1 /* YACC parser for C expressions, for GDB.
2 Copyright (C) 1986, 1989, 1990, 1991, 1993, 1994
3 Free Software Foundation, Inc.
5 This file is part of GDB.
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 2 of the License, or
10 (at your option) any later version.
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with this program; if not, write to the Free Software
19 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
21 /* Parse a C expression from text in a string,
22 and return the result as a struct expression pointer.
23 That structure contains arithmetic operations in reverse polish,
24 with constants represented by operations that are followed by special data.
25 See expression.h for the details of the format.
26 What is important here is that it can be built up sequentially
27 during the process of parsing; the lower levels of the tree always
28 come first in the result.
30 Note that malloc's and realloc's in this file are transformed to
31 xmalloc and xrealloc respectively by the same sed command in the
32 makefile that remaps any other malloc/realloc inserted by the parser
33 generator. Doing this with #defines and trying to control the interaction
34 with include files (<malloc.h> and <stdlib.h> for example) just became
35 too messy, particularly when such includes can be inserted at random
36 times by the parser generator. */
42 #include "expression.h"
44 #include "parser-defs.h"
47 #include "bfd.h" /* Required by objfiles.h. */
48 #include "symfile.h" /* Required by objfiles.h. */
49 #include "objfiles.h" /* For have_full_symbols and have_partial_symbols */
51 /* Remap normal yacc parser interface names (yyparse, yylex, yyerror, etc),
52 as well as gratuitiously global symbol names, so we can have multiple
53 yacc generated parsers in gdb. Note that these are only the variables
54 produced by yacc. If other parser generators (bison, byacc, etc) produce
55 additional global names that conflict at link time, then those parser
56 generators need to be fixed instead of adding those names to this list. */
58 #define yymaxdepth c_maxdepth
59 #define yyparse c_parse
61 #define yyerror c_error
64 #define yydebug c_debug
73 #define yyerrflag c_errflag
74 #define yynerrs c_nerrs
79 #define yystate c_state
85 #define yyreds c_reds /* With YYDEBUG defined */
86 #define yytoks c_toks /* With YYDEBUG defined */
89 #define yydefred c_yydefred
90 #define yydgoto c_yydgoto
91 #define yysindex c_yysindex
92 #define yyrindex c_yyrindex
93 #define yygindex c_yygindex
94 #define yytable c_yytable
95 #define yycheck c_yycheck
98 #define YYDEBUG 0 /* Default to no yydebug support */
102 yyparse PARAMS ((void));
105 yylex PARAMS ((void));
108 yyerror PARAMS ((char *));
112 /* Although the yacc "value" of an expression is not used,
113 since the result is stored in the structure being created,
114 other node types do have values. */
128 struct symtoken ssym;
131 enum exp_opcode opcode;
132 struct internalvar *ivar;
139 /* YYSTYPE gets defined by %union */
141 parse_number PARAMS ((char *, int, int, YYSTYPE *));
144 %type <voidval> exp exp1 type_exp start variable qualified_name lcurly
146 %type <tval> type typebase
147 %type <tvec> nonempty_typelist
148 /* %type <bval> block */
150 /* Fancy type parsing. */
151 %type <voidval> func_mod direct_abs_decl abs_decl
153 %type <lval> array_mod
155 %token <typed_val> INT
158 /* Both NAME and TYPENAME tokens represent symbols in the input,
159 and both convey their data as strings.
160 But a TYPENAME is a string that happens to be defined as a typedef
161 or builtin type name (such as int or char)
162 and a NAME is any other symbol.
163 Contexts where this distinction is not important can use the
164 nonterminal "name", which matches either NAME or TYPENAME. */
167 %token <ssym> NAME /* BLOCKNAME defined below to give it higher precedence. */
168 %token <tsym> TYPENAME
170 %type <ssym> name_not_typename
171 %type <tsym> typename
173 /* A NAME_OR_INT is a symbol which is not known in the symbol table,
174 but which would parse as a valid number in the current input radix.
175 E.g. "c" when input_radix==16. Depending on the parse, it will be
176 turned into a name or into a number. */
178 %token <ssym> NAME_OR_INT
180 %token STRUCT CLASS UNION ENUM SIZEOF UNSIGNED COLONCOLON
184 /* Special type cases, put in to allow the parser to distinguish different
186 %token SIGNED_KEYWORD LONG SHORT INT_KEYWORD CONST_KEYWORD VOLATILE_KEYWORD
187 %token <lval> LAST REGNAME
189 %token <ivar> VARIABLE
191 %token <opcode> ASSIGN_MODIFY
198 %right '=' ASSIGN_MODIFY
206 %left '<' '>' LEQ GEQ
211 %right UNARY INCREMENT DECREMENT
212 %right ARROW '.' '[' '('
213 %token <ssym> BLOCKNAME
225 { write_exp_elt_opcode(OP_TYPE);
226 write_exp_elt_type($1);
227 write_exp_elt_opcode(OP_TYPE);}
230 /* Expressions, including the comma operator. */
233 { write_exp_elt_opcode (BINOP_COMMA); }
236 /* Expressions, not including the comma operator. */
237 exp : '*' exp %prec UNARY
238 { write_exp_elt_opcode (UNOP_IND); }
240 exp : '&' exp %prec UNARY
241 { write_exp_elt_opcode (UNOP_ADDR); }
243 exp : '-' exp %prec UNARY
244 { write_exp_elt_opcode (UNOP_NEG); }
247 exp : '!' exp %prec UNARY
248 { write_exp_elt_opcode (UNOP_LOGICAL_NOT); }
251 exp : '~' exp %prec UNARY
252 { write_exp_elt_opcode (UNOP_COMPLEMENT); }
255 exp : INCREMENT exp %prec UNARY
256 { write_exp_elt_opcode (UNOP_PREINCREMENT); }
259 exp : DECREMENT exp %prec UNARY
260 { write_exp_elt_opcode (UNOP_PREDECREMENT); }
263 exp : exp INCREMENT %prec UNARY
264 { write_exp_elt_opcode (UNOP_POSTINCREMENT); }
267 exp : exp DECREMENT %prec UNARY
268 { write_exp_elt_opcode (UNOP_POSTDECREMENT); }
271 exp : SIZEOF exp %prec UNARY
272 { write_exp_elt_opcode (UNOP_SIZEOF); }
276 { write_exp_elt_opcode (STRUCTOP_PTR);
277 write_exp_string ($3);
278 write_exp_elt_opcode (STRUCTOP_PTR); }
281 exp : exp ARROW qualified_name
282 { /* exp->type::name becomes exp->*(&type::name) */
283 /* Note: this doesn't work if name is a
284 static member! FIXME */
285 write_exp_elt_opcode (UNOP_ADDR);
286 write_exp_elt_opcode (STRUCTOP_MPTR); }
288 exp : exp ARROW '*' exp
289 { write_exp_elt_opcode (STRUCTOP_MPTR); }
293 { write_exp_elt_opcode (STRUCTOP_STRUCT);
294 write_exp_string ($3);
295 write_exp_elt_opcode (STRUCTOP_STRUCT); }
298 exp : exp '.' qualified_name
299 { /* exp.type::name becomes exp.*(&type::name) */
300 /* Note: this doesn't work if name is a
301 static member! FIXME */
302 write_exp_elt_opcode (UNOP_ADDR);
303 write_exp_elt_opcode (STRUCTOP_MEMBER); }
306 exp : exp '.' '*' exp
307 { write_exp_elt_opcode (STRUCTOP_MEMBER); }
310 exp : exp '[' exp1 ']'
311 { write_exp_elt_opcode (BINOP_SUBSCRIPT); }
315 /* This is to save the value of arglist_len
316 being accumulated by an outer function call. */
317 { start_arglist (); }
318 arglist ')' %prec ARROW
319 { write_exp_elt_opcode (OP_FUNCALL);
320 write_exp_elt_longcst ((LONGEST) end_arglist ());
321 write_exp_elt_opcode (OP_FUNCALL); }
325 { start_arglist (); }
335 arglist : arglist ',' exp %prec ABOVE_COMMA
340 { $$ = end_arglist () - 1; }
342 exp : lcurly arglist rcurly %prec ARROW
343 { write_exp_elt_opcode (OP_ARRAY);
344 write_exp_elt_longcst ((LONGEST) 0);
345 write_exp_elt_longcst ((LONGEST) $3);
346 write_exp_elt_opcode (OP_ARRAY); }
349 exp : lcurly type rcurly exp %prec UNARY
350 { write_exp_elt_opcode (UNOP_MEMVAL);
351 write_exp_elt_type ($2);
352 write_exp_elt_opcode (UNOP_MEMVAL); }
355 exp : '(' type ')' exp %prec UNARY
356 { write_exp_elt_opcode (UNOP_CAST);
357 write_exp_elt_type ($2);
358 write_exp_elt_opcode (UNOP_CAST); }
365 /* Binary operators in order of decreasing precedence. */
368 { write_exp_elt_opcode (BINOP_REPEAT); }
372 { write_exp_elt_opcode (BINOP_MUL); }
376 { write_exp_elt_opcode (BINOP_DIV); }
380 { write_exp_elt_opcode (BINOP_REM); }
384 { write_exp_elt_opcode (BINOP_ADD); }
388 { write_exp_elt_opcode (BINOP_SUB); }
392 { write_exp_elt_opcode (BINOP_LSH); }
396 { write_exp_elt_opcode (BINOP_RSH); }
400 { write_exp_elt_opcode (BINOP_EQUAL); }
403 exp : exp NOTEQUAL exp
404 { write_exp_elt_opcode (BINOP_NOTEQUAL); }
408 { write_exp_elt_opcode (BINOP_LEQ); }
412 { write_exp_elt_opcode (BINOP_GEQ); }
416 { write_exp_elt_opcode (BINOP_LESS); }
420 { write_exp_elt_opcode (BINOP_GTR); }
424 { write_exp_elt_opcode (BINOP_BITWISE_AND); }
428 { write_exp_elt_opcode (BINOP_BITWISE_XOR); }
432 { write_exp_elt_opcode (BINOP_BITWISE_IOR); }
436 { write_exp_elt_opcode (BINOP_LOGICAL_AND); }
440 { write_exp_elt_opcode (BINOP_LOGICAL_OR); }
443 exp : exp '?' exp ':' exp %prec '?'
444 { write_exp_elt_opcode (TERNOP_COND); }
448 { write_exp_elt_opcode (BINOP_ASSIGN); }
451 exp : exp ASSIGN_MODIFY exp
452 { write_exp_elt_opcode (BINOP_ASSIGN_MODIFY);
453 write_exp_elt_opcode ($2);
454 write_exp_elt_opcode (BINOP_ASSIGN_MODIFY); }
458 { write_exp_elt_opcode (OP_LONG);
459 write_exp_elt_type ($1.type);
460 write_exp_elt_longcst ((LONGEST)($1.val));
461 write_exp_elt_opcode (OP_LONG); }
466 parse_number ($1.stoken.ptr, $1.stoken.length, 0, &val);
467 write_exp_elt_opcode (OP_LONG);
468 write_exp_elt_type (val.typed_val.type);
469 write_exp_elt_longcst ((LONGEST)val.typed_val.val);
470 write_exp_elt_opcode (OP_LONG);
476 { write_exp_elt_opcode (OP_DOUBLE);
477 write_exp_elt_type (builtin_type_double);
478 write_exp_elt_dblcst ($1);
479 write_exp_elt_opcode (OP_DOUBLE); }
486 { write_exp_elt_opcode (OP_LAST);
487 write_exp_elt_longcst ((LONGEST) $1);
488 write_exp_elt_opcode (OP_LAST); }
492 { write_exp_elt_opcode (OP_REGISTER);
493 write_exp_elt_longcst ((LONGEST) $1);
494 write_exp_elt_opcode (OP_REGISTER); }
498 { write_exp_elt_opcode (OP_INTERNALVAR);
499 write_exp_elt_intern ($1);
500 write_exp_elt_opcode (OP_INTERNALVAR); }
503 exp : SIZEOF '(' type ')' %prec UNARY
504 { write_exp_elt_opcode (OP_LONG);
505 write_exp_elt_type (builtin_type_int);
506 write_exp_elt_longcst ((LONGEST) TYPE_LENGTH ($3));
507 write_exp_elt_opcode (OP_LONG); }
511 { /* C strings are converted into array constants with
512 an explicit null byte added at the end. Thus
513 the array upper bound is the string length.
514 There is no such thing in C as a completely empty
516 char *sp = $1.ptr; int count = $1.length;
519 write_exp_elt_opcode (OP_LONG);
520 write_exp_elt_type (builtin_type_char);
521 write_exp_elt_longcst ((LONGEST)(*sp++));
522 write_exp_elt_opcode (OP_LONG);
524 write_exp_elt_opcode (OP_LONG);
525 write_exp_elt_type (builtin_type_char);
526 write_exp_elt_longcst ((LONGEST)'\0');
527 write_exp_elt_opcode (OP_LONG);
528 write_exp_elt_opcode (OP_ARRAY);
529 write_exp_elt_longcst ((LONGEST) 0);
530 write_exp_elt_longcst ((LONGEST) ($1.length));
531 write_exp_elt_opcode (OP_ARRAY); }
536 { write_exp_elt_opcode (OP_THIS);
537 write_exp_elt_opcode (OP_THIS); }
545 $$ = SYMBOL_BLOCK_VALUE ($1.sym);
549 lookup_symtab (copy_name ($1.stoken));
551 $$ = BLOCKVECTOR_BLOCK
552 (BLOCKVECTOR (tem), STATIC_BLOCK);
554 error ("No file or function \"%s\".",
555 copy_name ($1.stoken));
560 block : block COLONCOLON name
562 = lookup_symbol (copy_name ($3), $1,
563 VAR_NAMESPACE, (int *) NULL,
564 (struct symtab **) NULL);
565 if (!tem || SYMBOL_CLASS (tem) != LOC_BLOCK)
566 error ("No function \"%s\" in specified context.",
568 $$ = SYMBOL_BLOCK_VALUE (tem); }
571 variable: block COLONCOLON name
572 { struct symbol *sym;
573 sym = lookup_symbol (copy_name ($3), $1,
574 VAR_NAMESPACE, (int *) NULL,
575 (struct symtab **) NULL);
577 error ("No symbol \"%s\" in specified context.",
580 write_exp_elt_opcode (OP_VAR_VALUE);
581 /* block_found is set by lookup_symbol. */
582 write_exp_elt_block (block_found);
583 write_exp_elt_sym (sym);
584 write_exp_elt_opcode (OP_VAR_VALUE); }
587 qualified_name: typebase COLONCOLON name
589 struct type *type = $1;
590 if (TYPE_CODE (type) != TYPE_CODE_STRUCT
591 && TYPE_CODE (type) != TYPE_CODE_UNION)
592 error ("`%s' is not defined as an aggregate type.",
595 write_exp_elt_opcode (OP_SCOPE);
596 write_exp_elt_type (type);
597 write_exp_string ($3);
598 write_exp_elt_opcode (OP_SCOPE);
600 | typebase COLONCOLON '~' name
602 struct type *type = $1;
603 struct stoken tmp_token;
604 if (TYPE_CODE (type) != TYPE_CODE_STRUCT
605 && TYPE_CODE (type) != TYPE_CODE_UNION)
606 error ("`%s' is not defined as an aggregate type.",
609 if (!STREQ (type_name_no_tag (type), $4.ptr))
610 error ("invalid destructor `%s::~%s'",
611 type_name_no_tag (type), $4.ptr);
613 tmp_token.ptr = (char*) alloca ($4.length + 2);
614 tmp_token.length = $4.length + 1;
615 tmp_token.ptr[0] = '~';
616 memcpy (tmp_token.ptr+1, $4.ptr, $4.length);
617 tmp_token.ptr[tmp_token.length] = 0;
618 write_exp_elt_opcode (OP_SCOPE);
619 write_exp_elt_type (type);
620 write_exp_string (tmp_token);
621 write_exp_elt_opcode (OP_SCOPE);
625 variable: qualified_name
628 char *name = copy_name ($2);
630 struct minimal_symbol *msymbol;
633 lookup_symbol (name, (const struct block *) NULL,
634 VAR_NAMESPACE, (int *) NULL,
635 (struct symtab **) NULL);
638 write_exp_elt_opcode (OP_VAR_VALUE);
639 write_exp_elt_block (NULL);
640 write_exp_elt_sym (sym);
641 write_exp_elt_opcode (OP_VAR_VALUE);
645 msymbol = lookup_minimal_symbol (name, NULL, NULL);
648 write_exp_msymbol (msymbol,
649 lookup_function_type (builtin_type_int),
653 if (!have_full_symbols () && !have_partial_symbols ())
654 error ("No symbol table is loaded. Use the \"file\" command.");
656 error ("No symbol \"%s\" in current context.", name);
660 variable: name_not_typename
661 { struct symbol *sym = $1.sym;
665 if (symbol_read_needs_frame (sym))
667 if (innermost_block == 0 ||
668 contained_in (block_found,
670 innermost_block = block_found;
673 write_exp_elt_opcode (OP_VAR_VALUE);
674 /* We want to use the selected frame, not
675 another more inner frame which happens to
676 be in the same block. */
677 write_exp_elt_block (NULL);
678 write_exp_elt_sym (sym);
679 write_exp_elt_opcode (OP_VAR_VALUE);
681 else if ($1.is_a_field_of_this)
683 /* C++: it hangs off of `this'. Must
684 not inadvertently convert from a method call
686 if (innermost_block == 0 ||
687 contained_in (block_found, innermost_block))
688 innermost_block = block_found;
689 write_exp_elt_opcode (OP_THIS);
690 write_exp_elt_opcode (OP_THIS);
691 write_exp_elt_opcode (STRUCTOP_PTR);
692 write_exp_string ($1.stoken);
693 write_exp_elt_opcode (STRUCTOP_PTR);
697 struct minimal_symbol *msymbol;
698 register char *arg = copy_name ($1.stoken);
701 lookup_minimal_symbol (arg, NULL, NULL);
704 write_exp_msymbol (msymbol,
705 lookup_function_type (builtin_type_int),
708 else if (!have_full_symbols () && !have_partial_symbols ())
709 error ("No symbol table is loaded. Use the \"file\" command.");
711 error ("No symbol \"%s\" in current context.",
712 copy_name ($1.stoken));
719 /* "const" and "volatile" are curently ignored. A type qualifier
720 before the type is currently handled in the typebase rule.
721 The reason for recognizing these here (shift/reduce conflicts)
722 might be obsolete now that some pointer to member rules have
724 | typebase CONST_KEYWORD
725 | typebase VOLATILE_KEYWORD
727 { $$ = follow_types ($1); }
728 | typebase CONST_KEYWORD abs_decl
729 { $$ = follow_types ($1); }
730 | typebase VOLATILE_KEYWORD abs_decl
731 { $$ = follow_types ($1); }
735 { push_type (tp_pointer); $$ = 0; }
737 { push_type (tp_pointer); $$ = $2; }
739 { push_type (tp_reference); $$ = 0; }
741 { push_type (tp_reference); $$ = $2; }
745 direct_abs_decl: '(' abs_decl ')'
747 | direct_abs_decl array_mod
750 push_type (tp_array);
755 push_type (tp_array);
759 | direct_abs_decl func_mod
760 { push_type (tp_function); }
762 { push_type (tp_function); }
773 | '(' nonempty_typelist ')'
774 { free ((PTR)$2); $$ = 0; }
777 /* We used to try to recognize more pointer to member types here, but
778 that didn't work (shift/reduce conflicts meant that these rules never
779 got executed). The problem is that
780 int (foo::bar::baz::bizzle)
781 is a function type but
782 int (foo::bar::baz::bizzle::*)
783 is a pointer to member type. Stroustrup loses again! */
786 | typebase COLONCOLON '*'
787 { $$ = lookup_member_type (builtin_type_int, $1); }
790 typebase /* Implements (approximately): (type-qualifier)* type-specifier */
794 { $$ = builtin_type_int; }
796 { $$ = builtin_type_long; }
798 { $$ = builtin_type_short; }
800 { $$ = builtin_type_long; }
801 | UNSIGNED LONG INT_KEYWORD
802 { $$ = builtin_type_unsigned_long; }
804 { $$ = builtin_type_long_long; }
805 | LONG LONG INT_KEYWORD
806 { $$ = builtin_type_long_long; }
808 { $$ = builtin_type_unsigned_long_long; }
809 | UNSIGNED LONG LONG INT_KEYWORD
810 { $$ = builtin_type_unsigned_long_long; }
812 { $$ = builtin_type_short; }
813 | UNSIGNED SHORT INT_KEYWORD
814 { $$ = builtin_type_unsigned_short; }
816 { $$ = lookup_struct (copy_name ($2),
817 expression_context_block); }
819 { $$ = lookup_struct (copy_name ($2),
820 expression_context_block); }
822 { $$ = lookup_union (copy_name ($2),
823 expression_context_block); }
825 { $$ = lookup_enum (copy_name ($2),
826 expression_context_block); }
828 { $$ = lookup_unsigned_typename (TYPE_NAME($2.type)); }
830 { $$ = builtin_type_unsigned_int; }
831 | SIGNED_KEYWORD typename
832 { $$ = lookup_signed_typename (TYPE_NAME($2.type)); }
834 { $$ = builtin_type_int; }
835 | TEMPLATE name '<' type '>'
836 { $$ = lookup_template_type(copy_name($2), $4,
837 expression_context_block);
839 /* "const" and "volatile" are curently ignored. A type qualifier
840 after the type is handled in the ptype rule. I think these could
842 | CONST_KEYWORD typebase { $$ = $2; }
843 | VOLATILE_KEYWORD typebase { $$ = $2; }
849 $$.stoken.ptr = "int";
850 $$.stoken.length = 3;
851 $$.type = builtin_type_int;
855 $$.stoken.ptr = "long";
856 $$.stoken.length = 4;
857 $$.type = builtin_type_long;
861 $$.stoken.ptr = "short";
862 $$.stoken.length = 5;
863 $$.type = builtin_type_short;
869 { $$ = (struct type **) malloc (sizeof (struct type *) * 2);
870 $<ivec>$[0] = 1; /* Number of types in vector */
873 | nonempty_typelist ',' type
874 { int len = sizeof (struct type *) * (++($<ivec>1[0]) + 1);
875 $$ = (struct type **) realloc ((char *) $1, len);
876 $$[$<ivec>$[0]] = $3;
880 name : NAME { $$ = $1.stoken; }
881 | BLOCKNAME { $$ = $1.stoken; }
882 | TYPENAME { $$ = $1.stoken; }
883 | NAME_OR_INT { $$ = $1.stoken; }
886 name_not_typename : NAME
888 /* These would be useful if name_not_typename was useful, but it is just
889 a fake for "variable", so these cause reduce/reduce conflicts because
890 the parser can't tell whether NAME_OR_INT is a name_not_typename (=variable,
891 =exp) or just an exp. If name_not_typename was ever used in an lvalue
892 context where only a name could occur, this might be useful.
899 /* Take care of parsing a number (anything that starts with a digit).
900 Set yylval and return the token type; update lexptr.
901 LEN is the number of characters in it. */
903 /*** Needs some error checking for the float case ***/
906 parse_number (p, len, parsed_float, putithere)
912 /* FIXME: Shouldn't these be unsigned? We don't deal with negative values
913 here, and we do kind of silly things like cast to unsigned. */
914 register LONGEST n = 0;
915 register LONGEST prevn = 0;
920 register int base = input_radix;
923 /* Number of "L" suffixes encountered. */
926 /* We have found a "L" or "U" suffix. */
927 int found_suffix = 0;
929 unsigned LONGEST high_bit;
930 struct type *signed_type;
931 struct type *unsigned_type;
935 /* It's a float since it contains a point or an exponent. */
936 putithere->dval = atof (p);
940 /* Handle base-switching prefixes 0x, 0t, 0d, 0 */
974 if (c >= 'A' && c <= 'Z')
976 if (c != 'l' && c != 'u')
978 if (c >= '0' && c <= '9')
986 if (base > 10 && c >= 'a' && c <= 'f')
990 n += i = c - 'a' + 10;
1003 return ERROR; /* Char not a digit */
1006 return ERROR; /* Invalid digit in this base */
1008 /* Portably test for overflow (only works for nonzero values, so make
1009 a second check for zero). FIXME: Can't we just make n and prevn
1010 unsigned and avoid this? */
1011 if (c != 'l' && c != 'u' && (prevn >= n) && n != 0)
1012 unsigned_p = 1; /* Try something unsigned */
1014 /* Portably test for unsigned overflow.
1015 FIXME: This check is wrong; for example it doesn't find overflow
1016 on 0x123456789 when LONGEST is 32 bits. */
1017 if (c != 'l' && c != 'u' && n != 0)
1019 if ((unsigned_p && (unsigned LONGEST) prevn >= (unsigned LONGEST) n))
1020 error ("Numeric constant too large.");
1025 /* An integer constant is an int, a long, or a long long. An L
1026 suffix forces it to be long; an LL suffix forces it to be long
1027 long. If not forced to a larger size, it gets the first type of
1028 the above that it fits in. To figure out whether it fits, we
1029 shift it right and see whether anything remains. Note that we
1030 can't shift sizeof (LONGEST) * HOST_CHAR_BIT bits or more in one
1031 operation, because many compilers will warn about such a shift
1032 (which always produces a zero result). Sometimes TARGET_INT_BIT
1033 or TARGET_LONG_BIT will be that big, sometimes not. To deal with
1034 the case where it is we just always shift the value more than
1035 once, with fewer bits each time. */
1037 un = (unsigned LONGEST)n >> 2;
1039 && (un >> (TARGET_INT_BIT - 2)) == 0)
1041 high_bit = ((unsigned LONGEST)1) << (TARGET_INT_BIT-1);
1043 /* A large decimal (not hex or octal) constant (between INT_MAX
1044 and UINT_MAX) is a long or unsigned long, according to ANSI,
1045 never an unsigned int, but this code treats it as unsigned
1046 int. This probably should be fixed. GCC gives a warning on
1049 unsigned_type = builtin_type_unsigned_int;
1050 signed_type = builtin_type_int;
1052 else if (long_p <= 1
1053 && (un >> (TARGET_LONG_BIT - 2)) == 0)
1055 high_bit = ((unsigned LONGEST)1) << (TARGET_LONG_BIT-1);
1056 unsigned_type = builtin_type_unsigned_long;
1057 signed_type = builtin_type_long;
1061 high_bit = (((unsigned LONGEST)1)
1062 << (TARGET_LONG_LONG_BIT - 32 - 1)
1066 /* A long long does not fit in a LONGEST. */
1068 (unsigned LONGEST)1 << (sizeof (LONGEST) * HOST_CHAR_BIT - 1);
1069 unsigned_type = builtin_type_unsigned_long_long;
1070 signed_type = builtin_type_long_long;
1073 putithere->typed_val.val = n;
1075 /* If the high bit of the worked out type is set then this number
1076 has to be unsigned. */
1078 if (unsigned_p || (n & high_bit))
1080 putithere->typed_val.type = unsigned_type;
1084 putithere->typed_val.type = signed_type;
1094 enum exp_opcode opcode;
1097 static const struct token tokentab3[] =
1099 {">>=", ASSIGN_MODIFY, BINOP_RSH},
1100 {"<<=", ASSIGN_MODIFY, BINOP_LSH}
1103 static const struct token tokentab2[] =
1105 {"+=", ASSIGN_MODIFY, BINOP_ADD},
1106 {"-=", ASSIGN_MODIFY, BINOP_SUB},
1107 {"*=", ASSIGN_MODIFY, BINOP_MUL},
1108 {"/=", ASSIGN_MODIFY, BINOP_DIV},
1109 {"%=", ASSIGN_MODIFY, BINOP_REM},
1110 {"|=", ASSIGN_MODIFY, BINOP_BITWISE_IOR},
1111 {"&=", ASSIGN_MODIFY, BINOP_BITWISE_AND},
1112 {"^=", ASSIGN_MODIFY, BINOP_BITWISE_XOR},
1113 {"++", INCREMENT, BINOP_END},
1114 {"--", DECREMENT, BINOP_END},
1115 {"->", ARROW, BINOP_END},
1116 {"&&", ANDAND, BINOP_END},
1117 {"||", OROR, BINOP_END},
1118 {"::", COLONCOLON, BINOP_END},
1119 {"<<", LSH, BINOP_END},
1120 {">>", RSH, BINOP_END},
1121 {"==", EQUAL, BINOP_END},
1122 {"!=", NOTEQUAL, BINOP_END},
1123 {"<=", LEQ, BINOP_END},
1124 {">=", GEQ, BINOP_END}
1127 /* Read one token, getting characters through lexptr. */
1138 static char *tempbuf;
1139 static int tempbufsize;
1144 /* See if it is a special token of length 3. */
1145 for (i = 0; i < sizeof tokentab3 / sizeof tokentab3[0]; i++)
1146 if (STREQN (tokstart, tokentab3[i].operator, 3))
1149 yylval.opcode = tokentab3[i].opcode;
1150 return tokentab3[i].token;
1153 /* See if it is a special token of length 2. */
1154 for (i = 0; i < sizeof tokentab2 / sizeof tokentab2[0]; i++)
1155 if (STREQN (tokstart, tokentab2[i].operator, 2))
1158 yylval.opcode = tokentab2[i].opcode;
1159 return tokentab2[i].token;
1162 switch (c = *tokstart)
1174 /* We either have a character constant ('0' or '\177' for example)
1175 or we have a quoted symbol reference ('foo(int,int)' in C++
1180 c = parse_escape (&lexptr);
1182 yylval.typed_val.val = c;
1183 yylval.typed_val.type = builtin_type_char;
1188 namelen = skip_quoted (tokstart) - tokstart;
1191 lexptr = tokstart + namelen;
1192 if (lexptr[-1] != '\'')
1193 error ("Unmatched single quote.");
1198 error ("Invalid character constant.");
1208 if (paren_depth == 0)
1215 if (comma_terminates && paren_depth == 0)
1221 /* Might be a floating point number. */
1222 if (lexptr[1] < '0' || lexptr[1] > '9')
1223 goto symbol; /* Nope, must be a symbol. */
1224 /* FALL THRU into number case. */
1237 /* It's a number. */
1238 int got_dot = 0, got_e = 0, toktype;
1239 register char *p = tokstart;
1240 int hex = input_radix > 10;
1242 if (c == '0' && (p[1] == 'x' || p[1] == 'X'))
1247 else if (c == '0' && (p[1]=='t' || p[1]=='T' || p[1]=='d' || p[1]=='D'))
1255 /* This test includes !hex because 'e' is a valid hex digit
1256 and thus does not indicate a floating point number when
1257 the radix is hex. */
1258 if (!hex && !got_e && (*p == 'e' || *p == 'E'))
1259 got_dot = got_e = 1;
1260 /* This test does not include !hex, because a '.' always indicates
1261 a decimal floating point number regardless of the radix. */
1262 else if (!got_dot && *p == '.')
1264 else if (got_e && (p[-1] == 'e' || p[-1] == 'E')
1265 && (*p == '-' || *p == '+'))
1266 /* This is the sign of the exponent, not the end of the
1269 /* We will take any letters or digits. parse_number will
1270 complain if past the radix, or if L or U are not final. */
1271 else if ((*p < '0' || *p > '9')
1272 && ((*p < 'a' || *p > 'z')
1273 && (*p < 'A' || *p > 'Z')))
1276 toktype = parse_number (tokstart, p - tokstart, got_dot|got_e, &yylval);
1277 if (toktype == ERROR)
1279 char *err_copy = (char *) alloca (p - tokstart + 1);
1281 memcpy (err_copy, tokstart, p - tokstart);
1282 err_copy[p - tokstart] = 0;
1283 error ("Invalid number \"%s\".", err_copy);
1315 /* Build the gdb internal form of the input string in tempbuf,
1316 translating any standard C escape forms seen. Note that the
1317 buffer is null byte terminated *only* for the convenience of
1318 debugging gdb itself and printing the buffer contents when
1319 the buffer contains no embedded nulls. Gdb does not depend
1320 upon the buffer being null byte terminated, it uses the length
1321 string instead. This allows gdb to handle C strings (as well
1322 as strings in other languages) with embedded null bytes */
1324 tokptr = ++tokstart;
1328 /* Grow the static temp buffer if necessary, including allocating
1329 the first one on demand. */
1330 if (tempbufindex + 1 >= tempbufsize)
1332 tempbuf = (char *) realloc (tempbuf, tempbufsize += 64);
1338 /* Do nothing, loop will terminate. */
1342 c = parse_escape (&tokptr);
1347 tempbuf[tempbufindex++] = c;
1350 tempbuf[tempbufindex++] = *tokptr++;
1353 } while ((*tokptr != '"') && (*tokptr != '\0'));
1354 if (*tokptr++ != '"')
1356 error ("Unterminated string in expression.");
1358 tempbuf[tempbufindex] = '\0'; /* See note above */
1359 yylval.sval.ptr = tempbuf;
1360 yylval.sval.length = tempbufindex;
1365 if (!(c == '_' || c == '$'
1366 || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z')))
1367 /* We must have come across a bad character (e.g. ';'). */
1368 error ("Invalid character '%c' in expression.", c);
1370 /* It's a name. See how long it is. */
1372 for (c = tokstart[namelen];
1373 (c == '_' || c == '$' || (c >= '0' && c <= '9')
1374 || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '<');)
1379 while (tokstart[++i] && tokstart[i] != '>');
1380 if (tokstart[i] == '>')
1383 c = tokstart[++namelen];
1386 /* The token "if" terminates the expression and is NOT
1387 removed from the input stream. */
1388 if (namelen == 2 && tokstart[0] == 'i' && tokstart[1] == 'f')
1395 /* Handle the tokens $digits; also $ (short for $0) and $$ (short for $$1)
1396 and $$digits (equivalent to $<-digits> if you could type that).
1397 Make token type LAST, and put the number (the digits) in yylval. */
1400 if (*tokstart == '$')
1402 register int negate = 0;
1404 /* Double dollar means negate the number and add -1 as well.
1405 Thus $$ alone means -1. */
1406 if (namelen >= 2 && tokstart[1] == '$')
1413 /* Just dollars (one or two) */
1414 yylval.lval = - negate;
1417 /* Is the rest of the token digits? */
1418 for (; c < namelen; c++)
1419 if (!(tokstart[c] >= '0' && tokstart[c] <= '9'))
1423 yylval.lval = atoi (tokstart + 1 + negate);
1425 yylval.lval = - yylval.lval;
1430 /* Handle tokens that refer to machine registers:
1431 $ followed by a register name. */
1433 if (*tokstart == '$') {
1434 for (c = 0; c < NUM_REGS; c++)
1435 if (namelen - 1 == strlen (reg_names[c])
1436 && STREQN (tokstart + 1, reg_names[c], namelen - 1))
1441 for (c = 0; c < num_std_regs; c++)
1442 if (namelen - 1 == strlen (std_regs[c].name)
1443 && STREQN (tokstart + 1, std_regs[c].name, namelen - 1))
1445 yylval.lval = std_regs[c].regnum;
1449 /* Catch specific keywords. Should be done with a data structure. */
1453 if (STREQN (tokstart, "unsigned", 8))
1455 if (current_language->la_language == language_cplus
1456 && STREQN (tokstart, "template", 8))
1458 if (STREQN (tokstart, "volatile", 8))
1459 return VOLATILE_KEYWORD;
1462 if (STREQN (tokstart, "struct", 6))
1464 if (STREQN (tokstart, "signed", 6))
1465 return SIGNED_KEYWORD;
1466 if (STREQN (tokstart, "sizeof", 6))
1470 if (current_language->la_language == language_cplus
1471 && STREQN (tokstart, "class", 5))
1473 if (STREQN (tokstart, "union", 5))
1475 if (STREQN (tokstart, "short", 5))
1477 if (STREQN (tokstart, "const", 5))
1478 return CONST_KEYWORD;
1481 if (STREQN (tokstart, "enum", 4))
1483 if (STREQN (tokstart, "long", 4))
1485 if (current_language->la_language == language_cplus
1486 && STREQN (tokstart, "this", 4))
1488 static const char this_name[] =
1489 { CPLUS_MARKER, 't', 'h', 'i', 's', '\0' };
1491 if (lookup_symbol (this_name, expression_context_block,
1492 VAR_NAMESPACE, (int *) NULL,
1493 (struct symtab **) NULL))
1498 if (STREQN (tokstart, "int", 3))
1505 yylval.sval.ptr = tokstart;
1506 yylval.sval.length = namelen;
1508 /* Any other names starting in $ are debugger internal variables. */
1510 if (*tokstart == '$')
1512 yylval.ivar = lookup_internalvar (copy_name (yylval.sval) + 1);
1516 /* Use token-type BLOCKNAME for symbols that happen to be defined as
1517 functions or symtabs. If this is not so, then ...
1518 Use token-type TYPENAME for symbols that happen to be defined
1519 currently as names of types; NAME for other symbols.
1520 The caller is not constrained to care about the distinction. */
1522 char *tmp = copy_name (yylval.sval);
1524 int is_a_field_of_this = 0;
1527 sym = lookup_symbol (tmp, expression_context_block,
1529 current_language->la_language == language_cplus
1530 ? &is_a_field_of_this : (int *) NULL,
1531 (struct symtab **) NULL);
1532 /* Call lookup_symtab, not lookup_partial_symtab, in case there are
1533 no psymtabs (coff, xcoff, or some future change to blow away the
1534 psymtabs once once symbols are read). */
1535 if ((sym && SYMBOL_CLASS (sym) == LOC_BLOCK) ||
1536 lookup_symtab (tmp))
1538 yylval.ssym.sym = sym;
1539 yylval.ssym.is_a_field_of_this = is_a_field_of_this;
1542 if (sym && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
1545 /* Despite the following flaw, we need to keep this code enabled.
1546 Because we can get called from check_stub_method, if we don't
1547 handle nested types then it screws many operations in any
1548 program which uses nested types. */
1549 /* In "A::x", if x is a member function of A and there happens
1550 to be a type (nested or not, since the stabs don't make that
1551 distinction) named x, then this code incorrectly thinks we
1552 are dealing with nested types rather than a member function. */
1556 struct symbol *best_sym;
1558 /* Look ahead to detect nested types. This probably should be
1559 done in the grammar, but trying seemed to introduce a lot
1560 of shift/reduce and reduce/reduce conflicts. It's possible
1561 that it could be done, though. Or perhaps a non-grammar, but
1562 less ad hoc, approach would work well. */
1564 /* Since we do not currently have any way of distinguishing
1565 a nested type from a non-nested one (the stabs don't tell
1566 us whether a type is nested), we just ignore the
1573 /* Skip whitespace. */
1574 while (*p == ' ' || *p == '\t' || *p == '\n')
1576 if (*p == ':' && p[1] == ':')
1578 /* Skip the `::'. */
1580 /* Skip whitespace. */
1581 while (*p == ' ' || *p == '\t' || *p == '\n')
1584 while (*p == '_' || *p == '$' || (*p >= '0' && *p <= '9')
1585 || (*p >= 'a' && *p <= 'z')
1586 || (*p >= 'A' && *p <= 'Z'))
1590 struct symbol *cur_sym;
1591 /* As big as the whole rest of the expression, which is
1592 at least big enough. */
1593 char *ncopy = alloca (strlen (tmp)+strlen (namestart)+3);
1597 memcpy (tmp1, tmp, strlen (tmp));
1598 tmp1 += strlen (tmp);
1599 memcpy (tmp1, "::", 2);
1601 memcpy (tmp1, namestart, p - namestart);
1602 tmp1[p - namestart] = '\0';
1603 cur_sym = lookup_symbol (ncopy, expression_context_block,
1604 VAR_NAMESPACE, (int *) NULL,
1605 (struct symtab **) NULL);
1608 if (SYMBOL_CLASS (cur_sym) == LOC_TYPEDEF)
1626 yylval.tsym.type = SYMBOL_TYPE (best_sym);
1628 yylval.tsym.type = SYMBOL_TYPE (sym);
1632 if ((yylval.tsym.type = lookup_primitive_typename (tmp)) != 0)
1635 /* Input names that aren't symbols but ARE valid hex numbers,
1636 when the input radix permits them, can be names or numbers
1637 depending on the parse. Note we support radixes > 16 here. */
1639 ((tokstart[0] >= 'a' && tokstart[0] < 'a' + input_radix - 10) ||
1640 (tokstart[0] >= 'A' && tokstart[0] < 'A' + input_radix - 10)))
1642 YYSTYPE newlval; /* Its value is ignored. */
1643 hextype = parse_number (tokstart, namelen, 0, &newlval);
1646 yylval.ssym.sym = sym;
1647 yylval.ssym.is_a_field_of_this = is_a_field_of_this;
1652 /* Any other kind of symbol */
1653 yylval.ssym.sym = sym;
1654 yylval.ssym.is_a_field_of_this = is_a_field_of_this;
1663 error ("A %s in expression, near `%s'.", (msg ? msg : "error"), lexptr);