C.
   9  */
  10 #include <stdarg.h>
  11 #include <stdlib.h>
  12 #include <stdio.h>
  13 #include <string.h>
  14 #include <ctype.h>
  15 #include <unistd.h>
  16 #include <fcntl.h>
  17
  18 #include "lib.h"
  19 #include "token.h"
  20 #include "parse.h"
  21 #include "symbol.h"
  22 #include "scope.h"
  23 #include "expression.h"
  24 #include "target.h"
  25
  26 static int match_oplist(int op, ...)
  27 {
  28         va_list args;
  29
  30         va_start(args, op);
  31         for (;;) {
  32                 int nextop = va_arg(args, int);
  33                 if (!nextop)
  34                         return 0;
  35                 if (op == nextop)
  36                         return 1;
  37         }
  38 }
  39
  40 static struct token *comma_expression(struct token *, struct expression **);
  41
  42 struct token *parens_expression(struct token *token, struct expression **expr, const char *where)
  43 {
  44         token = expect(token, '(', where);
  45         if (match_op(token, '{')) {
  46                 struct expression *e = alloc_expression(token->pos, EXPR_STATEMENT);
  47                 struct statement *stmt = alloc_statement(token->pos, STMT_COMPOUND);
  48                 *expr = e;
  49                 e->statement = stmt;
  50                 start_symbol_scope();
  51                 token = compound_statement(token->next, stmt);
  52                 end_symbol_scope();
  53                 token = expect(token, '}', "at end of statement expression");
  54         } else
  55                 token = parse_expression(token, expr);
  56         return expect(token, ')', where);
  57 }
  58
  59 static struct token *string_expression(struct token *token, struct expression *expr)
  60 {
  61         struct string *string = token->string;
  62         struct token *next = token->next;
  63
  64         if (token_type(next) == TOKEN_STRING) {
  65                 int totlen = string->length;
  66                 char *data;
  67
  68                 do {
  69                         totlen += next->string->length-1;
  70                         next = next->next;
  71                 } while (token_type(next) == TOKEN_STRING);
  72
  73                 string = __alloc_string(totlen);
  74                 string->length = totlen;
  75                 data = string->data;
  76                 next = token;
  77                 do {
  78                         struct string *s = next->string;
  79                         int len = s->length;
  80
  81                         next = next->next;
  82                         memcpy(data, s->data, len);
  83                         data += len-1;
  84                 } while (token_type(next) == TOKEN_STRING);
  85         }
  86         expr->string = string;
  87         return next;
  88 }
  89
  90 static void get_int_value(struct expression *expr, const char *str)
  91 {
  92         unsigned long long value = 0;
  93         unsigned int base = 10, digit, bits;
  94         unsigned long modifiers, extramod;
  95
  96         switch (str[0]) {
  97         case 'x':
  98                 base = 18;      // the -= 2 for the octal case will
  99                 str++;          // skip the 'x'
 100         /* fallthrough */
 101         case 'o':
 102                 str++;          // skip the 'o' or 'x/X'
 103                 base -= 2;      // the fall-through will make this 8
 104         }
 105         while ((digit = hexval(*str)) < base) {
 106                 value = value * base + digit;
 107                 str++;
 108         }
 109         modifiers = 0;
 110         for (;;) {
 111                 char c = *str++;
 112                 if (c == 'u' || c == 'U') {
 113                         modifiers |= MOD_UNSIGNED;
 114                         continue;
 115                 }
 116                 if (c == 'l' || c == 'L') {
 117                         if (modifiers & MOD_LONG)
 118                                 modifiers |= MOD_LONGLONG;
 119                         modifiers |= MOD_LONG;
 120                         continue;
 121                 }
 122                 break;
 123         }
 124
 125         bits = BITS_IN_LONGLONG;
 126         extramod = 0;
 127         if (!(modifiers & MOD_LONGLONG)) {
 128                 if (value & (~1ULL << (BITS_IN_LONG-1))) {
 129                         extramod = MOD_LONGLONG | MOD_LONG;
 130                 } else {
 131                         bits = BITS_IN_LONG;
 132                         if (!(modifiers & MOD_LONG)) {
 133                                 if (value & (~1ULL << (BITS_IN_INT-1))) {
 134                                         extramod = MOD_LONG;
 135                                 } else
 136                                         bits = BITS_IN_INT;
 137                         }
 138                 }
 139         }
 140         if (!(modifiers & MOD_UNSIGNED)) {
 141                 if (value & (1ULL << (bits-1))) {
 142                         extramod |= MOD_UNSIGNED;
 143                 }
 144         }
 145         if (extramod) {
 146                 /*
 147                  * Special case: "int" gets promoted directly to "long"
 148                  * for normal decimal numbers..
 149                  */
 150                 modifiers |= extramod;
 151                 if (base == 10 && modifiers == MOD_UNSIGNED) {
 152                         modifiers = MOD_LONG;
 153                         if (BITS_IN_LONG == BITS_IN_INT)
 154                                 modifiers = MOD_LONG | MOD_UNSIGNED;
 155                 }
 156                 warn(expr->pos, "value is so big it is%s%s%s",
 157                         (modifiers & MOD_UNSIGNED) ? " unsigned":"",
 158                         (modifiers & MOD_LONG) ? " long":"",
 159                         (modifiers & MOD_LONGLONG) ? " long":"");
 160         }
 161
 162         expr->type = EXPR_VALUE;
 163         expr->ctype = ctype_integer(modifiers);
 164         expr->value = value;
 165 }
 166
 167 struct token *primary_expression(struct token *token, struct expression **tree)
 168 {
 169         struct expression *expr = NULL;
 170
 171         switch (token_type(token)) {
 172         case TOKEN_FP:
 173                 expr = alloc_expression(token->pos, EXPR_VALUE);
 174                 expr->ctype = &double_ctype;
 175                 expr->value = 0;
 176                 warn(token->pos, "FP values not yet implemented");
 177                 token = token->next;
 178                 break;
 179
 180         case TOKEN_CHAR:
 181                 expr = alloc_expression(token->pos, EXPR_VALUE);
 182                 expr->ctype = &int_ctype;
 183                 expr->value = (unsigned char) token->character;
 184                 token = token->next;
 185                 break;
 186
 187         case TOKEN_INTEGER:
 188                 expr = alloc_expression(token->pos, EXPR_VALUE);
 189                 get_int_value(expr, token->integer);
 190                 token = token->next;
 191                 break;
 192
 193         case TOKEN_IDENT: {
 194                 expr = alloc_expression(token->pos, EXPR_SYMBOL);
 195                 expr->symbol_name = token->ident;
 196                 expr->symbol = lookup_symbol(token->ident, NS_SYMBOL);
 197                 token = token->next;
 198                 break;
 199         }
 200
 201         case TOKEN_STRING: {
 202                 expr = alloc_expression(token->pos, EXPR_STRING);
 203                 token = string_expression(token, expr);
 204                 break;
 205         }
 206
 207         case TOKEN_SPECIAL:
 208                 if (token->special == '(') {
 209                         expr = alloc_expression(token->pos, EXPR_PREOP);
 210                         expr->op = '(';
 211                         token = parens_expression(token, &expr->unop, "in expression");
 212                         break;
 213                 }
 214         default:
 215                 ;
 216         }
 217         *tree = expr;
 218         return token;
 219 }
 220
 221 static struct token *expression_list(struct token *token, struct expression_list **list)
 222 {
 223         while (!match_op(token, ')')) {
 224                 struct expression *expr = NULL;
 225                 token = assignment_expression(token, &expr);
 226                 if (!expr)
 227                         break;
 228                 add_expression(list, expr);
 229                 if (!match_op(token, ','))
 230                         break;
 231                 token = token->next;
 232         }
 233         return token;
 234 }
 235
 236 static struct token *postfix_expression(struct token *token, struct expression **tree)
 237 {
 238         struct expression *expr = NULL;
 239
 240         token = primary_expression(token, &expr);
 241         while (expr && token_type(token) == TOKEN_SPECIAL) {
 242                 switch (token->special) {
 243                 case '[': {                     /* Array dereference */
 244                         struct expression *deref = alloc_expression(token->pos, EXPR_PREOP);
 245                         struct expression *add = alloc_expression(token->pos, EXPR_BINOP);
 246
 247                         deref->op = '*';
 248                         deref->unop = add;
 249
 250                         add->op = '+';
 251                         add->left = expr;
 252                         token = parse_expression(token->next, &add->right);
 253                         token = expect(token, ']', "at end of array dereference");
 254                         expr = deref;
 255                         continue;
 256                 }
 257                 case SPECIAL_INCREMENT:         /* Post-increment */
 258                 case SPECIAL_DECREMENT: {       /* Post-decrement */
 259                         struct expression *post = alloc_expression(token->pos, EXPR_POSTOP);
 260                         post->op = token->special;
 261                         post->unop = expr;
 262                         expr = post;
 263                         token = token->next;
 264                         continue;
 265                 }
 266                 case '.':                       /* Structure member dereference */
 267                 case SPECIAL_DEREFERENCE: {     /* Structure pointer member dereference */
 268                         struct expression *deref = alloc_expression(token->pos, EXPR_DEREF);
 269                         deref->op = token->special;
 270                         deref->deref = expr;
 271                         token = token->next;
 272                         if (token_type(token) != TOKEN_IDENT) {
 273                                 warn(token->pos, "Expected member name");
 274                                 break;
 275                         }
 276                         deref->member = token->ident;
 277                         token = token->next;
 278                         expr = deref;
 279                         continue;
 280                 }
 281
 282                 case '(': {                     /* Function call */
 283                         struct expression *call = alloc_expression(token->pos, EXPR_CALL);
 284                         call->op = '(';
 285                         call->fn = expr;
 286                         token = expression_list(token->next, &call->args);
 287                         token = expect(token, ')', "in function call");
 288                         expr = call;
 289                         continue;
 290                 }
 291
 292                 default:
 293                         break;
 294                 }
 295                 break;
 296         }
 297         *tree = expr;
 298         return token;
 299 }
 300
 301 static struct token *cast_expression(struct token *token, struct expression **tree);
 302 static struct token *unary_expression(struct token *token, struct expression **tree)
 303 {
 304         if (token_type(token) == TOKEN_IDENT &&
 305             (token->ident == &sizeof_ident ||
 306              token->ident == &__alignof___ident)) {
 307                 struct expression *sizeof_ex = alloc_expression(token->pos, EXPR_SIZEOF);
 308                 *tree = sizeof_ex;
 309                 tree = &sizeof_ex->unop;
 310                 token = token->next;
 311                 if (!match_op(token, '(') || !lookup_type(token->next))
 312                         return unary_expression(token, &sizeof_ex->cast_expression);
 313                 token = typename(token->next, &sizeof_ex->cast_type);
 314                 return expect(token, ')', "at end of sizeof type-name");
 315         }
 316
 317         if (token_type(token) == TOKEN_SPECIAL) {
 318                 if (match_oplist(token->special,
 319                     SPECIAL_INCREMENT, SPECIAL_DECREMENT,
 320                     '&', '*', '+', '-', '~', '!', 0)) {
 321                         struct expression *unary = alloc_expression(token->pos, EXPR_PREOP);
 322                         unary->op = token->special;
 323                         *tree = unary;
 324                         return cast_expression(token->next, &unary->unop);
 325                 }
 326         }
 327
 328         return postfix_expression(token, tree);
 329 }
 330
 331 /*
 332  * Ambiguity: a '(' can be either a cast-expression or
 333  * a primary-expression depending on whether it is followed
 334  * by a type or not.
 335  */
 336 static struct token *cast_expression(struct token *token, struct expression **tree)
 337 {
 338         if (match_op(token, '(')) {
 339                 struct token *next = token->next;
 340                 if (lookup_type(next)) {
 341                         struct expression *cast = alloc_expression(next->pos, EXPR_CAST);
 342                         struct symbol *sym;
 343
 344                         token = typename(next, &sym);
 345                         cast->cast_type = sym->ctype.base_type;
 346                         token = expect(token, ')', "at end of cast operator");
 347                         *tree = cast;
 348                         if (match_op(token, '{'))
 349                                 return initializer(&cast->cast_expression, token);
 350                         token = cast_expression(token, &cast->cast_expression);
 351                         return token;
 352                 }
 353         }
 354         return unary_expression(token, tree);
 355 }
 356
 357 /* Generic left-to-right binop parsing */
 358 static struct token *lr_binop_expression(struct token *token, struct expression **tree,
 359         enum expression_type type, struct token *(*inner)(struct token *, struct expression **), ...)
 360 {
 361         struct expression *left = NULL;
 362         struct token * next = inner(token, &left);
 363
 364         if (left) {
 365                 while (token_type(next) == TOKEN_SPECIAL) {
 366                         struct expression *top, *right = NULL;
 367                         int op = next->special;
 368                         va_list args;
 369
 370                         va_start(args, inner);
 371                         for (;;) {
 372                                 int nextop = va_arg(args, int);
 373                                 if (!nextop)
 374                                         goto out;
 375                                 if (op == nextop)
 376                                         break;
 377                         }
 378                         va_end(args);
 379                         top = alloc_expression(next->pos, type);
 380                         next = inner(next->next, &right);
 381                         if (!right) {
 382                                 warn(next->pos, "No right hand side of '%s'-expression", show_special(op));
 383                                 break;
 384                         }
 385                         top->op = op;
 386                         top->left = left;
 387                         top->right = right;
 388                         left = top;
 389                 }
 390         }
 391 out:
 392         *tree = left;
 393         return next;
 394 }
 395
 396 static struct token *multiplicative_expression(struct token *token, struct expression **tree)
 397 {
 398         return lr_binop_expression(token, tree, EXPR_BINOP, cast_expression, '*', '/', '%', 0);
 399 }
 400
 401 static struct token *additive_expression(struct token *token, struct expression **tree)
 402 {
 403         return lr_binop_expression(token, tree, EXPR_BINOP, multiplicative_expression, '+', '-', 0);
 404 }
 405
 406 static struct token *shift_expression(struct token *token, struct expression **tree)
 407 {
 408         return lr_binop_expression(token, tree, EXPR_BINOP, additive_expression, SPECIAL_LEFTSHIFT, SPECIAL_RIGHTSHIFT, 0);
 409 }
 410
 411 static struct token *relational_expression(struct token *token, struct expression **tree)
 412 {
 413         return lr_binop_expression(token, tree, EXPR_COMPARE, shift_expression, '<', '>', SPECIAL_LTE, SPECIAL_GTE, 0);
 414 }
 415
 416 static struct token *equality_expression(struct token *token, struct expression **tree)
 417 {
 418         return lr_binop_expression(token, tree, EXPR_COMPARE, relational_expression, SPECIAL_EQUAL, SPECIAL_NOTEQUAL, 0);
 419 }
 420
 421 static struct token *bitwise_and_expression(struct token *token, struct expression **tree)
 422 {
 423         return lr_binop_expression(token, tree, EXPR_BINOP, equality_expression, '&', 0);
 424 }
 425
 426 static struct token *bitwise_xor_expression(struct token *token, struct expression **tree)
 427 {
 428         return lr_binop_expression(token, tree, EXPR_BINOP, bitwise_and_expression, '^', 0);
 429 }
 430
 431 static struct token *bitwise_or_expression(struct token *token, struct expression **tree)
 432 {
 433         return lr_binop_expression(token, tree, EXPR_BINOP, bitwise_xor_expression, '|', 0);
 434 }
 435
 436 static struct token *logical_and_expression(struct token *token, struct expression **tree)
 437 {
 438         return lr_binop_expression(token, tree, EXPR_LOGICAL, bitwise_or_expression, SPECIAL_LOGICAL_AND, 0);
 439 }
 440
 441 static struct token *logical_or_expression(struct token *token, struct expression **tree)
 442 {
 443         return lr_binop_expression(token, tree, EXPR_LOGICAL, logical_and_expression, SPECIAL_LOGICAL_OR, 0);
 444 }
 445
 446 struct token *conditional_expression(struct token *token, struct expression **tree)
 447 {
 448         token = logical_or_expression(token, tree);
 449         if (match_op(token, '?')) {
 450                 struct expression *expr = alloc_expression(token->pos, EXPR_CONDITIONAL);
 451                 expr->op = token->special;
 452                 expr->left = *tree;
 453                 *tree = expr;
 454                 token = parse_expression(token->next, &expr->cond_true);
 455                 token = expect(token, ':', "in conditional expression");
 456                 token = conditional_expression(token, &expr->cond_false);
 457         }
 458         return token;
 459 }
 460
 461 struct token *assignment_expression(struct token *token, struct expression **tree)
 462 {
 463         token = conditional_expression(token, tree);
 464         if (token_type(token) == TOKEN_SPECIAL) {
 465                 static const int assignments[] = {
 466                         '=',
 467                         SPECIAL_ADD_ASSIGN, SPECIAL_SUB_ASSIGN,
 468                         SPECIAL_MUL_ASSIGN, SPECIAL_DIV_ASSIGN,
 469                         SPECIAL_MOD_ASSIGN, SPECIAL_SHL_ASSIGN,
 470                         SPECIAL_SHR_ASSIGN, SPECIAL_AND_ASSIGN,
 471                         SPECIAL_OR_ASSIGN,  SPECIAL_XOR_ASSIGN };
 472                 int i, op = token->special;
 473                 for (i = 0; i < sizeof(assignments)/sizeof(int); i++)
 474                         if (assignments[i] == op) {
 475                                 struct expression * expr = alloc_expression(token->pos, EXPR_ASSIGNMENT);
 476                                 expr->left = *tree;
 477                                 expr->op = op;
 478                                 *tree = expr;
 479                                 return assignment_expression(token->next, &expr->right);
 480                         }
 481         }
 482         return token;
 483 }
 484
 485 static struct token *comma_expression(struct token *token, struct expression **tree)
 486 {
 487         return lr_binop_expression(token, tree, EXPR_COMMA, assignment_expression, ',', 0);
 488 }
 489
 490 struct token *parse_expression(struct token *token, struct expression **tree)
 491 {
 492         return comma_expression(token,tree);
 493 }
 494
 495