1 /* Source-language-related definitions for GDB.
3 Copyright (C) 1991-2018 Free Software Foundation, Inc.
5 Contributed by the Department of Computer Science at the State University
6 of New York at Buffalo.
8 This file is part of GDB.
10 This program is free software; you can redistribute it and/or modify
11 it under the terms of the GNU General Public License as published by
12 the Free Software Foundation; either version 3 of the License, or
13 (at your option) any later version.
15 This program is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 GNU General Public License for more details.
20 You should have received a copy of the GNU General Public License
21 along with this program. If not, see <http://www.gnu.org/licenses/>. */
23 #if !defined (LANGUAGE_H)
27 #include "common/function-view.h"
28 #include "expression.h"
30 /* Forward decls for prototypes. */
35 struct value_print_options;
36 struct type_print_options;
37 struct lang_varobj_ops;
39 struct compile_instance;
40 struct completion_match_for_lcd;
42 #define MAX_FORTRAN_DIMS 7 /* Maximum number of F77 array dims. */
45 range_mode_auto: range_check set automatically to default of language.
46 range_mode_manual: range_check set manually by user. */
48 extern enum range_mode
50 range_mode_auto, range_mode_manual
55 range_check_on: Ranges are checked in GDB expressions, producing errors.
56 range_check_warn: Ranges are checked, producing warnings.
57 range_check_off: Ranges are not checked in GDB expressions. */
59 extern enum range_check
61 range_check_off, range_check_warn, range_check_on
66 case_mode_auto: case_sensitivity set upon selection of scope.
67 case_mode_manual: case_sensitivity set only by user. */
71 case_mode_auto, case_mode_manual
76 array_row_major: Arrays are in row major order.
77 array_column_major: Arrays are in column major order. */
79 extern enum array_ordering
81 array_row_major, array_column_major
86 /* case_sensitivity ==
87 case_sensitive_on: Case sensitivity in name matching is used.
88 case_sensitive_off: Case sensitivity in name matching is not used. */
90 extern enum case_sensitivity
92 case_sensitive_on, case_sensitive_off
98 macro_expansion_no: No macro expansion is available.
99 macro_expansion_c: C-like macro expansion is available. */
103 macro_expansion_no, macro_expansion_c
107 /* Per architecture (OS/ABI) language information. */
109 struct language_arch_info
111 /* Its primitive types. This is a vector ended by a NULL pointer.
112 These types can be specified by name in parsing types in
113 expressions, regardless of whether the program being debugged
114 actually defines such a type. */
115 struct type **primitive_type_vector;
117 /* Symbol wrappers around primitive_type_vector, so that the symbol lookup
118 machinery can return them. */
119 struct symbol **primitive_type_symbols;
121 /* Type of elements of strings. */
122 struct type *string_char_type;
124 /* Symbol name of type to use as boolean type, if defined. */
125 const char *bool_type_symbol;
126 /* Otherwise, this is the default boolean builtin type. */
127 struct type *bool_type_default;
130 /* Structure tying together assorted information about a language. */
134 /* Name of the language. */
138 /* Natural or official name of the language. */
140 const char *la_natural_name;
142 /* its symtab language-enum (defs.h). */
144 enum language la_language;
146 /* Default range checking. */
148 enum range_check la_range_check;
150 /* Default case sensitivity. */
151 enum case_sensitivity la_case_sensitivity;
153 /* Multi-dimensional array ordering. */
154 enum array_ordering la_array_ordering;
156 /* Style of macro expansion, if any, supported by this language. */
157 enum macro_expansion la_macro_expansion;
159 /* A NULL-terminated array of file extensions for this language.
160 The extension must include the ".", like ".c". If this
161 language doesn't need to provide any filename extensions, this
164 const char *const *la_filename_extensions;
166 /* Definitions related to expression printing, prefixifying, and
169 const struct exp_descriptor *la_exp_desc;
171 /* Parser function. */
173 int (*la_parser) (struct parser_state *);
175 /* Given an expression *EXPP created by prefixifying the result of
176 la_parser, perform any remaining processing necessary to complete
177 its translation. *EXPP may change; la_post_parser is responsible
178 for releasing its previous contents, if necessary. If
179 VOID_CONTEXT_P, then no value is expected from the expression. */
181 void (*la_post_parser) (expression_up *expp, int void_context_p);
183 void (*la_printchar) (int ch, struct type *chtype,
184 struct ui_file * stream);
186 void (*la_printstr) (struct ui_file * stream, struct type *elttype,
187 const gdb_byte *string, unsigned int length,
188 const char *encoding, int force_ellipses,
189 const struct value_print_options *);
191 void (*la_emitchar) (int ch, struct type *chtype,
192 struct ui_file * stream, int quoter);
194 /* Print a type using syntax appropriate for this language. */
196 void (*la_print_type) (struct type *, const char *, struct ui_file *, int,
197 int, const struct type_print_options *);
199 /* Print a typedef using syntax appropriate for this language.
200 TYPE is the underlying type. NEW_SYMBOL is the symbol naming
201 the type. STREAM is the output stream on which to print. */
203 void (*la_print_typedef) (struct type *type, struct symbol *new_symbol,
204 struct ui_file *stream);
206 /* Print a value using syntax appropriate for this language.
208 TYPE is the type of the sub-object to be printed.
210 EMBEDDED_OFFSET is the offset into the outermost object of the
211 sub-object represented by TYPE. This is the object which this
212 call should print. Note that the enclosing type is not
215 ADDRESS is the address in the inferior of the enclosing object.
217 STREAM is the stream on which the value is to be printed.
219 RECURSE is the recursion depth. It is zero-based.
221 OPTIONS are the formatting options to be used when
224 void (*la_val_print) (struct type *type,
225 int embedded_offset, CORE_ADDR address,
226 struct ui_file *stream, int recurse,
228 const struct value_print_options *options);
230 /* Print a top-level value using syntax appropriate for this language. */
232 void (*la_value_print) (struct value *, struct ui_file *,
233 const struct value_print_options *);
235 /* Given a symbol VAR, the corresponding block VAR_BLOCK (if any) and a
236 stack frame id FRAME, read the value of the variable and return (pointer
237 to a) struct value containing the value.
239 VAR_BLOCK is needed if there's a possibility for VAR to be outside
240 FRAME. This is what happens if FRAME correspond to a nested function
241 and VAR is defined in the outer function. If callers know that VAR is
242 located in FRAME or is global/static, NULL can be passed as VAR_BLOCK.
244 Throw an error if the variable cannot be found. */
246 struct value *(*la_read_var_value) (struct symbol *var,
247 const struct block *var_block,
248 struct frame_info *frame);
250 /* PC is possibly an unknown languages trampoline.
251 If that PC falls in a trampoline belonging to this language,
252 return the address of the first pc in the real function, or 0
253 if it isn't a language tramp for this language. */
254 CORE_ADDR (*skip_trampoline) (struct frame_info *, CORE_ADDR);
256 /* Now come some hooks for lookup_symbol. */
258 /* If this is non-NULL, specifies the name that of the implicit
259 local variable that refers to the current object instance. */
261 const char *la_name_of_this;
263 /* True if the symbols names should be stored in GDB's data structures
264 for minimal/partial/full symbols using their linkage (aka mangled)
265 form; false if the symbol names should be demangled first.
267 Most languages implement symbol lookup by comparing the demangled
268 names, in which case it is advantageous to store that information
269 already demangled, and so would set this field to false.
271 On the other hand, some languages have opted for doing symbol
272 lookups by comparing mangled names instead, for reasons usually
273 specific to the language. Those languages should set this field
276 And finally, other languages such as C or Asm do not have
277 the concept of mangled vs demangled name, so those languages
278 should set this field to true as well, to prevent any accidental
279 demangling through an unrelated language's demangler. */
281 const bool la_store_sym_names_in_linkage_form_p;
283 /* This is a function that lookup_symbol will call when it gets to
284 the part of symbol lookup where C looks up static and global
287 struct block_symbol (*la_lookup_symbol_nonlocal)
288 (const struct language_defn *,
290 const struct block *,
293 /* Find the definition of the type with the given name. */
294 struct type *(*la_lookup_transparent_type) (const char *);
296 /* Return demangled language symbol, or NULL. */
297 char *(*la_demangle) (const char *mangled, int options);
299 /* Demangle a symbol according to this language's rules. Unlike
300 la_demangle, this does not take any options.
302 *DEMANGLED will be set by this function.
304 If this function returns 0, then *DEMANGLED must always be set
307 If this function returns 1, the implementation may set this to
308 a xmalloc'd string holding the demangled form. However, it is
309 not required to. The string, if any, is owned by the caller.
311 The resulting string should be of the form that will be
312 installed into a symbol. */
313 int (*la_sniff_from_mangled_name) (const char *mangled, char **demangled);
315 /* Return class name of a mangled method name or NULL. */
316 char *(*la_class_name_from_physname) (const char *physname);
318 /* Table for printing expressions. */
320 const struct op_print *la_op_print_tab;
322 /* Zero if the language has first-class arrays. True if there are no
323 array values, and array objects decay to pointers, as in C. */
327 /* Index to use for extracting the first element of a string. */
328 char string_lower_bound;
330 /* The list of characters forming word boundaries. */
331 const char *(*la_word_break_characters) (void);
333 /* Add to the completion tracker all symbols which are possible
334 completions for TEXT. WORD is the entire command on which the
335 completion is being made. If CODE is TYPE_CODE_UNDEF, then all
336 symbols should be examined; otherwise, only STRUCT_DOMAIN
337 symbols whose type has a code of CODE should be matched. */
338 void (*la_collect_symbol_completion_matches)
339 (completion_tracker &tracker,
340 complete_symbol_mode mode,
341 symbol_name_match_type match_type,
344 enum type_code code);
346 /* The per-architecture (OS/ABI) language information. */
347 void (*la_language_arch_info) (struct gdbarch *,
348 struct language_arch_info *);
350 /* Print the index of an element of an array. */
351 void (*la_print_array_index) (struct value *index_value,
352 struct ui_file *stream,
353 const struct value_print_options *options);
355 /* Return non-zero if TYPE should be passed (and returned) by
356 reference at the language level. */
357 int (*la_pass_by_reference) (struct type *type);
359 /* Obtain a string from the inferior, storing it in a newly allocated
360 buffer in BUFFER, which should be freed by the caller. If the
361 in- and out-parameter *LENGTH is specified at -1, the string is
362 read until a null character of the appropriate width is found -
363 otherwise the string is read to the length of characters specified.
364 On completion, *LENGTH will hold the size of the string in characters.
365 If a *LENGTH of -1 was specified it will count only actual
366 characters, excluding any eventual terminating null character.
367 Otherwise *LENGTH will include all characters - including any nulls.
368 CHARSET will hold the encoding used in the string. */
369 void (*la_get_string) (struct value *value, gdb_byte **buffer, int *length,
370 struct type **chartype, const char **charset);
372 /* Return an expression that can be used for a location
373 watchpoint. TYPE is a pointer type that points to the memory
374 to watch, and ADDR is the address of the watched memory. */
375 gdb::unique_xmalloc_ptr<char> (*la_watch_location_expression)
376 (struct type *type, CORE_ADDR addr);
378 /* Return a pointer to the function that should be used to match a
379 symbol name against LOOKUP_NAME, according to this language's
380 rules. The matching algorithm depends on LOOKUP_NAME. For
381 example, on Ada, the matching algorithm depends on the symbol
382 name (wild/full/verbatim matching), and on whether we're doing
383 a normal lookup or a completion match lookup.
385 This field may be NULL, in which case
386 default_symbol_name_matcher is used to perform the
388 symbol_name_matcher_ftype *(*la_get_symbol_name_matcher)
389 (const lookup_name_info &);
391 /* Find all symbols in the current program space matching NAME in
392 DOMAIN, according to this language's rules.
394 The search is done in BLOCK only.
395 The caller is responsible for iterating up through superblocks
398 For each one, call CALLBACK with the symbol. If CALLBACK
399 returns false, the iteration ends at that point.
401 This field may not be NULL. If the language does not need any
402 special processing here, 'iterate_over_symbols' should be
403 used as the definition. */
404 void (*la_iterate_over_symbols)
405 (const struct block *block, const lookup_name_info &name,
407 gdb::function_view<symbol_found_callback_ftype> callback);
409 /* Hash the given symbol search name. Use
410 default_search_name_hash if no special treatment is
412 unsigned int (*la_search_name_hash) (const char *name);
414 /* Various operations on varobj. */
415 const struct lang_varobj_ops *la_varobj_ops;
417 /* If this language allows compilation from the gdb command line,
418 this method should be non-NULL. When called it should return
419 an instance of struct gcc_context appropriate to the language.
420 When defined this method must never return NULL; instead it
421 should throw an exception on failure. The returned compiler
422 instance is owned by its caller and must be deallocated by
423 calling its 'destroy' method. */
425 struct compile_instance *(*la_get_compile_instance) (void);
427 /* This method must be defined if 'la_get_gcc_context' is defined.
428 If 'la_get_gcc_context' is not defined, then this method is
431 This takes the user-supplied text and returns a new bit of code
434 INST is the compiler instance being used.
435 INPUT is the user's input text.
436 GDBARCH is the architecture to use.
437 EXPR_BLOCK is the block in which the expression is being
439 EXPR_PC is the PC at which the expression is being parsed. */
441 std::string (*la_compute_program) (struct compile_instance *inst,
443 struct gdbarch *gdbarch,
444 const struct block *expr_block,
447 /* Add fields above this point, so the magic number is always last. */
448 /* Magic number for compat checking. */
454 #define LANG_MAGIC 910823L
456 /* Pointer to the language_defn for our current language. This pointer
457 always points to *some* valid struct; it can be used without checking
460 The current language affects expression parsing and evaluation
461 (FIXME: it might be cleaner to make the evaluation-related stuff
462 separate exp_opcodes for each different set of semantics. We
463 should at least think this through more clearly with respect to
464 what happens if the language is changed between parsing and
465 evaluation) and printing of things like types and arrays. It does
466 *not* affect symbol-reading-- each source file in a symbol-file has
467 its own language and we should keep track of that regardless of the
468 language when symbols are read. If we want some manual setting for
469 the language of symbol files (e.g. detecting when ".c" files are
470 C++), it should be a separate setting from the current_language. */
472 extern const struct language_defn *current_language;
474 /* Pointer to the language_defn expected by the user, e.g. the language
475 of main(), or the language we last mentioned in a message, or C. */
477 extern const struct language_defn *expected_language;
480 language_mode_auto: current_language automatically set upon selection
481 of scope (e.g. stack frame)
482 language_mode_manual: current_language set only by user. */
484 extern enum language_mode
486 language_mode_auto, language_mode_manual
490 struct type *language_bool_type (const struct language_defn *l,
491 struct gdbarch *gdbarch);
493 struct type *language_string_char_type (const struct language_defn *l,
494 struct gdbarch *gdbarch);
496 /* Look up type NAME in language L, and return its definition for architecture
497 GDBARCH. Returns NULL if not found. */
499 struct type *language_lookup_primitive_type (const struct language_defn *l,
500 struct gdbarch *gdbarch,
503 /* Wrapper around language_lookup_primitive_type to return the
504 corresponding symbol. */
507 language_lookup_primitive_type_as_symbol (const struct language_defn *l,
508 struct gdbarch *gdbarch,
512 /* These macros define the behaviour of the expression
515 /* Should we range check values against the domain of their type? */
516 #define RANGE_CHECK (range_check != range_check_off)
518 /* "cast" really means conversion. */
519 /* FIXME -- should be a setting in language_defn. */
520 #define CAST_IS_CONVERSION(LANG) ((LANG)->la_language == language_c || \
521 (LANG)->la_language == language_cplus || \
522 (LANG)->la_language == language_objc)
524 extern void language_info (int);
526 extern enum language set_language (enum language);
529 /* This page contains functions that return things that are
530 specific to languages. Each of these functions is based on
531 the current setting of working_lang, which the user sets
532 with the "set language" command. */
534 #define LA_PRINT_TYPE(type,varstring,stream,show,level,flags) \
535 (current_language->la_print_type(type,varstring,stream,show,level,flags))
537 #define LA_PRINT_TYPEDEF(type,new_symbol,stream) \
538 (current_language->la_print_typedef(type,new_symbol,stream))
540 #define LA_VALUE_PRINT(val,stream,options) \
541 (current_language->la_value_print(val,stream,options))
543 #define LA_PRINT_CHAR(ch, type, stream) \
544 (current_language->la_printchar(ch, type, stream))
545 #define LA_PRINT_STRING(stream, elttype, string, length, encoding, force_ellipses, options) \
546 (current_language->la_printstr(stream, elttype, string, length, \
547 encoding, force_ellipses,options))
548 #define LA_EMIT_CHAR(ch, type, stream, quoter) \
549 (current_language->la_emitchar(ch, type, stream, quoter))
550 #define LA_GET_STRING(value, buffer, length, chartype, encoding) \
551 (current_language->la_get_string(value, buffer, length, chartype, encoding))
553 #define LA_PRINT_ARRAY_INDEX(index_value, stream, options) \
554 (current_language->la_print_array_index(index_value, stream, options))
556 #define LA_ITERATE_OVER_SYMBOLS(BLOCK, NAME, DOMAIN, CALLBACK) \
557 (current_language->la_iterate_over_symbols (BLOCK, NAME, DOMAIN, CALLBACK))
559 /* Test a character to decide whether it can be printed in literal form
560 or needs to be printed in another representation. For example,
561 in C the literal form of the character with octal value 141 is 'a'
562 and the "other representation" is '\141'. The "other representation"
563 is program language dependent. */
565 #define PRINT_LITERAL_FORM(c) \
567 && ((c) < 0x7F || (c) >= 0xA0) \
568 && (!sevenbit_strings || (c) < 0x80))
570 /* Type predicates */
572 extern int pointer_type (struct type *);
576 extern void range_error (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
578 /* Data: Does this value represent "truth" to the current language? */
580 extern int value_true (struct value *);
582 /* Misc: The string representing a particular enum language. */
584 extern enum language language_enum (const char *str);
586 extern const struct language_defn *language_def (enum language);
588 extern const char *language_str (enum language);
590 /* Check for a language-specific trampoline. */
592 extern CORE_ADDR skip_language_trampoline (struct frame_info *, CORE_ADDR pc);
594 /* Return demangled language symbol, or NULL. */
595 extern char *language_demangle (const struct language_defn *current_language,
596 const char *mangled, int options);
598 /* A wrapper for la_sniff_from_mangled_name. The arguments and result
599 are as for the method. */
601 extern int language_sniff_from_mangled_name (const struct language_defn *lang,
605 /* Return class name from physname, or NULL. */
606 extern char *language_class_name_from_physname (const struct language_defn *,
607 const char *physname);
609 /* Splitting strings into words. */
610 extern const char *default_word_break_characters (void);
612 /* Print the index of an array element using the C99 syntax. */
613 extern void default_print_array_index (struct value *index_value,
614 struct ui_file *stream,
615 const struct value_print_options *options);
617 /* Return non-zero if TYPE should be passed (and returned) by
618 reference at the language level. */
619 int language_pass_by_reference (struct type *type);
621 /* Return zero; by default, types are passed by value at the language
622 level. The target ABI may pass or return some structs by reference
623 independent of this. */
624 int default_pass_by_reference (struct type *type);
626 /* The default implementation of la_print_typedef. */
627 void default_print_typedef (struct type *type, struct symbol *new_symbol,
628 struct ui_file *stream);
630 void default_get_string (struct value *value, gdb_byte **buffer, int *length,
631 struct type **char_type, const char **charset);
633 /* Default name hashing function. */
635 /* Produce an unsigned hash value from SEARCH_NAME that is consistent
636 with strcmp_iw, strcmp, and, at least on Ada symbols, wild_match.
637 That is, two identifiers equivalent according to any of those three
638 comparison operators hash to the same value. */
639 extern unsigned int default_search_name_hash (const char *search_name);
641 void c_get_string (struct value *value, gdb_byte **buffer, int *length,
642 struct type **char_type, const char **charset);
644 /* The default implementation of la_symbol_name_matcher. Matches with
646 extern bool default_symbol_name_matcher
647 (const char *symbol_search_name,
648 const lookup_name_info &lookup_name,
649 completion_match_result *comp_match_res);
651 /* Get LANG's symbol_name_matcher method for LOOKUP_NAME. Returns
652 default_symbol_name_matcher if not set. LANG is used as a hint;
653 the function may ignore it depending on the current language and
654 LOOKUP_NAME. Specifically, if the current language is Ada, this
655 may return an Ada matcher regardless of LANG. */
656 symbol_name_matcher_ftype *get_symbol_name_matcher
657 (const language_defn *lang, const lookup_name_info &lookup_name);
659 /* The languages supported by GDB. */
661 extern const struct language_defn auto_language_defn;
662 extern const struct language_defn unknown_language_defn;
663 extern const struct language_defn minimal_language_defn;
665 extern const struct language_defn ada_language_defn;
666 extern const struct language_defn asm_language_defn;
667 extern const struct language_defn c_language_defn;
668 extern const struct language_defn cplus_language_defn;
669 extern const struct language_defn d_language_defn;
670 extern const struct language_defn f_language_defn;
671 extern const struct language_defn go_language_defn;
672 extern const struct language_defn m2_language_defn;
673 extern const struct language_defn objc_language_defn;
674 extern const struct language_defn opencl_language_defn;
675 extern const struct language_defn pascal_language_defn;
676 extern const struct language_defn rust_language_defn;
678 /* Save the current language and restore it upon destruction. */
680 class scoped_restore_current_language
684 explicit scoped_restore_current_language ()
685 : m_lang (current_language->la_language)
689 ~scoped_restore_current_language ()
691 set_language (m_lang);
694 scoped_restore_current_language (const scoped_restore_current_language &)
696 scoped_restore_current_language &operator=
697 (const scoped_restore_current_language &) = delete;
701 enum language m_lang;
704 #endif /* defined (LANGUAGE_H) */