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 "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 class 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.
180 If COMPLETING is non-zero, then the expression has been parsed
181 for completion, not evaluation. */
183 void (*la_post_parser
) (expression_up
*expp
, int void_context_p
,
186 void (*la_printchar
) (int ch
, struct type
*chtype
,
187 struct ui_file
* stream
);
189 void (*la_printstr
) (struct ui_file
* stream
, struct type
*elttype
,
190 const gdb_byte
*string
, unsigned int length
,
191 const char *encoding
, int force_ellipses
,
192 const struct value_print_options
*);
194 void (*la_emitchar
) (int ch
, struct type
*chtype
,
195 struct ui_file
* stream
, int quoter
);
197 /* Print a type using syntax appropriate for this language. */
199 void (*la_print_type
) (struct type
*, const char *, struct ui_file
*, int,
200 int, const struct type_print_options
*);
202 /* Print a typedef using syntax appropriate for this language.
203 TYPE is the underlying type. NEW_SYMBOL is the symbol naming
204 the type. STREAM is the output stream on which to print. */
206 void (*la_print_typedef
) (struct type
*type
, struct symbol
*new_symbol
,
207 struct ui_file
*stream
);
209 /* Print a value using syntax appropriate for this language.
211 TYPE is the type of the sub-object to be printed.
213 EMBEDDED_OFFSET is the offset into the outermost object of the
214 sub-object represented by TYPE. This is the object which this
215 call should print. Note that the enclosing type is not
218 ADDRESS is the address in the inferior of the enclosing object.
220 STREAM is the stream on which the value is to be printed.
222 RECURSE is the recursion depth. It is zero-based.
224 OPTIONS are the formatting options to be used when
227 void (*la_val_print
) (struct type
*type
,
228 int embedded_offset
, CORE_ADDR address
,
229 struct ui_file
*stream
, int recurse
,
231 const struct value_print_options
*options
);
233 /* Print a top-level value using syntax appropriate for this language. */
235 void (*la_value_print
) (struct value
*, struct ui_file
*,
236 const struct value_print_options
*);
238 /* Given a symbol VAR, the corresponding block VAR_BLOCK (if any) and a
239 stack frame id FRAME, read the value of the variable and return (pointer
240 to a) struct value containing the value.
242 VAR_BLOCK is needed if there's a possibility for VAR to be outside
243 FRAME. This is what happens if FRAME correspond to a nested function
244 and VAR is defined in the outer function. If callers know that VAR is
245 located in FRAME or is global/static, NULL can be passed as VAR_BLOCK.
247 Throw an error if the variable cannot be found. */
249 struct value
*(*la_read_var_value
) (struct symbol
*var
,
250 const struct block
*var_block
,
251 struct frame_info
*frame
);
253 /* PC is possibly an unknown languages trampoline.
254 If that PC falls in a trampoline belonging to this language,
255 return the address of the first pc in the real function, or 0
256 if it isn't a language tramp for this language. */
257 CORE_ADDR (*skip_trampoline
) (struct frame_info
*, CORE_ADDR
);
259 /* Now come some hooks for lookup_symbol. */
261 /* If this is non-NULL, specifies the name that of the implicit
262 local variable that refers to the current object instance. */
264 const char *la_name_of_this
;
266 /* True if the symbols names should be stored in GDB's data structures
267 for minimal/partial/full symbols using their linkage (aka mangled)
268 form; false if the symbol names should be demangled first.
270 Most languages implement symbol lookup by comparing the demangled
271 names, in which case it is advantageous to store that information
272 already demangled, and so would set this field to false.
274 On the other hand, some languages have opted for doing symbol
275 lookups by comparing mangled names instead, for reasons usually
276 specific to the language. Those languages should set this field
279 And finally, other languages such as C or Asm do not have
280 the concept of mangled vs demangled name, so those languages
281 should set this field to true as well, to prevent any accidental
282 demangling through an unrelated language's demangler. */
284 const bool la_store_sym_names_in_linkage_form_p
;
286 /* This is a function that lookup_symbol will call when it gets to
287 the part of symbol lookup where C looks up static and global
290 struct block_symbol (*la_lookup_symbol_nonlocal
)
291 (const struct language_defn
*,
293 const struct block
*,
296 /* Find the definition of the type with the given name. */
297 struct type
*(*la_lookup_transparent_type
) (const char *);
299 /* Return demangled language symbol, or NULL. */
300 char *(*la_demangle
) (const char *mangled
, int options
);
302 /* Demangle a symbol according to this language's rules. Unlike
303 la_demangle, this does not take any options.
305 *DEMANGLED will be set by this function.
307 If this function returns 0, then *DEMANGLED must always be set
310 If this function returns 1, the implementation may set this to
311 a xmalloc'd string holding the demangled form. However, it is
312 not required to. The string, if any, is owned by the caller.
314 The resulting string should be of the form that will be
315 installed into a symbol. */
316 int (*la_sniff_from_mangled_name
) (const char *mangled
, char **demangled
);
318 /* Return class name of a mangled method name or NULL. */
319 char *(*la_class_name_from_physname
) (const char *physname
);
321 /* Table for printing expressions. */
323 const struct op_print
*la_op_print_tab
;
325 /* Zero if the language has first-class arrays. True if there are no
326 array values, and array objects decay to pointers, as in C. */
330 /* Index to use for extracting the first element of a string. */
331 char string_lower_bound
;
333 /* The list of characters forming word boundaries. */
334 const char *(*la_word_break_characters
) (void);
336 /* Add to the completion tracker all symbols which are possible
337 completions for TEXT. WORD is the entire command on which the
338 completion is being made. If CODE is TYPE_CODE_UNDEF, then all
339 symbols should be examined; otherwise, only STRUCT_DOMAIN
340 symbols whose type has a code of CODE should be matched. */
341 void (*la_collect_symbol_completion_matches
)
342 (completion_tracker
&tracker
,
343 complete_symbol_mode mode
,
344 symbol_name_match_type match_type
,
347 enum type_code code
);
349 /* The per-architecture (OS/ABI) language information. */
350 void (*la_language_arch_info
) (struct gdbarch
*,
351 struct language_arch_info
*);
353 /* Print the index of an element of an array. */
354 void (*la_print_array_index
) (struct value
*index_value
,
355 struct ui_file
*stream
,
356 const struct value_print_options
*options
);
358 /* Return non-zero if TYPE should be passed (and returned) by
359 reference at the language level. */
360 int (*la_pass_by_reference
) (struct type
*type
);
362 /* Obtain a string from the inferior, storing it in a newly allocated
363 buffer in BUFFER, which should be freed by the caller. If the
364 in- and out-parameter *LENGTH is specified at -1, the string is
365 read until a null character of the appropriate width is found -
366 otherwise the string is read to the length of characters specified.
367 On completion, *LENGTH will hold the size of the string in characters.
368 If a *LENGTH of -1 was specified it will count only actual
369 characters, excluding any eventual terminating null character.
370 Otherwise *LENGTH will include all characters - including any nulls.
371 CHARSET will hold the encoding used in the string. */
372 void (*la_get_string
) (struct value
*value
,
373 gdb::unique_xmalloc_ptr
<gdb_byte
> *buffer
,
374 int *length
, struct type
**chartype
,
375 const char **charset
);
377 /* Return an expression that can be used for a location
378 watchpoint. TYPE is a pointer type that points to the memory
379 to watch, and ADDR is the address of the watched memory. */
380 gdb::unique_xmalloc_ptr
<char> (*la_watch_location_expression
)
381 (struct type
*type
, CORE_ADDR addr
);
383 /* Return a pointer to the function that should be used to match a
384 symbol name against LOOKUP_NAME, according to this language's
385 rules. The matching algorithm depends on LOOKUP_NAME. For
386 example, on Ada, the matching algorithm depends on the symbol
387 name (wild/full/verbatim matching), and on whether we're doing
388 a normal lookup or a completion match lookup.
390 This field may be NULL, in which case
391 default_symbol_name_matcher is used to perform the
393 symbol_name_matcher_ftype
*(*la_get_symbol_name_matcher
)
394 (const lookup_name_info
&);
396 /* Find all symbols in the current program space matching NAME in
397 DOMAIN, according to this language's rules.
399 The search is done in BLOCK only.
400 The caller is responsible for iterating up through superblocks
403 For each one, call CALLBACK with the symbol. If CALLBACK
404 returns false, the iteration ends at that point.
406 This field may not be NULL. If the language does not need any
407 special processing here, 'iterate_over_symbols' should be
408 used as the definition. */
409 void (*la_iterate_over_symbols
)
410 (const struct block
*block
, const lookup_name_info
&name
,
412 gdb::function_view
<symbol_found_callback_ftype
> callback
);
414 /* Hash the given symbol search name. Use
415 default_search_name_hash if no special treatment is
417 unsigned int (*la_search_name_hash
) (const char *name
);
419 /* Various operations on varobj. */
420 const struct lang_varobj_ops
*la_varobj_ops
;
422 /* If this language allows compilation from the gdb command line,
423 this method should be non-NULL. When called it should return
424 an instance of struct gcc_context appropriate to the language.
425 When defined this method must never return NULL; instead it
426 should throw an exception on failure. The returned compiler
427 instance is owned by its caller and must be deallocated by
428 calling its 'destroy' method. */
430 compile_instance
*(*la_get_compile_instance
) (void);
432 /* This method must be defined if 'la_get_gcc_context' is defined.
433 If 'la_get_gcc_context' is not defined, then this method is
436 This takes the user-supplied text and returns a new bit of code
439 INST is the compiler instance being used.
440 INPUT is the user's input text.
441 GDBARCH is the architecture to use.
442 EXPR_BLOCK is the block in which the expression is being
444 EXPR_PC is the PC at which the expression is being parsed. */
446 std::string (*la_compute_program
) (compile_instance
*inst
,
448 struct gdbarch
*gdbarch
,
449 const struct block
*expr_block
,
452 /* Add fields above this point, so the magic number is always last. */
453 /* Magic number for compat checking. */
459 #define LANG_MAGIC 910823L
461 /* Pointer to the language_defn for our current language. This pointer
462 always points to *some* valid struct; it can be used without checking
465 The current language affects expression parsing and evaluation
466 (FIXME: it might be cleaner to make the evaluation-related stuff
467 separate exp_opcodes for each different set of semantics. We
468 should at least think this through more clearly with respect to
469 what happens if the language is changed between parsing and
470 evaluation) and printing of things like types and arrays. It does
471 *not* affect symbol-reading-- each source file in a symbol-file has
472 its own language and we should keep track of that regardless of the
473 language when symbols are read. If we want some manual setting for
474 the language of symbol files (e.g. detecting when ".c" files are
475 C++), it should be a separate setting from the current_language. */
477 extern const struct language_defn
*current_language
;
479 /* Pointer to the language_defn expected by the user, e.g. the language
480 of main(), or the language we last mentioned in a message, or C. */
482 extern const struct language_defn
*expected_language
;
485 language_mode_auto: current_language automatically set upon selection
486 of scope (e.g. stack frame)
487 language_mode_manual: current_language set only by user. */
489 extern enum language_mode
491 language_mode_auto
, language_mode_manual
495 struct type
*language_bool_type (const struct language_defn
*l
,
496 struct gdbarch
*gdbarch
);
498 struct type
*language_string_char_type (const struct language_defn
*l
,
499 struct gdbarch
*gdbarch
);
501 /* Look up type NAME in language L, and return its definition for architecture
502 GDBARCH. Returns NULL if not found. */
504 struct type
*language_lookup_primitive_type (const struct language_defn
*l
,
505 struct gdbarch
*gdbarch
,
508 /* Wrapper around language_lookup_primitive_type to return the
509 corresponding symbol. */
512 language_lookup_primitive_type_as_symbol (const struct language_defn
*l
,
513 struct gdbarch
*gdbarch
,
517 /* These macros define the behaviour of the expression
520 /* Should we range check values against the domain of their type? */
521 #define RANGE_CHECK (range_check != range_check_off)
523 /* "cast" really means conversion. */
524 /* FIXME -- should be a setting in language_defn. */
525 #define CAST_IS_CONVERSION(LANG) ((LANG)->la_language == language_c || \
526 (LANG)->la_language == language_cplus || \
527 (LANG)->la_language == language_objc)
529 extern void language_info (int);
531 extern enum language
set_language (enum language
);
534 /* This page contains functions that return things that are
535 specific to languages. Each of these functions is based on
536 the current setting of working_lang, which the user sets
537 with the "set language" command. */
539 #define LA_PRINT_TYPE(type,varstring,stream,show,level,flags) \
540 (current_language->la_print_type(type,varstring,stream,show,level,flags))
542 #define LA_PRINT_TYPEDEF(type,new_symbol,stream) \
543 (current_language->la_print_typedef(type,new_symbol,stream))
545 #define LA_VALUE_PRINT(val,stream,options) \
546 (current_language->la_value_print(val,stream,options))
548 #define LA_PRINT_CHAR(ch, type, stream) \
549 (current_language->la_printchar(ch, type, stream))
550 #define LA_PRINT_STRING(stream, elttype, string, length, encoding, force_ellipses, options) \
551 (current_language->la_printstr(stream, elttype, string, length, \
552 encoding, force_ellipses,options))
553 #define LA_EMIT_CHAR(ch, type, stream, quoter) \
554 (current_language->la_emitchar(ch, type, stream, quoter))
555 #define LA_GET_STRING(value, buffer, length, chartype, encoding) \
556 (current_language->la_get_string(value, buffer, length, chartype, encoding))
558 #define LA_PRINT_ARRAY_INDEX(index_value, stream, options) \
559 (current_language->la_print_array_index(index_value, stream, options))
561 #define LA_ITERATE_OVER_SYMBOLS(BLOCK, NAME, DOMAIN, CALLBACK) \
562 (current_language->la_iterate_over_symbols (BLOCK, NAME, DOMAIN, CALLBACK))
564 /* Test a character to decide whether it can be printed in literal form
565 or needs to be printed in another representation. For example,
566 in C the literal form of the character with octal value 141 is 'a'
567 and the "other representation" is '\141'. The "other representation"
568 is program language dependent. */
570 #define PRINT_LITERAL_FORM(c) \
572 && ((c) < 0x7F || (c) >= 0xA0) \
573 && (!sevenbit_strings || (c) < 0x80))
575 /* Type predicates */
577 extern int pointer_type (struct type
*);
581 extern void range_error (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
583 /* Data: Does this value represent "truth" to the current language? */
585 extern int value_true (struct value
*);
587 /* Misc: The string representing a particular enum language. */
589 extern enum language
language_enum (const char *str
);
591 extern const struct language_defn
*language_def (enum language
);
593 extern const char *language_str (enum language
);
595 /* Check for a language-specific trampoline. */
597 extern CORE_ADDR
skip_language_trampoline (struct frame_info
*, CORE_ADDR pc
);
599 /* Return demangled language symbol, or NULL. */
600 extern char *language_demangle (const struct language_defn
*current_language
,
601 const char *mangled
, int options
);
603 /* A wrapper for la_sniff_from_mangled_name. The arguments and result
604 are as for the method. */
606 extern int language_sniff_from_mangled_name (const struct language_defn
*lang
,
610 /* Return class name from physname, or NULL. */
611 extern char *language_class_name_from_physname (const struct language_defn
*,
612 const char *physname
);
614 /* Splitting strings into words. */
615 extern const char *default_word_break_characters (void);
617 /* Print the index of an array element using the C99 syntax. */
618 extern void default_print_array_index (struct value
*index_value
,
619 struct ui_file
*stream
,
620 const struct value_print_options
*options
);
622 /* Return non-zero if TYPE should be passed (and returned) by
623 reference at the language level. */
624 int language_pass_by_reference (struct type
*type
);
626 /* Return zero; by default, types are passed by value at the language
627 level. The target ABI may pass or return some structs by reference
628 independent of this. */
629 int default_pass_by_reference (struct type
*type
);
631 /* The default implementation of la_print_typedef. */
632 void default_print_typedef (struct type
*type
, struct symbol
*new_symbol
,
633 struct ui_file
*stream
);
635 void default_get_string (struct value
*value
,
636 gdb::unique_xmalloc_ptr
<gdb_byte
> *buffer
,
637 int *length
, struct type
**char_type
,
638 const char **charset
);
640 /* Default name hashing function. */
642 /* Produce an unsigned hash value from SEARCH_NAME that is consistent
643 with strcmp_iw, strcmp, and, at least on Ada symbols, wild_match.
644 That is, two identifiers equivalent according to any of those three
645 comparison operators hash to the same value. */
646 extern unsigned int default_search_name_hash (const char *search_name
);
648 void c_get_string (struct value
*value
,
649 gdb::unique_xmalloc_ptr
<gdb_byte
> *buffer
,
650 int *length
, struct type
**char_type
,
651 const char **charset
);
653 /* The default implementation of la_symbol_name_matcher. Matches with
655 extern bool default_symbol_name_matcher
656 (const char *symbol_search_name
,
657 const lookup_name_info
&lookup_name
,
658 completion_match_result
*comp_match_res
);
660 /* Get LANG's symbol_name_matcher method for LOOKUP_NAME. Returns
661 default_symbol_name_matcher if not set. LANG is used as a hint;
662 the function may ignore it depending on the current language and
663 LOOKUP_NAME. Specifically, if the current language is Ada, this
664 may return an Ada matcher regardless of LANG. */
665 symbol_name_matcher_ftype
*get_symbol_name_matcher
666 (const language_defn
*lang
, const lookup_name_info
&lookup_name
);
668 /* The languages supported by GDB. */
670 extern const struct language_defn auto_language_defn
;
671 extern const struct language_defn unknown_language_defn
;
672 extern const struct language_defn minimal_language_defn
;
674 extern const struct language_defn ada_language_defn
;
675 extern const struct language_defn asm_language_defn
;
676 extern const struct language_defn c_language_defn
;
677 extern const struct language_defn cplus_language_defn
;
678 extern const struct language_defn d_language_defn
;
679 extern const struct language_defn f_language_defn
;
680 extern const struct language_defn go_language_defn
;
681 extern const struct language_defn m2_language_defn
;
682 extern const struct language_defn objc_language_defn
;
683 extern const struct language_defn opencl_language_defn
;
684 extern const struct language_defn pascal_language_defn
;
685 extern const struct language_defn rust_language_defn
;
687 /* Save the current language and restore it upon destruction. */
689 class scoped_restore_current_language
693 explicit scoped_restore_current_language ()
694 : m_lang (current_language
->la_language
)
698 ~scoped_restore_current_language ()
700 set_language (m_lang
);
703 scoped_restore_current_language (const scoped_restore_current_language
&)
705 scoped_restore_current_language
&operator=
706 (const scoped_restore_current_language
&) = delete;
710 enum language m_lang
;
713 /* If language_mode is language_mode_auto,
714 then switch current language to the language of SYM
715 and restore current language upon destruction.
719 class scoped_switch_to_sym_language_if_auto
723 explicit scoped_switch_to_sym_language_if_auto (const struct symbol
*sym
)
725 if (language_mode
== language_mode_auto
)
727 m_lang
= current_language
->la_language
;
729 set_language (SYMBOL_LANGUAGE (sym
));
734 /* Assign to m_lang to silence a GCC warning. See
735 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80635. */
736 m_lang
= language_unknown
;
740 ~scoped_switch_to_sym_language_if_auto ()
743 set_language (m_lang
);
746 DISABLE_COPY_AND_ASSIGN (scoped_switch_to_sym_language_if_auto
);
750 enum language m_lang
;
753 #endif /* defined (LANGUAGE_H) */