1 /* Source-language-related definitions for GDB.
3 Copyright (C) 1991-2019 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 "gdbsupport/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 class compile_instance;
40 struct completion_match_for_lcd;
41 class innermost_block_tracker;
43 #define MAX_FORTRAN_DIMS 7 /* Maximum number of F77 array dims. */
46 range_mode_auto: range_check set automatically to default of language.
47 range_mode_manual: range_check set manually by user. */
49 extern enum range_mode
51 range_mode_auto, range_mode_manual
56 range_check_on: Ranges are checked in GDB expressions, producing errors.
57 range_check_warn: Ranges are checked, producing warnings.
58 range_check_off: Ranges are not checked in GDB expressions. */
60 extern enum range_check
62 range_check_off, range_check_warn, range_check_on
67 case_mode_auto: case_sensitivity set upon selection of scope.
68 case_mode_manual: case_sensitivity set only by user. */
72 case_mode_auto, case_mode_manual
77 array_row_major: Arrays are in row major order.
78 array_column_major: Arrays are in column major order. */
80 extern enum array_ordering
82 array_row_major, array_column_major
87 /* case_sensitivity ==
88 case_sensitive_on: Case sensitivity in name matching is used.
89 case_sensitive_off: Case sensitivity in name matching is not used. */
91 extern enum case_sensitivity
93 case_sensitive_on, case_sensitive_off
99 macro_expansion_no: No macro expansion is available.
100 macro_expansion_c: C-like macro expansion is available. */
104 macro_expansion_no, macro_expansion_c
108 /* Per architecture (OS/ABI) language information. */
110 struct language_arch_info
112 /* Its primitive types. This is a vector ended by a NULL pointer.
113 These types can be specified by name in parsing types in
114 expressions, regardless of whether the program being debugged
115 actually defines such a type. */
116 struct type **primitive_type_vector;
118 /* Symbol wrappers around primitive_type_vector, so that the symbol lookup
119 machinery can return them. */
120 struct symbol **primitive_type_symbols;
122 /* Type of elements of strings. */
123 struct type *string_char_type;
125 /* Symbol name of type to use as boolean type, if defined. */
126 const char *bool_type_symbol;
127 /* Otherwise, this is the default boolean builtin type. */
128 struct type *bool_type_default;
131 /* Structure tying together assorted information about a language. */
135 /* Name of the language. */
139 /* Natural or official name of the language. */
141 const char *la_natural_name;
143 /* its symtab language-enum (defs.h). */
145 enum language la_language;
147 /* Default range checking. */
149 enum range_check la_range_check;
151 /* Default case sensitivity. */
152 enum case_sensitivity la_case_sensitivity;
154 /* Multi-dimensional array ordering. */
155 enum array_ordering la_array_ordering;
157 /* Style of macro expansion, if any, supported by this language. */
158 enum macro_expansion la_macro_expansion;
160 /* A NULL-terminated array of file extensions for this language.
161 The extension must include the ".", like ".c". If this
162 language doesn't need to provide any filename extensions, this
165 const char *const *la_filename_extensions;
167 /* Definitions related to expression printing, prefixifying, and
170 const struct exp_descriptor *la_exp_desc;
172 /* Parser function. */
174 int (*la_parser) (struct parser_state *);
176 /* Given an expression *EXPP created by prefixifying the result of
177 la_parser, perform any remaining processing necessary to complete
178 its translation. *EXPP may change; la_post_parser is responsible
179 for releasing its previous contents, if necessary. If
180 VOID_CONTEXT_P, then no value is expected from the expression.
181 If COMPLETING is non-zero, then the expression has been parsed
182 for completion, not evaluation. */
184 void (*la_post_parser) (expression_up *expp, int void_context_p,
185 int completing, innermost_block_tracker *tracker);
187 void (*la_printchar) (int ch, struct type *chtype,
188 struct ui_file * stream);
190 void (*la_printstr) (struct ui_file * stream, struct type *elttype,
191 const gdb_byte *string, unsigned int length,
192 const char *encoding, int force_ellipses,
193 const struct value_print_options *);
195 void (*la_emitchar) (int ch, struct type *chtype,
196 struct ui_file * stream, int quoter);
198 /* Print a type using syntax appropriate for this language. */
200 void (*la_print_type) (struct type *, const char *, struct ui_file *, int,
201 int, const struct type_print_options *);
203 /* Print a typedef using syntax appropriate for this language.
204 TYPE is the underlying type. NEW_SYMBOL is the symbol naming
205 the type. STREAM is the output stream on which to print. */
207 void (*la_print_typedef) (struct type *type, struct symbol *new_symbol,
208 struct ui_file *stream);
210 /* Print a value using syntax appropriate for this language.
212 TYPE is the type of the sub-object to be printed.
214 EMBEDDED_OFFSET is the offset into the outermost object of the
215 sub-object represented by TYPE. This is the object which this
216 call should print. Note that the enclosing type is not
219 ADDRESS is the address in the inferior of the enclosing object.
221 STREAM is the stream on which the value is to be printed.
223 RECURSE is the recursion depth. It is zero-based.
225 OPTIONS are the formatting options to be used when
228 void (*la_val_print) (struct type *type,
229 int embedded_offset, CORE_ADDR address,
230 struct ui_file *stream, int recurse,
232 const struct value_print_options *options);
234 /* Print a top-level value using syntax appropriate for this language. */
236 void (*la_value_print) (struct value *, struct ui_file *,
237 const struct value_print_options *);
239 /* Given a symbol VAR, the corresponding block VAR_BLOCK (if any) and a
240 stack frame id FRAME, read the value of the variable and return (pointer
241 to a) struct value containing the value.
243 VAR_BLOCK is needed if there's a possibility for VAR to be outside
244 FRAME. This is what happens if FRAME correspond to a nested function
245 and VAR is defined in the outer function. If callers know that VAR is
246 located in FRAME or is global/static, NULL can be passed as VAR_BLOCK.
248 Throw an error if the variable cannot be found. */
250 struct value *(*la_read_var_value) (struct symbol *var,
251 const struct block *var_block,
252 struct frame_info *frame);
254 /* PC is possibly an unknown languages trampoline.
255 If that PC falls in a trampoline belonging to this language,
256 return the address of the first pc in the real function, or 0
257 if it isn't a language tramp for this language. */
258 CORE_ADDR (*skip_trampoline) (struct frame_info *, CORE_ADDR);
260 /* Now come some hooks for lookup_symbol. */
262 /* If this is non-NULL, specifies the name that of the implicit
263 local variable that refers to the current object instance. */
265 const char *la_name_of_this;
267 /* True if the symbols names should be stored in GDB's data structures
268 for minimal/partial/full symbols using their linkage (aka mangled)
269 form; false if the symbol names should be demangled first.
271 Most languages implement symbol lookup by comparing the demangled
272 names, in which case it is advantageous to store that information
273 already demangled, and so would set this field to false.
275 On the other hand, some languages have opted for doing symbol
276 lookups by comparing mangled names instead, for reasons usually
277 specific to the language. Those languages should set this field
280 And finally, other languages such as C or Asm do not have
281 the concept of mangled vs demangled name, so those languages
282 should set this field to true as well, to prevent any accidental
283 demangling through an unrelated language's demangler. */
285 const bool la_store_sym_names_in_linkage_form_p;
287 /* This is a function that lookup_symbol will call when it gets to
288 the part of symbol lookup where C looks up static and global
291 struct block_symbol (*la_lookup_symbol_nonlocal)
292 (const struct language_defn *,
294 const struct block *,
297 /* Find the definition of the type with the given name. */
298 struct type *(*la_lookup_transparent_type) (const char *);
300 /* Return demangled language symbol, or NULL. */
301 char *(*la_demangle) (const char *mangled, int options);
303 /* Demangle a symbol according to this language's rules. Unlike
304 la_demangle, this does not take any options.
306 *DEMANGLED will be set by this function.
308 If this function returns 0, then *DEMANGLED must always be set
311 If this function returns 1, the implementation may set this to
312 a xmalloc'd string holding the demangled form. However, it is
313 not required to. The string, if any, is owned by the caller.
315 The resulting string should be of the form that will be
316 installed into a symbol. */
317 int (*la_sniff_from_mangled_name) (const char *mangled, char **demangled);
319 /* Return class name of a mangled method name or NULL. */
320 char *(*la_class_name_from_physname) (const char *physname);
322 /* Table for printing expressions. */
324 const struct op_print *la_op_print_tab;
326 /* Zero if the language has first-class arrays. True if there are no
327 array values, and array objects decay to pointers, as in C. */
331 /* Index to use for extracting the first element of a string. */
332 char string_lower_bound;
334 /* The list of characters forming word boundaries. */
335 const char *(*la_word_break_characters) (void);
337 /* Add to the completion tracker all symbols which are possible
338 completions for TEXT. WORD is the entire command on which the
339 completion is being made. If CODE is TYPE_CODE_UNDEF, then all
340 symbols should be examined; otherwise, only STRUCT_DOMAIN
341 symbols whose type has a code of CODE should be matched. */
342 void (*la_collect_symbol_completion_matches)
343 (completion_tracker &tracker,
344 complete_symbol_mode mode,
345 symbol_name_match_type match_type,
348 enum type_code code);
350 /* The per-architecture (OS/ABI) language information. */
351 void (*la_language_arch_info) (struct gdbarch *,
352 struct language_arch_info *);
354 /* Print the index of an element of an array. */
355 void (*la_print_array_index) (struct value *index_value,
356 struct ui_file *stream,
357 const struct value_print_options *options);
359 /* Return non-zero if TYPE should be passed (and returned) by
360 reference at the language level. */
361 int (*la_pass_by_reference) (struct type *type);
363 /* Obtain a string from the inferior, storing it in a newly allocated
364 buffer in BUFFER, which should be freed by the caller. If the
365 in- and out-parameter *LENGTH is specified at -1, the string is
366 read until a null character of the appropriate width is found -
367 otherwise the string is read to the length of characters specified.
368 On completion, *LENGTH will hold the size of the string in characters.
369 If a *LENGTH of -1 was specified it will count only actual
370 characters, excluding any eventual terminating null character.
371 Otherwise *LENGTH will include all characters - including any nulls.
372 CHARSET will hold the encoding used in the string. */
373 void (*la_get_string) (struct value *value,
374 gdb::unique_xmalloc_ptr<gdb_byte> *buffer,
375 int *length, struct type **chartype,
376 const char **charset);
378 /* Return an expression that can be used for a location
379 watchpoint. TYPE is a pointer type that points to the memory
380 to watch, and ADDR is the address of the watched memory. */
381 gdb::unique_xmalloc_ptr<char> (*la_watch_location_expression)
382 (struct type *type, CORE_ADDR addr);
384 /* Return a pointer to the function that should be used to match a
385 symbol name against LOOKUP_NAME, according to this language's
386 rules. The matching algorithm depends on LOOKUP_NAME. For
387 example, on Ada, the matching algorithm depends on the symbol
388 name (wild/full/verbatim matching), and on whether we're doing
389 a normal lookup or a completion match lookup.
391 This field may be NULL, in which case
392 default_symbol_name_matcher is used to perform the
394 symbol_name_matcher_ftype *(*la_get_symbol_name_matcher)
395 (const lookup_name_info &);
397 /* Find all symbols in the current program space matching NAME in
398 DOMAIN, according to this language's rules.
400 The search is done in BLOCK only.
401 The caller is responsible for iterating up through superblocks
404 For each one, call CALLBACK with the symbol. If CALLBACK
405 returns false, the iteration ends at that point.
407 This field may not be NULL. If the language does not need any
408 special processing here, 'iterate_over_symbols' should be
409 used as the definition. */
410 void (*la_iterate_over_symbols)
411 (const struct block *block, const lookup_name_info &name,
413 gdb::function_view<symbol_found_callback_ftype> callback);
415 /* Hash the given symbol search name. Use
416 default_search_name_hash if no special treatment is
418 unsigned int (*la_search_name_hash) (const char *name);
420 /* Various operations on varobj. */
421 const struct lang_varobj_ops *la_varobj_ops;
423 /* If this language allows compilation from the gdb command line,
424 this method should be non-NULL. When called it should return
425 an instance of struct gcc_context appropriate to the language.
426 When defined this method must never return NULL; instead it
427 should throw an exception on failure. The returned compiler
428 instance is owned by its caller and must be deallocated by
429 calling its 'destroy' method. */
431 compile_instance *(*la_get_compile_instance) (void);
433 /* This method must be defined if 'la_get_gcc_context' is defined.
434 If 'la_get_gcc_context' is not defined, then this method is
437 This takes the user-supplied text and returns a new bit of code
440 INST is the compiler instance being used.
441 INPUT is the user's input text.
442 GDBARCH is the architecture to use.
443 EXPR_BLOCK is the block in which the expression is being
445 EXPR_PC is the PC at which the expression is being parsed. */
447 std::string (*la_compute_program) (compile_instance *inst,
449 struct gdbarch *gdbarch,
450 const struct block *expr_block,
453 /* Return true if TYPE is a string type. */
454 bool (*la_is_string_type_p) (struct type *type);
456 /* This string is used by the 'set print max-depth' setting. When GDB
457 replaces a struct or union (during value printing) that is "too
458 deep" this string is displayed instead. */
459 const char *la_struct_too_deep_ellipsis;
463 /* Pointer to the language_defn for our current language. This pointer
464 always points to *some* valid struct; it can be used without checking
467 The current language affects expression parsing and evaluation
468 (FIXME: it might be cleaner to make the evaluation-related stuff
469 separate exp_opcodes for each different set of semantics. We
470 should at least think this through more clearly with respect to
471 what happens if the language is changed between parsing and
472 evaluation) and printing of things like types and arrays. It does
473 *not* affect symbol-reading-- each source file in a symbol-file has
474 its own language and we should keep track of that regardless of the
475 language when symbols are read. If we want some manual setting for
476 the language of symbol files (e.g. detecting when ".c" files are
477 C++), it should be a separate setting from the current_language. */
479 extern const struct language_defn *current_language;
481 /* Pointer to the language_defn expected by the user, e.g. the language
482 of main(), or the language we last mentioned in a message, or C. */
484 extern const struct language_defn *expected_language;
487 language_mode_auto: current_language automatically set upon selection
488 of scope (e.g. stack frame)
489 language_mode_manual: current_language set only by user. */
491 extern enum language_mode
493 language_mode_auto, language_mode_manual
497 struct type *language_bool_type (const struct language_defn *l,
498 struct gdbarch *gdbarch);
500 struct type *language_string_char_type (const struct language_defn *l,
501 struct gdbarch *gdbarch);
503 /* Look up type NAME in language L, and return its definition for architecture
504 GDBARCH. Returns NULL if not found. */
506 struct type *language_lookup_primitive_type (const struct language_defn *l,
507 struct gdbarch *gdbarch,
510 /* Wrapper around language_lookup_primitive_type to return the
511 corresponding symbol. */
514 language_lookup_primitive_type_as_symbol (const struct language_defn *l,
515 struct gdbarch *gdbarch,
519 /* These macros define the behaviour of the expression
522 /* Should we range check values against the domain of their type? */
523 #define RANGE_CHECK (range_check != range_check_off)
525 /* "cast" really means conversion. */
526 /* FIXME -- should be a setting in language_defn. */
527 #define CAST_IS_CONVERSION(LANG) ((LANG)->la_language == language_c || \
528 (LANG)->la_language == language_cplus || \
529 (LANG)->la_language == language_objc)
531 extern void language_info (int);
533 extern enum language set_language (enum language);
536 /* This page contains functions that return things that are
537 specific to languages. Each of these functions is based on
538 the current setting of working_lang, which the user sets
539 with the "set language" command. */
541 #define LA_PRINT_TYPE(type,varstring,stream,show,level,flags) \
542 (current_language->la_print_type(type,varstring,stream,show,level,flags))
544 #define LA_PRINT_TYPEDEF(type,new_symbol,stream) \
545 (current_language->la_print_typedef(type,new_symbol,stream))
547 #define LA_VALUE_PRINT(val,stream,options) \
548 (current_language->la_value_print(val,stream,options))
550 #define LA_PRINT_CHAR(ch, type, stream) \
551 (current_language->la_printchar(ch, type, stream))
552 #define LA_PRINT_STRING(stream, elttype, string, length, encoding, force_ellipses, options) \
553 (current_language->la_printstr(stream, elttype, string, length, \
554 encoding, force_ellipses,options))
555 #define LA_EMIT_CHAR(ch, type, stream, quoter) \
556 (current_language->la_emitchar(ch, type, stream, quoter))
557 #define LA_GET_STRING(value, buffer, length, chartype, encoding) \
558 (current_language->la_get_string(value, buffer, length, chartype, encoding))
560 #define LA_PRINT_ARRAY_INDEX(index_value, stream, options) \
561 (current_language->la_print_array_index(index_value, stream, options))
563 #define LA_ITERATE_OVER_SYMBOLS(BLOCK, NAME, DOMAIN, CALLBACK) \
564 (current_language->la_iterate_over_symbols (BLOCK, NAME, DOMAIN, CALLBACK))
566 /* Test a character to decide whether it can be printed in literal form
567 or needs to be printed in another representation. For example,
568 in C the literal form of the character with octal value 141 is 'a'
569 and the "other representation" is '\141'. The "other representation"
570 is program language dependent. */
572 #define PRINT_LITERAL_FORM(c) \
574 && ((c) < 0x7F || (c) >= 0xA0) \
575 && (!sevenbit_strings || (c) < 0x80))
577 /* Type predicates */
579 extern int pointer_type (struct type *);
581 /* Return true if TYPE is a string type, otherwise return false. This
582 default implementation only detects TYPE_CODE_STRING. */
583 extern bool default_is_string_type_p (struct type *type);
587 extern void range_error (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
589 /* Data: Does this value represent "truth" to the current language? */
591 extern int value_true (struct value *);
593 /* Misc: The string representing a particular enum language. */
595 extern enum language language_enum (const char *str);
597 extern const struct language_defn *language_def (enum language);
599 extern const char *language_str (enum language);
601 /* Check for a language-specific trampoline. */
603 extern CORE_ADDR skip_language_trampoline (struct frame_info *, CORE_ADDR pc);
605 /* Return demangled language symbol, or NULL. */
606 extern char *language_demangle (const struct language_defn *current_language,
607 const char *mangled, int options);
609 /* A wrapper for la_sniff_from_mangled_name. The arguments and result
610 are as for the method. */
612 extern int language_sniff_from_mangled_name (const struct language_defn *lang,
616 /* Return class name from physname, or NULL. */
617 extern char *language_class_name_from_physname (const struct language_defn *,
618 const char *physname);
620 /* Splitting strings into words. */
621 extern const char *default_word_break_characters (void);
623 /* Print the index of an array element using the C99 syntax. */
624 extern void default_print_array_index (struct value *index_value,
625 struct ui_file *stream,
626 const struct value_print_options *options);
628 /* Return non-zero if TYPE should be passed (and returned) by
629 reference at the language level. */
630 int language_pass_by_reference (struct type *type);
632 /* Return zero; by default, types are passed by value at the language
633 level. The target ABI may pass or return some structs by reference
634 independent of this. */
635 int default_pass_by_reference (struct type *type);
637 /* The default implementation of la_print_typedef. */
638 void default_print_typedef (struct type *type, struct symbol *new_symbol,
639 struct ui_file *stream);
641 void default_get_string (struct value *value,
642 gdb::unique_xmalloc_ptr<gdb_byte> *buffer,
643 int *length, struct type **char_type,
644 const char **charset);
646 /* Default name hashing function. */
648 /* Produce an unsigned hash value from SEARCH_NAME that is consistent
649 with strcmp_iw, strcmp, and, at least on Ada symbols, wild_match.
650 That is, two identifiers equivalent according to any of those three
651 comparison operators hash to the same value. */
652 extern unsigned int default_search_name_hash (const char *search_name);
654 void c_get_string (struct value *value,
655 gdb::unique_xmalloc_ptr<gdb_byte> *buffer,
656 int *length, struct type **char_type,
657 const char **charset);
659 /* The default implementation of la_symbol_name_matcher. Matches with
661 extern bool default_symbol_name_matcher
662 (const char *symbol_search_name,
663 const lookup_name_info &lookup_name,
664 completion_match_result *comp_match_res);
666 /* Get LANG's symbol_name_matcher method for LOOKUP_NAME. Returns
667 default_symbol_name_matcher if not set. LANG is used as a hint;
668 the function may ignore it depending on the current language and
669 LOOKUP_NAME. Specifically, if the current language is Ada, this
670 may return an Ada matcher regardless of LANG. */
671 symbol_name_matcher_ftype *get_symbol_name_matcher
672 (const language_defn *lang, const lookup_name_info &lookup_name);
674 /* The languages supported by GDB. */
676 extern const struct language_defn auto_language_defn;
677 extern const struct language_defn unknown_language_defn;
678 extern const struct language_defn minimal_language_defn;
680 extern const struct language_defn ada_language_defn;
681 extern const struct language_defn asm_language_defn;
682 extern const struct language_defn c_language_defn;
683 extern const struct language_defn cplus_language_defn;
684 extern const struct language_defn d_language_defn;
685 extern const struct language_defn f_language_defn;
686 extern const struct language_defn go_language_defn;
687 extern const struct language_defn m2_language_defn;
688 extern const struct language_defn objc_language_defn;
689 extern const struct language_defn opencl_language_defn;
690 extern const struct language_defn pascal_language_defn;
691 extern const struct language_defn rust_language_defn;
693 /* Save the current language and restore it upon destruction. */
695 class scoped_restore_current_language
699 explicit scoped_restore_current_language ()
700 : m_lang (current_language->la_language)
704 ~scoped_restore_current_language ()
706 set_language (m_lang);
709 scoped_restore_current_language (const scoped_restore_current_language &)
711 scoped_restore_current_language &operator=
712 (const scoped_restore_current_language &) = delete;
716 enum language m_lang;
719 /* If language_mode is language_mode_auto,
720 then switch current language to the language of SYM
721 and restore current language upon destruction.
725 class scoped_switch_to_sym_language_if_auto
729 explicit scoped_switch_to_sym_language_if_auto (const struct symbol *sym)
731 if (language_mode == language_mode_auto)
733 m_lang = current_language->la_language;
735 set_language (SYMBOL_LANGUAGE (sym));
740 /* Assign to m_lang to silence a GCC warning. See
741 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80635. */
742 m_lang = language_unknown;
746 ~scoped_switch_to_sym_language_if_auto ()
749 set_language (m_lang);
752 DISABLE_COPY_AND_ASSIGN (scoped_switch_to_sym_language_if_auto);
756 enum language m_lang;
759 #endif /* defined (LANGUAGE_H) */