4 * Copyright (C) 2003 Transmeta Corp.
5 * 2003-2004 Linus Torvalds
7 * Licensed under the Open Software License version 1.1
9 * This is the expression parsing part of parsing C.
27 #include "expression.h"
30 static int match_oplist(int op
, ...)
37 nextop
= va_arg(args
, int);
38 } while (nextop
!= 0 && nextop
!= op
);
44 static struct token
*comma_expression(struct token
*, struct expression
**);
46 struct token
*parens_expression(struct token
*token
, struct expression
**expr
, const char *where
)
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
);
55 token
= compound_statement(token
->next
, stmt
);
57 token
= expect(token
, '}', "at end of statement expression");
59 token
= parse_expression(token
, expr
);
60 return expect(token
, ')', where
);
64 * Handle __func__, __FUNCTION__ and __PRETTY_FUNCTION__ token
67 static int convert_one_fn_token(struct token
*token
)
69 struct symbol
*sym
= current_fn
;
72 struct ident
*ident
= sym
->ident
;
75 struct string
*string
;
77 string
= __alloc_string(len
+1);
78 memcpy(string
->data
, ident
->name
, len
);
79 string
->data
[len
] = 0;
80 string
->length
= len
+1;
81 token_type(token
) = TOKEN_STRING
;
82 token
->string
= string
;
89 static int convert_function(struct token
*next
)
93 struct token
*token
= next
;
95 switch (token_type(token
)) {
99 if (token
->ident
== &__func___ident
||
100 token
->ident
== &__FUNCTION___ident
||
101 token
->ident
== &__PRETTY_FUNCTION___ident
) {
102 if (!convert_one_fn_token(token
))
116 static struct token
*parse_type(struct token
*token
, struct expression
**tree
)
119 *tree
= alloc_expression(token
->pos
, EXPR_TYPE
);
120 (*tree
)->flags
= Int_const_expr
; /* sic */
121 token
= typename(token
, &sym
, NULL
);
123 sparse_error(token
->pos
,
124 "type expression should not include identifier "
125 "\"%s\"", sym
->ident
->name
);
126 (*tree
)->symbol
= sym
;
130 static struct token
*builtin_types_compatible_p_expr(struct token
*token
,
131 struct expression
**tree
)
133 struct expression
*expr
= alloc_expression(
134 token
->pos
, EXPR_COMPARE
);
135 expr
->flags
= Int_const_expr
;
136 expr
->op
= SPECIAL_EQUAL
;
138 if (!match_op(token
, '('))
139 return expect(token
, '(',
140 "after __builtin_types_compatible_p");
142 token
= parse_type(token
, &expr
->left
);
143 if (!match_op(token
, ','))
144 return expect(token
, ',',
145 "in __builtin_types_compatible_p");
147 token
= parse_type(token
, &expr
->right
);
148 if (!match_op(token
, ')'))
149 return expect(token
, ')',
150 "at end of __builtin_types_compatible_p");
157 static struct token
*builtin_offsetof_expr(struct token
*token
,
158 struct expression
**tree
)
160 struct expression
*expr
= NULL
;
161 struct expression
**p
= &expr
;
166 if (!match_op(token
, '('))
167 return expect(token
, '(', "after __builtin_offset");
170 token
= typename(token
, &sym
, NULL
);
172 sparse_error(token
->pos
,
173 "type expression should not include identifier "
174 "\"%s\"", sym
->ident
->name
);
176 if (!match_op(token
, ','))
177 return expect(token
, ',', "in __builtin_offset");
180 struct expression
*e
;
186 return expect(token
, ')', "at end of __builtin_offset");
187 case SPECIAL_DEREFERENCE
:
188 e
= alloc_expression(token
->pos
, EXPR_OFFSETOF
);
189 e
->flags
= Int_const_expr
;
196 e
= alloc_expression(token
->pos
, EXPR_OFFSETOF
);
197 e
->flags
= Int_const_expr
;
199 if (token_type(token
) != TOKEN_IDENT
) {
200 sparse_error(token
->pos
, "Expected member name");
203 e
->ident
= token
->ident
;
208 e
= alloc_expression(token
->pos
, EXPR_OFFSETOF
);
209 e
->flags
= Int_const_expr
;
211 token
= parse_expression(token
, &e
->index
);
212 token
= expect(token
, ']',
213 "at end of array dereference");
219 op
= token_type(token
) == TOKEN_SPECIAL
? token
->special
: 0;
223 static struct token
*string_expression(struct token
*token
, struct expression
*expr
)
225 struct string
*string
= token
->string
;
226 struct token
*next
= token
->next
;
227 int stringtype
= token_type(token
);
229 convert_function(token
);
231 if (token_type(next
) == stringtype
) {
232 int totlen
= string
->length
-1;
236 totlen
+= next
->string
->length
-1;
238 } while (token_type(next
) == stringtype
);
240 if (totlen
> MAX_STRING
) {
241 warning(token
->pos
, "trying to concatenate %d-character string (%d bytes max)", totlen
, MAX_STRING
);
245 string
= __alloc_string(totlen
+1);
246 string
->length
= totlen
+1;
250 struct string
*s
= next
->string
;
251 int len
= s
->length
-1;
258 memcpy(data
, s
->data
, len
);
260 } while (token_type(next
) == stringtype
);
263 expr
->string
= string
;
268 #define ULLONG_MAX (~0ULL)
271 static void get_number_value(struct expression
*expr
, struct token
*token
)
273 const char *str
= token
->number
;
274 unsigned long long value
;
276 int size
= 0, want_unsigned
= 0;
277 int overflow
= 0, do_warn
= 0;
278 int try_unsigned
= 1;
282 value
= strtoull(str
, &end
, 0);
285 if (value
== ULLONG_MAX
&& errno
== ERANGE
)
291 } else if (c
== 'u' || c
== 'U') {
295 } else if (c
== 'l' || c
== 'L') {
308 /* OK, it's a valid integer */
309 /* decimals can be unsigned only if directly specified as such */
310 if (str
[0] != '0' && !want_unsigned
)
313 bits
= bits_in_int
- 1;
314 if (!(value
& (~1ULL << bits
))) {
315 if (!(value
& (1ULL << bits
))) {
317 } else if (try_unsigned
) {
326 bits
= bits_in_long
- 1;
327 if (!(value
& (~1ULL << bits
))) {
328 if (!(value
& (1ULL << bits
))) {
330 } else if (try_unsigned
) {
339 bits
= bits_in_longlong
- 1;
340 if (value
& (~1ULL << bits
))
342 if (!(value
& (1ULL << bits
)))
345 warning(expr
->pos
, "decimal constant %s is too big for long long",
350 warning(expr
->pos
, "constant %s is so big it is%s%s%s",
352 want_unsigned
? " unsigned":"",
353 size
> 0 ? " long":"",
354 size
> 1 ? " long":"");
357 "decimal constant %s is between LONG_MAX and ULONG_MAX."
358 " For C99 that means long long, C90 compilers are very "
359 "likely to produce unsigned long (and a warning) here",
361 expr
->type
= EXPR_VALUE
;
362 expr
->flags
= Int_const_expr
;
363 expr
->ctype
= ctype_integer(size
, want_unsigned
);
367 error_die(expr
->pos
, "constant %s is too big even for unsigned long long",
371 expr
->fvalue
= string_to_ld(str
, &end
);
378 if (*end
== 'f' || *end
== 'F')
379 expr
->ctype
= &float_ctype
;
380 else if (*end
== 'l' || *end
== 'L')
381 expr
->ctype
= &ldouble_ctype
;
383 expr
->ctype
= &double_ctype
;
387 expr
->flags
= Float_literal
;
388 expr
->type
= EXPR_FVALUE
;
392 error_die(expr
->pos
, "constant %s is not a valid number", show_token(token
));
395 struct token
*primary_expression(struct token
*token
, struct expression
**tree
)
397 struct expression
*expr
= NULL
;
399 switch (token_type(token
)) {
401 case TOKEN_WIDE_CHAR
:
402 expr
= alloc_expression(token
->pos
, EXPR_VALUE
);
403 expr
->flags
= Int_const_expr
;
404 expr
->ctype
= token_type(token
) == TOKEN_CHAR
? &int_ctype
: &long_ctype
;
405 expr
->value
= (unsigned char) token
->character
;
410 expr
= alloc_expression(token
->pos
, EXPR_VALUE
);
411 get_number_value(expr
, token
); /* will see if it's an integer */
415 case TOKEN_ZERO_IDENT
: {
416 expr
= alloc_expression(token
->pos
, EXPR_SYMBOL
);
417 expr
->flags
= Int_const_expr
;
418 expr
->ctype
= &int_ctype
;
419 expr
->symbol
= &zero_int
;
420 expr
->symbol_name
= token
->ident
;
426 struct symbol
*sym
= lookup_symbol(token
->ident
, NS_SYMBOL
| NS_TYPEDEF
);
427 struct token
*next
= token
->next
;
430 if (convert_function(token
))
432 if (token
->ident
== &__builtin_types_compatible_p_ident
) {
433 token
= builtin_types_compatible_p_expr(token
, &expr
);
436 if (token
->ident
== &__builtin_offsetof_ident
) {
437 token
= builtin_offsetof_expr(token
, &expr
);
440 } else if (sym
->enum_member
) {
441 expr
= alloc_expression(token
->pos
, EXPR_VALUE
);
442 *expr
= *sym
->initializer
;
443 /* we want the right position reported, thus the copy */
444 expr
->pos
= token
->pos
;
445 expr
->flags
= Int_const_expr
;
450 expr
= alloc_expression(token
->pos
, EXPR_SYMBOL
);
453 * We support types as real first-class citizens, with type
456 * if (typeof(a) == int) ..
458 if (sym
&& sym
->namespace == NS_TYPEDEF
) {
459 sparse_error(token
->pos
, "typename in expression");
462 expr
->symbol_name
= token
->ident
;
469 case TOKEN_WIDE_STRING
: {
471 expr
= alloc_expression(token
->pos
, EXPR_STRING
);
472 expr
->wide
= token_type(token
) == TOKEN_WIDE_STRING
;
473 token
= string_expression(token
, expr
);
478 if (token
->special
== '(') {
479 expr
= alloc_expression(token
->pos
, EXPR_PREOP
);
481 token
= parens_expression(token
, &expr
->unop
, "in expression");
483 expr
->flags
= expr
->unop
->flags
;
486 if (token
->special
== '[' && lookup_type(token
->next
)) {
487 expr
= alloc_expression(token
->pos
, EXPR_TYPE
);
488 expr
->flags
= Int_const_expr
; /* sic */
489 token
= typename(token
->next
, &expr
->symbol
, NULL
);
490 token
= expect(token
, ']', "in type expression");
501 static struct token
*expression_list(struct token
*token
, struct expression_list
**list
)
503 while (!match_op(token
, ')')) {
504 struct expression
*expr
= NULL
;
505 token
= assignment_expression(token
, &expr
);
508 add_expression(list
, expr
);
509 if (!match_op(token
, ','))
517 * extend to deal with the ambiguous C grammar for parsing
518 * a cast expressions followed by an initializer.
520 static struct token
*postfix_expression(struct token
*token
, struct expression
**tree
, struct expression
*cast_init_expr
)
522 struct expression
*expr
= cast_init_expr
;
525 token
= primary_expression(token
, &expr
);
527 while (expr
&& token_type(token
) == TOKEN_SPECIAL
) {
528 switch (token
->special
) {
529 case '[': { /* Array dereference */
530 struct expression
*deref
= alloc_expression(token
->pos
, EXPR_PREOP
);
531 struct expression
*add
= alloc_expression(token
->pos
, EXPR_BINOP
);
538 token
= parse_expression(token
->next
, &add
->right
);
539 token
= expect(token
, ']', "at end of array dereference");
543 case SPECIAL_INCREMENT
: /* Post-increment */
544 case SPECIAL_DECREMENT
: { /* Post-decrement */
545 struct expression
*post
= alloc_expression(token
->pos
, EXPR_POSTOP
);
546 post
->op
= token
->special
;
552 case SPECIAL_DEREFERENCE
: { /* Structure pointer member dereference */
553 /* "x->y" is just shorthand for "(*x).y" */
554 struct expression
*inner
= alloc_expression(token
->pos
, EXPR_PREOP
);
560 case '.': { /* Structure member dereference */
561 struct expression
*deref
= alloc_expression(token
->pos
, EXPR_DEREF
);
565 if (token_type(token
) != TOKEN_IDENT
) {
566 sparse_error(token
->pos
, "Expected member name");
569 deref
->member
= token
->ident
;
575 case '(': { /* Function call */
576 struct expression
*call
= alloc_expression(token
->pos
, EXPR_CALL
);
579 token
= expression_list(token
->next
, &call
->args
);
580 token
= expect(token
, ')', "in function call");
594 static struct token
*cast_expression(struct token
*token
, struct expression
**tree
);
595 static struct token
*unary_expression(struct token
*token
, struct expression
**tree
);
597 static struct token
*type_info_expression(struct token
*token
,
598 struct expression
**tree
, int type
)
600 struct expression
*expr
= alloc_expression(token
->pos
, type
);
604 expr
->flags
= Int_const_expr
; /* XXX: VLA support will need that changed */
606 if (!match_op(token
, '(') || !lookup_type(token
->next
))
607 return unary_expression(token
, &expr
->cast_expression
);
609 token
= typename(token
->next
, &expr
->cast_type
, NULL
);
611 if (!match_op(token
, ')')) {
612 static const char * error
[] = {
613 [EXPR_SIZEOF
] = "at end of sizeof",
614 [EXPR_ALIGNOF
] = "at end of __alignof__",
615 [EXPR_PTRSIZEOF
] = "at end of __sizeof_ptr__"
617 return expect(token
, ')', error
[type
]);
622 * C99 ambiguity: the typename might have been the beginning
623 * of a typed initializer expression..
625 if (match_op(token
, '{')) {
626 struct expression
*cast
= alloc_expression(p
->pos
, EXPR_CAST
);
627 cast
->cast_type
= expr
->cast_type
;
628 expr
->cast_type
= NULL
;
629 expr
->cast_expression
= cast
;
630 token
= initializer(&cast
->cast_expression
, token
);
631 token
= postfix_expression(token
, &expr
->cast_expression
, cast
);
636 static struct token
*unary_expression(struct token
*token
, struct expression
**tree
)
638 if (token_type(token
) == TOKEN_IDENT
) {
639 struct ident
*ident
= token
->ident
;
640 if (ident
->reserved
) {
641 static const struct {
644 } type_information
[] = {
645 { &sizeof_ident
, EXPR_SIZEOF
},
646 { &__alignof___ident
, EXPR_ALIGNOF
},
647 { &__alignof_ident
, EXPR_ALIGNOF
},
648 { &__sizeof_ptr___ident
, EXPR_PTRSIZEOF
},
651 for (i
= 0; i
< 3; i
++) {
652 if (ident
== type_information
[i
].id
)
653 return type_info_expression(token
, tree
, type_information
[i
].type
);
658 if (token_type(token
) == TOKEN_SPECIAL
) {
659 if (match_oplist(token
->special
,
660 SPECIAL_INCREMENT
, SPECIAL_DECREMENT
,
662 struct expression
*unop
;
663 struct expression
*unary
;
666 next
= cast_expression(token
->next
, &unop
);
668 sparse_error(token
->pos
, "Syntax error in unary expression");
672 unary
= alloc_expression(token
->pos
, EXPR_PREOP
);
673 unary
->op
= token
->special
;
678 /* possibly constant ones */
679 if (match_oplist(token
->special
, '+', '-', '~', '!', 0)) {
680 struct expression
*unop
;
681 struct expression
*unary
;
684 next
= cast_expression(token
->next
, &unop
);
686 sparse_error(token
->pos
, "Syntax error in unary expression");
690 unary
= alloc_expression(token
->pos
, EXPR_PREOP
);
691 unary
->op
= token
->special
;
693 unary
->flags
= unop
->flags
& Int_const_expr
;
697 /* Gcc extension: &&label gives the address of a label */
698 if (match_op(token
, SPECIAL_LOGICAL_AND
) &&
699 token_type(token
->next
) == TOKEN_IDENT
) {
700 struct expression
*label
= alloc_expression(token
->pos
, EXPR_LABEL
);
701 struct symbol
*sym
= label_symbol(token
->next
);
702 if (!(sym
->ctype
.modifiers
& MOD_ADDRESSABLE
)) {
703 sym
->ctype
.modifiers
|= MOD_ADDRESSABLE
;
704 add_symbol(&function_computed_target_list
, sym
);
706 label
->label_symbol
= sym
;
708 return token
->next
->next
;
713 return postfix_expression(token
, tree
, NULL
);
717 * Ambiguity: a '(' can be either a cast-expression or
718 * a primary-expression depending on whether it is followed
721 * additional ambiguity: a "cast expression" followed by
722 * an initializer is really a postfix-expression.
724 static struct token
*cast_expression(struct token
*token
, struct expression
**tree
)
726 if (match_op(token
, '(')) {
727 struct token
*next
= token
->next
;
728 if (lookup_type(next
)) {
729 struct expression
*cast
= alloc_expression(next
->pos
, EXPR_CAST
);
730 struct expression
*v
;
734 token
= typename(next
, &sym
, &is_force
);
735 cast
->cast_type
= sym
;
736 token
= expect(token
, ')', "at end of cast operator");
737 if (match_op(token
, '{')) {
740 "[force] in compound literal");
741 token
= initializer(&cast
->cast_expression
, token
);
742 return postfix_expression(token
, tree
, cast
);
746 cast
->type
= EXPR_FORCE_CAST
;
747 token
= cast_expression(token
, &v
);
750 cast
->cast_expression
= v
;
751 if (v
->flags
& Int_const_expr
)
752 cast
->flags
= Int_const_expr
;
753 else if (v
->flags
& Float_literal
) /* and _not_ int */
754 cast
->flags
= Int_const_expr
| Float_literal
;
758 return unary_expression(token
, tree
);
762 * Generic left-to-right binop parsing
764 * This _really_ needs to be inlined, because that makes the inner
765 * function call statically deterministic rather than a totally
766 * unpredictable indirect call. But gcc-3 is so "clever" that it
767 * doesn't do so by default even when you tell it to inline it.
769 * Making it a macro avoids the inlining problem, and also means
770 * that we can pass in the op-comparison as an expression rather
771 * than create a data structure for it.
774 #define LR_BINOP_EXPRESSION(__token, tree, type, inner, compare) \
775 struct expression *left = NULL; \
776 struct token * next = inner(__token, &left); \
779 while (token_type(next) == TOKEN_SPECIAL) { \
780 struct expression *top, *right = NULL; \
781 int op = next->special; \
785 top = alloc_expression(next->pos, type); \
786 next = inner(next->next, &right); \
788 sparse_error(next->pos, "No right hand side of '%s'-expression", show_special(op)); \
791 top->flags = left->flags & right->flags \
795 top->right = right; \
803 static struct token *multiplicative_expression(struct token *token, struct expression **tree)
806 token
, tree
, EXPR_BINOP
, cast_expression
,
807 (op
== '*') || (op
== '/') || (op
== '%')
811 static struct token
*additive_expression(struct token
*token
, struct expression
**tree
)
814 token
, tree
, EXPR_BINOP
, multiplicative_expression
,
815 (op
== '+') || (op
== '-')
819 static struct token
*shift_expression(struct token
*token
, struct expression
**tree
)
822 token
, tree
, EXPR_BINOP
, additive_expression
,
823 (op
== SPECIAL_LEFTSHIFT
) || (op
== SPECIAL_RIGHTSHIFT
)
827 static struct token
*relational_expression(struct token
*token
, struct expression
**tree
)
830 token
, tree
, EXPR_COMPARE
, shift_expression
,
831 (op
== '<') || (op
== '>') ||
832 (op
== SPECIAL_LTE
) || (op
== SPECIAL_GTE
)
836 static struct token
*equality_expression(struct token
*token
, struct expression
**tree
)
839 token
, tree
, EXPR_COMPARE
, relational_expression
,
840 (op
== SPECIAL_EQUAL
) || (op
== SPECIAL_NOTEQUAL
)
844 static struct token
*bitwise_and_expression(struct token
*token
, struct expression
**tree
)
847 token
, tree
, EXPR_BINOP
, equality_expression
,
852 static struct token
*bitwise_xor_expression(struct token
*token
, struct expression
**tree
)
855 token
, tree
, EXPR_BINOP
, bitwise_and_expression
,
860 static struct token
*bitwise_or_expression(struct token
*token
, struct expression
**tree
)
863 token
, tree
, EXPR_BINOP
, bitwise_xor_expression
,
868 static struct token
*logical_and_expression(struct token
*token
, struct expression
**tree
)
871 token
, tree
, EXPR_LOGICAL
, bitwise_or_expression
,
872 (op
== SPECIAL_LOGICAL_AND
)
876 static struct token
*logical_or_expression(struct token
*token
, struct expression
**tree
)
879 token
, tree
, EXPR_LOGICAL
, logical_and_expression
,
880 (op
== SPECIAL_LOGICAL_OR
)
884 struct token
*conditional_expression(struct token
*token
, struct expression
**tree
)
886 token
= logical_or_expression(token
, tree
);
887 if (*tree
&& match_op(token
, '?')) {
888 struct expression
*expr
= alloc_expression(token
->pos
, EXPR_CONDITIONAL
);
889 expr
->op
= token
->special
;
892 token
= parse_expression(token
->next
, &expr
->cond_true
);
893 token
= expect(token
, ':', "in conditional expression");
894 token
= conditional_expression(token
, &expr
->cond_false
);
895 if (expr
->left
&& expr
->cond_false
) {
896 int is_const
= expr
->left
->flags
&
897 expr
->cond_false
->flags
&
900 is_const
&= expr
->cond_true
->flags
;
901 expr
->flags
= is_const
;
907 struct token
*assignment_expression(struct token
*token
, struct expression
**tree
)
909 token
= conditional_expression(token
, tree
);
910 if (*tree
&& token_type(token
) == TOKEN_SPECIAL
) {
911 static const int assignments
[] = {
913 SPECIAL_ADD_ASSIGN
, SPECIAL_SUB_ASSIGN
,
914 SPECIAL_MUL_ASSIGN
, SPECIAL_DIV_ASSIGN
,
915 SPECIAL_MOD_ASSIGN
, SPECIAL_SHL_ASSIGN
,
916 SPECIAL_SHR_ASSIGN
, SPECIAL_AND_ASSIGN
,
917 SPECIAL_OR_ASSIGN
, SPECIAL_XOR_ASSIGN
};
918 int i
, op
= token
->special
;
919 for (i
= 0; i
< ARRAY_SIZE(assignments
); i
++)
920 if (assignments
[i
] == op
) {
921 struct expression
* expr
= alloc_expression(token
->pos
, EXPR_ASSIGNMENT
);
925 return assignment_expression(token
->next
, &expr
->right
);
931 static struct token
*comma_expression(struct token
*token
, struct expression
**tree
)
934 token
, tree
, EXPR_COMMA
, assignment_expression
,
939 struct token
*parse_expression(struct token
*token
, struct expression
**tree
)
941 return comma_expression(token
,tree
);