1 /* Source-language-related definitions for GDB.
3 Copyright (C) 1991-2020 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 /* In a language (particularly C++) a function argument of an aggregate
132 type (i.e. class/struct/union) may be implicitly passed by reference
133 even though it is declared a call-by-value argument in the source.
134 The struct below puts together necessary information for GDB to be
135 able to detect and carry out pass-by-reference semantics for a
136 particular type. This type is referred as T in the inlined comments
139 The default values of the fields are chosen to give correct semantics
140 for primitive types and for simple aggregate types, such as
146 struct language_pass_by_ref_info
148 /* True if an argument of type T can be passed to a function by value
149 (i.e. not through an implicit reference). False, otherwise. */
150 bool trivially_copyable
= true;
152 /* True if a copy of a value of type T can be initialized by
153 memcpy'ing the value bit-by-bit. False, otherwise.
154 E.g. If T has a user-defined copy ctor, this should be false. */
155 bool trivially_copy_constructible
= true;
157 /* True if a value of type T can be destructed simply by reclaiming
158 the memory area occupied by the value. False, otherwise.
159 E.g. If T has a user-defined destructor, this should be false. */
160 bool trivially_destructible
= true;
162 /* True if it is allowed to create a copy of a value of type T.
164 E.g. If T has a deleted copy ctor, this should be false. */
165 bool copy_constructible
= true;
167 /* True if a value of type T can be destructed. False, otherwise.
168 E.g. If T has a deleted destructor, this should be false. */
169 bool destructible
= true;
172 /* Structure tying together assorted information about a language. */
176 /* Name of the language. */
180 /* Natural or official name of the language. */
182 const char *la_natural_name
;
184 /* its symtab language-enum (defs.h). */
186 enum language la_language
;
188 /* Default range checking. */
190 enum range_check la_range_check
;
192 /* Default case sensitivity. */
193 enum case_sensitivity la_case_sensitivity
;
195 /* Multi-dimensional array ordering. */
196 enum array_ordering la_array_ordering
;
198 /* Style of macro expansion, if any, supported by this language. */
199 enum macro_expansion la_macro_expansion
;
201 /* A NULL-terminated array of file extensions for this language.
202 The extension must include the ".", like ".c". If this
203 language doesn't need to provide any filename extensions, this
206 const char *const *la_filename_extensions
;
208 /* Definitions related to expression printing, prefixifying, and
211 const struct exp_descriptor
*la_exp_desc
;
213 /* Parser function. */
215 int (*la_parser
) (struct parser_state
*);
217 /* Given an expression *EXPP created by prefixifying the result of
218 la_parser, perform any remaining processing necessary to complete
219 its translation. *EXPP may change; la_post_parser is responsible
220 for releasing its previous contents, if necessary. If
221 VOID_CONTEXT_P, then no value is expected from the expression.
222 If COMPLETING is non-zero, then the expression has been parsed
223 for completion, not evaluation. */
225 void (*la_post_parser
) (expression_up
*expp
, int void_context_p
,
226 int completing
, innermost_block_tracker
*tracker
);
228 void (*la_printchar
) (int ch
, struct type
*chtype
,
229 struct ui_file
* stream
);
231 void (*la_printstr
) (struct ui_file
* stream
, struct type
*elttype
,
232 const gdb_byte
*string
, unsigned int length
,
233 const char *encoding
, int force_ellipses
,
234 const struct value_print_options
*);
236 void (*la_emitchar
) (int ch
, struct type
*chtype
,
237 struct ui_file
* stream
, int quoter
);
239 /* Print a type using syntax appropriate for this language. */
241 void (*la_print_type
) (struct type
*, const char *, struct ui_file
*, int,
242 int, const struct type_print_options
*);
244 /* Print a typedef using syntax appropriate for this language.
245 TYPE is the underlying type. NEW_SYMBOL is the symbol naming
246 the type. STREAM is the output stream on which to print. */
248 void (*la_print_typedef
) (struct type
*type
, struct symbol
*new_symbol
,
249 struct ui_file
*stream
);
251 /* Print a value using syntax appropriate for this language.
253 TYPE is the type of the sub-object to be printed.
255 EMBEDDED_OFFSET is the offset into the outermost object of the
256 sub-object represented by TYPE. This is the object which this
257 call should print. Note that the enclosing type is not
260 ADDRESS is the address in the inferior of the enclosing object.
262 STREAM is the stream on which the value is to be printed.
264 RECURSE is the recursion depth. It is zero-based.
266 OPTIONS are the formatting options to be used when
269 void (*la_val_print
) (struct type
*type
,
270 int embedded_offset
, CORE_ADDR address
,
271 struct ui_file
*stream
, int recurse
,
273 const struct value_print_options
*options
);
275 /* Print a top-level value using syntax appropriate for this language. */
277 void (*la_value_print
) (struct value
*, struct ui_file
*,
278 const struct value_print_options
*);
280 /* Given a symbol VAR, the corresponding block VAR_BLOCK (if any) and a
281 stack frame id FRAME, read the value of the variable and return (pointer
282 to a) struct value containing the value.
284 VAR_BLOCK is needed if there's a possibility for VAR to be outside
285 FRAME. This is what happens if FRAME correspond to a nested function
286 and VAR is defined in the outer function. If callers know that VAR is
287 located in FRAME or is global/static, NULL can be passed as VAR_BLOCK.
289 Throw an error if the variable cannot be found. */
291 struct value
*(*la_read_var_value
) (struct symbol
*var
,
292 const struct block
*var_block
,
293 struct frame_info
*frame
);
295 /* PC is possibly an unknown languages trampoline.
296 If that PC falls in a trampoline belonging to this language,
297 return the address of the first pc in the real function, or 0
298 if it isn't a language tramp for this language. */
299 CORE_ADDR (*skip_trampoline
) (struct frame_info
*, CORE_ADDR
);
301 /* Now come some hooks for lookup_symbol. */
303 /* If this is non-NULL, specifies the name that of the implicit
304 local variable that refers to the current object instance. */
306 const char *la_name_of_this
;
308 /* True if the symbols names should be stored in GDB's data structures
309 for minimal/partial/full symbols using their linkage (aka mangled)
310 form; false if the symbol names should be demangled first.
312 Most languages implement symbol lookup by comparing the demangled
313 names, in which case it is advantageous to store that information
314 already demangled, and so would set this field to false.
316 On the other hand, some languages have opted for doing symbol
317 lookups by comparing mangled names instead, for reasons usually
318 specific to the language. Those languages should set this field
321 And finally, other languages such as C or Asm do not have
322 the concept of mangled vs demangled name, so those languages
323 should set this field to true as well, to prevent any accidental
324 demangling through an unrelated language's demangler. */
326 const bool la_store_sym_names_in_linkage_form_p
;
328 /* This is a function that lookup_symbol will call when it gets to
329 the part of symbol lookup where C looks up static and global
332 struct block_symbol (*la_lookup_symbol_nonlocal
)
333 (const struct language_defn
*,
335 const struct block
*,
338 /* Find the definition of the type with the given name. */
339 struct type
*(*la_lookup_transparent_type
) (const char *);
341 /* Return demangled language symbol, or NULL. */
342 char *(*la_demangle
) (const char *mangled
, int options
);
344 /* Demangle a symbol according to this language's rules. Unlike
345 la_demangle, this does not take any options.
347 *DEMANGLED will be set by this function.
349 If this function returns 0, then *DEMANGLED must always be set
352 If this function returns 1, the implementation may set this to
353 a xmalloc'd string holding the demangled form. However, it is
354 not required to. The string, if any, is owned by the caller.
356 The resulting string should be of the form that will be
357 installed into a symbol. */
358 int (*la_sniff_from_mangled_name
) (const char *mangled
, char **demangled
);
360 /* Return class name of a mangled method name or NULL. */
361 char *(*la_class_name_from_physname
) (const char *physname
);
363 /* Table for printing expressions. */
365 const struct op_print
*la_op_print_tab
;
367 /* Zero if the language has first-class arrays. True if there are no
368 array values, and array objects decay to pointers, as in C. */
372 /* Index to use for extracting the first element of a string. */
373 char string_lower_bound
;
375 /* The list of characters forming word boundaries. */
376 const char *(*la_word_break_characters
) (void);
378 /* Add to the completion tracker all symbols which are possible
379 completions for TEXT. WORD is the entire command on which the
380 completion is being made. If CODE is TYPE_CODE_UNDEF, then all
381 symbols should be examined; otherwise, only STRUCT_DOMAIN
382 symbols whose type has a code of CODE should be matched. */
383 void (*la_collect_symbol_completion_matches
)
384 (completion_tracker
&tracker
,
385 complete_symbol_mode mode
,
386 symbol_name_match_type match_type
,
389 enum type_code code
);
391 /* The per-architecture (OS/ABI) language information. */
392 void (*la_language_arch_info
) (struct gdbarch
*,
393 struct language_arch_info
*);
395 /* Print the index of an element of an array. */
396 void (*la_print_array_index
) (struct value
*index_value
,
397 struct ui_file
*stream
,
398 const struct value_print_options
*options
);
400 /* Return information about whether TYPE should be passed
401 (and returned) by reference at the language level. */
402 struct language_pass_by_ref_info (*la_pass_by_reference
)
405 /* Return an expression that can be used for a location
406 watchpoint. TYPE is a pointer type that points to the memory
407 to watch, and ADDR is the address of the watched memory. */
408 gdb::unique_xmalloc_ptr
<char> (*la_watch_location_expression
)
409 (struct type
*type
, CORE_ADDR addr
);
411 /* Return a pointer to the function that should be used to match a
412 symbol name against LOOKUP_NAME, according to this language's
413 rules. The matching algorithm depends on LOOKUP_NAME. For
414 example, on Ada, the matching algorithm depends on the symbol
415 name (wild/full/verbatim matching), and on whether we're doing
416 a normal lookup or a completion match lookup.
418 This field may be NULL, in which case
419 default_symbol_name_matcher is used to perform the
421 symbol_name_matcher_ftype
*(*la_get_symbol_name_matcher
)
422 (const lookup_name_info
&);
424 /* Find all symbols in the current program space matching NAME in
425 DOMAIN, according to this language's rules.
427 The search is done in BLOCK only.
428 The caller is responsible for iterating up through superblocks
431 For each one, call CALLBACK with the symbol. If CALLBACK
432 returns false, the iteration ends at that point.
434 This field may not be NULL. If the language does not need any
435 special processing here, 'iterate_over_symbols' should be
436 used as the definition. */
437 bool (*la_iterate_over_symbols
)
438 (const struct block
*block
, const lookup_name_info
&name
,
440 gdb::function_view
<symbol_found_callback_ftype
> callback
);
442 /* Hash the given symbol search name. Use
443 default_search_name_hash if no special treatment is
445 unsigned int (*la_search_name_hash
) (const char *name
);
447 /* Various operations on varobj. */
448 const struct lang_varobj_ops
*la_varobj_ops
;
450 /* If this language allows compilation from the gdb command line,
451 this method should be non-NULL. When called it should return
452 an instance of struct gcc_context appropriate to the language.
453 When defined this method must never return NULL; instead it
454 should throw an exception on failure. The returned compiler
455 instance is owned by its caller and must be deallocated by
456 calling its 'destroy' method. */
458 compile_instance
*(*la_get_compile_instance
) (void);
460 /* This method must be defined if 'la_get_gcc_context' is defined.
461 If 'la_get_gcc_context' is not defined, then this method is
464 This takes the user-supplied text and returns a new bit of code
467 INST is the compiler instance being used.
468 INPUT is the user's input text.
469 GDBARCH is the architecture to use.
470 EXPR_BLOCK is the block in which the expression is being
472 EXPR_PC is the PC at which the expression is being parsed. */
474 std::string (*la_compute_program
) (compile_instance
*inst
,
476 struct gdbarch
*gdbarch
,
477 const struct block
*expr_block
,
480 /* Return true if TYPE is a string type. */
481 bool (*la_is_string_type_p
) (struct type
*type
);
483 /* This string is used by the 'set print max-depth' setting. When GDB
484 replaces a struct or union (during value printing) that is "too
485 deep" this string is displayed instead. */
486 const char *la_struct_too_deep_ellipsis
;
490 /* Pointer to the language_defn for our current language. This pointer
491 always points to *some* valid struct; it can be used without checking
494 The current language affects expression parsing and evaluation
495 (FIXME: it might be cleaner to make the evaluation-related stuff
496 separate exp_opcodes for each different set of semantics. We
497 should at least think this through more clearly with respect to
498 what happens if the language is changed between parsing and
499 evaluation) and printing of things like types and arrays. It does
500 *not* affect symbol-reading-- each source file in a symbol-file has
501 its own language and we should keep track of that regardless of the
502 language when symbols are read. If we want some manual setting for
503 the language of symbol files (e.g. detecting when ".c" files are
504 C++), it should be a separate setting from the current_language. */
506 extern const struct language_defn
*current_language
;
508 /* Pointer to the language_defn expected by the user, e.g. the language
509 of main(), or the language we last mentioned in a message, or C. */
511 extern const struct language_defn
*expected_language
;
513 /* Warning issued when current_language and the language of the current
514 frame do not match. */
516 extern const char lang_frame_mismatch_warn
[];
519 language_mode_auto: current_language automatically set upon selection
520 of scope (e.g. stack frame)
521 language_mode_manual: current_language set only by user. */
523 extern enum language_mode
525 language_mode_auto
, language_mode_manual
529 struct type
*language_bool_type (const struct language_defn
*l
,
530 struct gdbarch
*gdbarch
);
532 struct type
*language_string_char_type (const struct language_defn
*l
,
533 struct gdbarch
*gdbarch
);
535 /* Look up type NAME in language L, and return its definition for architecture
536 GDBARCH. Returns NULL if not found. */
538 struct type
*language_lookup_primitive_type (const struct language_defn
*l
,
539 struct gdbarch
*gdbarch
,
542 /* Wrapper around language_lookup_primitive_type to return the
543 corresponding symbol. */
546 language_lookup_primitive_type_as_symbol (const struct language_defn
*l
,
547 struct gdbarch
*gdbarch
,
551 /* These macros define the behaviour of the expression
554 /* Should we range check values against the domain of their type? */
555 #define RANGE_CHECK (range_check != range_check_off)
557 /* "cast" really means conversion. */
558 /* FIXME -- should be a setting in language_defn. */
559 #define CAST_IS_CONVERSION(LANG) ((LANG)->la_language == language_c || \
560 (LANG)->la_language == language_cplus || \
561 (LANG)->la_language == language_objc)
563 extern void language_info (int);
565 extern enum language
set_language (enum language
);
568 /* This page contains functions that return things that are
569 specific to languages. Each of these functions is based on
570 the current setting of working_lang, which the user sets
571 with the "set language" command. */
573 #define LA_PRINT_TYPE(type,varstring,stream,show,level,flags) \
574 (current_language->la_print_type(type,varstring,stream,show,level,flags))
576 #define LA_PRINT_TYPEDEF(type,new_symbol,stream) \
577 (current_language->la_print_typedef(type,new_symbol,stream))
579 #define LA_VALUE_PRINT(val,stream,options) \
580 (current_language->la_value_print(val,stream,options))
582 #define LA_PRINT_CHAR(ch, type, stream) \
583 (current_language->la_printchar(ch, type, stream))
584 #define LA_PRINT_STRING(stream, elttype, string, length, encoding, force_ellipses, options) \
585 (current_language->la_printstr(stream, elttype, string, length, \
586 encoding, force_ellipses,options))
587 #define LA_EMIT_CHAR(ch, type, stream, quoter) \
588 (current_language->la_emitchar(ch, type, stream, quoter))
590 #define LA_PRINT_ARRAY_INDEX(index_value, stream, options) \
591 (current_language->la_print_array_index(index_value, stream, options))
593 #define LA_ITERATE_OVER_SYMBOLS(BLOCK, NAME, DOMAIN, CALLBACK) \
594 (current_language->la_iterate_over_symbols (BLOCK, NAME, DOMAIN, CALLBACK))
596 /* Test a character to decide whether it can be printed in literal form
597 or needs to be printed in another representation. For example,
598 in C the literal form of the character with octal value 141 is 'a'
599 and the "other representation" is '\141'. The "other representation"
600 is program language dependent. */
602 #define PRINT_LITERAL_FORM(c) \
604 && ((c) < 0x7F || (c) >= 0xA0) \
605 && (!sevenbit_strings || (c) < 0x80))
607 /* Type predicates */
609 extern int pointer_type (struct type
*);
611 /* Return true if TYPE is a string type, otherwise return false. This
612 default implementation only detects TYPE_CODE_STRING. */
613 extern bool default_is_string_type_p (struct type
*type
);
617 extern void range_error (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
619 /* Data: Does this value represent "truth" to the current language? */
621 extern int value_true (struct value
*);
623 /* Misc: The string representing a particular enum language. */
625 extern enum language
language_enum (const char *str
);
627 extern const struct language_defn
*language_def (enum language
);
629 extern const char *language_str (enum language
);
631 /* Check for a language-specific trampoline. */
633 extern CORE_ADDR
skip_language_trampoline (struct frame_info
*, CORE_ADDR pc
);
635 /* Return demangled language symbol, or NULL. */
636 extern char *language_demangle (const struct language_defn
*current_language
,
637 const char *mangled
, int options
);
639 /* A wrapper for la_sniff_from_mangled_name. The arguments and result
640 are as for the method. */
642 extern int language_sniff_from_mangled_name (const struct language_defn
*lang
,
646 /* Return class name from physname, or NULL. */
647 extern char *language_class_name_from_physname (const struct language_defn
*,
648 const char *physname
);
650 /* Splitting strings into words. */
651 extern const char *default_word_break_characters (void);
653 /* Print the index of an array element using the C99 syntax. */
654 extern void default_print_array_index (struct value
*index_value
,
655 struct ui_file
*stream
,
656 const struct value_print_options
*options
);
658 /* Return information about whether TYPE should be passed
659 (and returned) by reference at the language level. */
660 struct language_pass_by_ref_info
language_pass_by_reference (struct type
*type
);
662 /* Return a default struct that provides pass-by-reference information
663 about the given TYPE. Languages should update the default values
665 struct language_pass_by_ref_info
default_pass_by_reference (struct type
*type
);
667 /* The default implementation of la_print_typedef. */
668 void default_print_typedef (struct type
*type
, struct symbol
*new_symbol
,
669 struct ui_file
*stream
);
671 /* Default name hashing function. */
673 /* Produce an unsigned hash value from SEARCH_NAME that is consistent
674 with strcmp_iw, strcmp, and, at least on Ada symbols, wild_match.
675 That is, two identifiers equivalent according to any of those three
676 comparison operators hash to the same value. */
677 extern unsigned int default_search_name_hash (const char *search_name
);
679 void c_get_string (struct value
*value
,
680 gdb::unique_xmalloc_ptr
<gdb_byte
> *buffer
,
681 int *length
, struct type
**char_type
,
682 const char **charset
);
684 /* The default implementation of la_symbol_name_matcher. Matches with
686 extern bool default_symbol_name_matcher
687 (const char *symbol_search_name
,
688 const lookup_name_info
&lookup_name
,
689 completion_match_result
*comp_match_res
);
691 /* Get LANG's symbol_name_matcher method for LOOKUP_NAME. Returns
692 default_symbol_name_matcher if not set. LANG is used as a hint;
693 the function may ignore it depending on the current language and
694 LOOKUP_NAME. Specifically, if the current language is Ada, this
695 may return an Ada matcher regardless of LANG. */
696 symbol_name_matcher_ftype
*get_symbol_name_matcher
697 (const language_defn
*lang
, const lookup_name_info
&lookup_name
);
699 /* The languages supported by GDB. */
701 extern const struct language_defn auto_language_defn
;
702 extern const struct language_defn unknown_language_defn
;
703 extern const struct language_defn minimal_language_defn
;
705 extern const struct language_defn ada_language_defn
;
706 extern const struct language_defn asm_language_defn
;
707 extern const struct language_defn c_language_defn
;
708 extern const struct language_defn cplus_language_defn
;
709 extern const struct language_defn d_language_defn
;
710 extern const struct language_defn f_language_defn
;
711 extern const struct language_defn go_language_defn
;
712 extern const struct language_defn m2_language_defn
;
713 extern const struct language_defn objc_language_defn
;
714 extern const struct language_defn opencl_language_defn
;
715 extern const struct language_defn pascal_language_defn
;
716 extern const struct language_defn rust_language_defn
;
718 /* Save the current language and restore it upon destruction. */
720 class scoped_restore_current_language
724 explicit scoped_restore_current_language ()
725 : m_lang (current_language
->la_language
)
729 ~scoped_restore_current_language ()
731 set_language (m_lang
);
734 scoped_restore_current_language (const scoped_restore_current_language
&)
736 scoped_restore_current_language
&operator=
737 (const scoped_restore_current_language
&) = delete;
741 enum language m_lang
;
744 /* If language_mode is language_mode_auto,
745 then switch current language to the language of SYM
746 and restore current language upon destruction.
750 class scoped_switch_to_sym_language_if_auto
754 explicit scoped_switch_to_sym_language_if_auto (const struct symbol
*sym
)
756 if (language_mode
== language_mode_auto
)
758 m_lang
= current_language
->la_language
;
760 set_language (sym
->language ());
765 /* Assign to m_lang to silence a GCC warning. See
766 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80635. */
767 m_lang
= language_unknown
;
771 ~scoped_switch_to_sym_language_if_auto ()
774 set_language (m_lang
);
777 DISABLE_COPY_AND_ASSIGN (scoped_switch_to_sym_language_if_auto
);
781 enum language m_lang
;
784 #endif /* defined (LANGUAGE_H) */