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.
174 As we move over from the old structure based languages to a class
175 hierarchy of languages this structure will continue to contain a
176 mixture of both data and function pointers.
178 Once the class hierarchy of languages in place the first task is to
179 remove the function pointers from this structure and convert them into
180 member functions on the different language classes.
182 The current plan it to keep the constant data that describes a language
183 in this structure, and have each language pass in an instance of this
184 structure at construction time. */
188 /* Name of the language. */
192 /* Natural or official name of the language. */
194 const char *la_natural_name
;
196 /* its symtab language-enum (defs.h). */
198 enum language la_language
;
200 /* Default range checking. */
202 enum range_check la_range_check
;
204 /* Default case sensitivity. */
205 enum case_sensitivity la_case_sensitivity
;
207 /* Multi-dimensional array ordering. */
208 enum array_ordering la_array_ordering
;
210 /* Style of macro expansion, if any, supported by this language. */
211 enum macro_expansion la_macro_expansion
;
213 /* A NULL-terminated array of file extensions for this language.
214 The extension must include the ".", like ".c". If this
215 language doesn't need to provide any filename extensions, this
218 const char *const *la_filename_extensions
;
220 /* Definitions related to expression printing, prefixifying, and
223 const struct exp_descriptor
*la_exp_desc
;
225 /* Parser function. */
227 int (*la_parser
) (struct parser_state
*);
229 /* Given an expression *EXPP created by prefixifying the result of
230 la_parser, perform any remaining processing necessary to complete
231 its translation. *EXPP may change; la_post_parser is responsible
232 for releasing its previous contents, if necessary. If
233 VOID_CONTEXT_P, then no value is expected from the expression.
234 If COMPLETING is non-zero, then the expression has been parsed
235 for completion, not evaluation. */
237 void (*la_post_parser
) (expression_up
*expp
, int void_context_p
,
238 int completing
, innermost_block_tracker
*tracker
);
240 void (*la_printchar
) (int ch
, struct type
*chtype
,
241 struct ui_file
* stream
);
243 void (*la_printstr
) (struct ui_file
* stream
, struct type
*elttype
,
244 const gdb_byte
*string
, unsigned int length
,
245 const char *encoding
, int force_ellipses
,
246 const struct value_print_options
*);
248 void (*la_emitchar
) (int ch
, struct type
*chtype
,
249 struct ui_file
* stream
, int quoter
);
251 /* Print a type using syntax appropriate for this language. */
253 void (*la_print_type
) (struct type
*, const char *, struct ui_file
*, int,
254 int, const struct type_print_options
*);
256 /* Print a typedef using syntax appropriate for this language.
257 TYPE is the underlying type. NEW_SYMBOL is the symbol naming
258 the type. STREAM is the output stream on which to print. */
260 void (*la_print_typedef
) (struct type
*type
, struct symbol
*new_symbol
,
261 struct ui_file
*stream
);
263 /* Print a value using syntax appropriate for this language.
264 RECURSE is the recursion depth. It is zero-based. */
266 void (*la_value_print_inner
) (struct value
*, struct ui_file
*,
268 const struct value_print_options
*);
270 /* Print a top-level value using syntax appropriate for this language. */
272 void (*la_value_print
) (struct value
*, struct ui_file
*,
273 const struct value_print_options
*);
275 /* PC is possibly an unknown languages trampoline.
276 If that PC falls in a trampoline belonging to this language,
277 return the address of the first pc in the real function, or 0
278 if it isn't a language tramp for this language. */
279 CORE_ADDR (*skip_trampoline
) (struct frame_info
*, CORE_ADDR
);
281 /* Now come some hooks for lookup_symbol. */
283 /* If this is non-NULL, specifies the name that of the implicit
284 local variable that refers to the current object instance. */
286 const char *la_name_of_this
;
288 /* True if the symbols names should be stored in GDB's data structures
289 for minimal/partial/full symbols using their linkage (aka mangled)
290 form; false if the symbol names should be demangled first.
292 Most languages implement symbol lookup by comparing the demangled
293 names, in which case it is advantageous to store that information
294 already demangled, and so would set this field to false.
296 On the other hand, some languages have opted for doing symbol
297 lookups by comparing mangled names instead, for reasons usually
298 specific to the language. Those languages should set this field
301 And finally, other languages such as C or Asm do not have
302 the concept of mangled vs demangled name, so those languages
303 should set this field to true as well, to prevent any accidental
304 demangling through an unrelated language's demangler. */
306 const bool la_store_sym_names_in_linkage_form_p
;
308 /* This is a function that lookup_symbol will call when it gets to
309 the part of symbol lookup where C looks up static and global
312 struct block_symbol (*la_lookup_symbol_nonlocal
)
313 (const struct language_defn
*,
315 const struct block
*,
318 /* Return demangled language symbol, or NULL. */
319 char *(*la_demangle
) (const char *mangled
, int options
);
321 /* Demangle a symbol according to this language's rules. Unlike
322 la_demangle, this does not take any options.
324 *DEMANGLED will be set by this function.
326 If this function returns 0, then *DEMANGLED must always be set
329 If this function returns 1, the implementation may set this to
330 a xmalloc'd string holding the demangled form. However, it is
331 not required to. The string, if any, is owned by the caller.
333 The resulting string should be of the form that will be
334 installed into a symbol. */
335 int (*la_sniff_from_mangled_name
) (const char *mangled
, char **demangled
);
337 /* Return class name of a mangled method name or NULL. */
338 char *(*la_class_name_from_physname
) (const char *physname
);
340 /* Table for printing expressions. */
342 const struct op_print
*la_op_print_tab
;
344 /* Zero if the language has first-class arrays. True if there are no
345 array values, and array objects decay to pointers, as in C. */
349 /* Index to use for extracting the first element of a string. */
350 char string_lower_bound
;
352 /* The list of characters forming word boundaries. */
353 const char *(*la_word_break_characters
) (void);
355 /* Add to the completion tracker all symbols which are possible
356 completions for TEXT. WORD is the entire command on which the
357 completion is being made. If CODE is TYPE_CODE_UNDEF, then all
358 symbols should be examined; otherwise, only STRUCT_DOMAIN
359 symbols whose type has a code of CODE should be matched. */
360 void (*la_collect_symbol_completion_matches
)
361 (completion_tracker
&tracker
,
362 complete_symbol_mode mode
,
363 symbol_name_match_type match_type
,
366 enum type_code code
);
368 /* Return an expression that can be used for a location
369 watchpoint. TYPE is a pointer type that points to the memory
370 to watch, and ADDR is the address of the watched memory. */
371 gdb::unique_xmalloc_ptr
<char> (*la_watch_location_expression
)
372 (struct type
*type
, CORE_ADDR addr
);
374 /* Return a pointer to the function that should be used to match a
375 symbol name against LOOKUP_NAME, according to this language's
376 rules. The matching algorithm depends on LOOKUP_NAME. For
377 example, on Ada, the matching algorithm depends on the symbol
378 name (wild/full/verbatim matching), and on whether we're doing
379 a normal lookup or a completion match lookup.
381 This field may be NULL, in which case
382 default_symbol_name_matcher is used to perform the
384 symbol_name_matcher_ftype
*(*la_get_symbol_name_matcher
)
385 (const lookup_name_info
&);
387 /* Hash the given symbol search name. Use
388 default_search_name_hash if no special treatment is
390 unsigned int (*la_search_name_hash
) (const char *name
);
392 /* Various operations on varobj. */
393 const struct lang_varobj_ops
*la_varobj_ops
;
395 /* This method must be defined if 'la_get_gcc_context' is defined.
396 If 'la_get_gcc_context' is not defined, then this method is
399 This takes the user-supplied text and returns a new bit of code
402 INST is the compiler instance being used.
403 INPUT is the user's input text.
404 GDBARCH is the architecture to use.
405 EXPR_BLOCK is the block in which the expression is being
407 EXPR_PC is the PC at which the expression is being parsed. */
409 std::string (*la_compute_program
) (compile_instance
*inst
,
411 struct gdbarch
*gdbarch
,
412 const struct block
*expr_block
,
415 /* Return true if TYPE is a string type. */
416 bool (*la_is_string_type_p
) (struct type
*type
);
418 /* This string is used by the 'set print max-depth' setting. When GDB
419 replaces a struct or union (during value printing) that is "too
420 deep" this string is displayed instead. */
421 const char *la_struct_too_deep_ellipsis
;
425 /* Base class from which all other language classes derive. */
427 struct language_defn
: language_data
429 language_defn (enum language lang
, const language_data
&init_data
)
430 : language_data (init_data
)
432 /* We should only ever create one instance of each language. */
433 gdb_assert (languages
[lang
] == nullptr);
434 languages
[lang
] = this;
437 /* Print the index of an element of an array. This default
438 implementation prints using C99 syntax. */
440 virtual void print_array_index (struct type
*index_type
,
442 struct ui_file
*stream
,
443 const value_print_options
*options
) const;
445 /* Given a symbol VAR, the corresponding block VAR_BLOCK (if any) and a
446 stack frame id FRAME, read the value of the variable and return (pointer
447 to a) struct value containing the value.
449 VAR_BLOCK is needed if there's a possibility for VAR to be outside
450 FRAME. This is what happens if FRAME correspond to a nested function
451 and VAR is defined in the outer function. If callers know that VAR is
452 located in FRAME or is global/static, NULL can be passed as VAR_BLOCK.
454 Throw an error if the variable cannot be found. */
456 virtual struct value
*read_var_value (struct symbol
*var
,
457 const struct block
*var_block
,
458 struct frame_info
*frame
) const;
460 /* Return information about whether TYPE should be passed
461 (and returned) by reference at the language level. The default
462 implementation returns a LANGUAGE_PASS_BY_REF_INFO initialised in its
465 virtual struct language_pass_by_ref_info pass_by_reference_info
466 (struct type
*type
) const
471 /* The per-architecture (OS/ABI) language information. */
473 virtual void language_arch_info (struct gdbarch
*,
474 struct language_arch_info
*) const = 0;
476 /* Find the definition of the type with the given name. */
478 virtual struct type
*lookup_transparent_type (const char *name
) const
480 return basic_lookup_transparent_type (name
);
483 /* Find all symbols in the current program space matching NAME in
484 DOMAIN, according to this language's rules.
486 The search is done in BLOCK only.
487 The caller is responsible for iterating up through superblocks
490 For each one, call CALLBACK with the symbol. If CALLBACK
491 returns false, the iteration ends at that point.
493 This field may not be NULL. If the language does not need any
494 special processing here, 'iterate_over_symbols' should be
495 used as the definition. */
496 virtual bool iterate_over_symbols
497 (const struct block
*block
, const lookup_name_info
&name
,
499 gdb::function_view
<symbol_found_callback_ftype
> callback
) const
501 return ::iterate_over_symbols (block
, name
, domain
, callback
);
504 /* If this language allows compilation from the gdb command line, then
505 this method will return an instance of struct gcc_context appropriate
506 to the language. If compilation for this language is generally
507 supported, but something goes wrong then an exception is thrown. The
508 returned compiler instance is owned by its caller and must be
509 deallocated by the caller. If compilation is not supported for this
510 language then this method returns NULL. */
512 virtual compile_instance
*get_compile_instance () const
517 /* List of all known languages. */
518 static const struct language_defn
*languages
[nr_languages
];
521 /* Pointer to the language_defn for our current language. This pointer
522 always points to *some* valid struct; it can be used without checking
525 The current language affects expression parsing and evaluation
526 (FIXME: it might be cleaner to make the evaluation-related stuff
527 separate exp_opcodes for each different set of semantics. We
528 should at least think this through more clearly with respect to
529 what happens if the language is changed between parsing and
530 evaluation) and printing of things like types and arrays. It does
531 *not* affect symbol-reading-- each source file in a symbol-file has
532 its own language and we should keep track of that regardless of the
533 language when symbols are read. If we want some manual setting for
534 the language of symbol files (e.g. detecting when ".c" files are
535 C++), it should be a separate setting from the current_language. */
537 extern const struct language_defn
*current_language
;
539 /* Pointer to the language_defn expected by the user, e.g. the language
540 of main(), or the language we last mentioned in a message, or C. */
542 extern const struct language_defn
*expected_language
;
544 /* Warning issued when current_language and the language of the current
545 frame do not match. */
547 extern const char lang_frame_mismatch_warn
[];
550 language_mode_auto: current_language automatically set upon selection
551 of scope (e.g. stack frame)
552 language_mode_manual: current_language set only by user. */
554 extern enum language_mode
556 language_mode_auto
, language_mode_manual
560 struct type
*language_bool_type (const struct language_defn
*l
,
561 struct gdbarch
*gdbarch
);
563 struct type
*language_string_char_type (const struct language_defn
*l
,
564 struct gdbarch
*gdbarch
);
566 /* Look up type NAME in language L, and return its definition for architecture
567 GDBARCH. Returns NULL if not found. */
569 struct type
*language_lookup_primitive_type (const struct language_defn
*l
,
570 struct gdbarch
*gdbarch
,
573 /* Wrapper around language_lookup_primitive_type to return the
574 corresponding symbol. */
577 language_lookup_primitive_type_as_symbol (const struct language_defn
*l
,
578 struct gdbarch
*gdbarch
,
582 /* These macros define the behaviour of the expression
585 /* Should we range check values against the domain of their type? */
586 #define RANGE_CHECK (range_check != range_check_off)
588 /* "cast" really means conversion. */
589 /* FIXME -- should be a setting in language_defn. */
590 #define CAST_IS_CONVERSION(LANG) ((LANG)->la_language == language_c || \
591 (LANG)->la_language == language_cplus || \
592 (LANG)->la_language == language_objc)
594 extern void language_info (int);
596 extern enum language
set_language (enum language
);
599 /* This page contains functions that return things that are
600 specific to languages. Each of these functions is based on
601 the current setting of working_lang, which the user sets
602 with the "set language" command. */
604 #define LA_PRINT_TYPE(type,varstring,stream,show,level,flags) \
605 (current_language->la_print_type(type,varstring,stream,show,level,flags))
607 #define LA_PRINT_TYPEDEF(type,new_symbol,stream) \
608 (current_language->la_print_typedef(type,new_symbol,stream))
610 #define LA_VALUE_PRINT(val,stream,options) \
611 (current_language->la_value_print(val,stream,options))
613 #define LA_PRINT_CHAR(ch, type, stream) \
614 (current_language->la_printchar(ch, type, stream))
615 #define LA_PRINT_STRING(stream, elttype, string, length, encoding, force_ellipses, options) \
616 (current_language->la_printstr(stream, elttype, string, length, \
617 encoding, force_ellipses,options))
618 #define LA_EMIT_CHAR(ch, type, stream, quoter) \
619 (current_language->la_emitchar(ch, type, stream, quoter))
621 #define LA_PRINT_ARRAY_INDEX(index_type, index_value, stream, options) \
622 (current_language->print_array_index(index_type, index_value, stream, \
625 #define LA_ITERATE_OVER_SYMBOLS(BLOCK, NAME, DOMAIN, CALLBACK) \
626 (current_language->iterate_over_symbols (BLOCK, NAME, DOMAIN, CALLBACK))
628 /* Test a character to decide whether it can be printed in literal form
629 or needs to be printed in another representation. For example,
630 in C the literal form of the character with octal value 141 is 'a'
631 and the "other representation" is '\141'. The "other representation"
632 is program language dependent. */
634 #define PRINT_LITERAL_FORM(c) \
636 && ((c) < 0x7F || (c) >= 0xA0) \
637 && (!sevenbit_strings || (c) < 0x80))
639 /* Type predicates */
641 extern int pointer_type (struct type
*);
643 /* Return true if TYPE is a string type, otherwise return false. This
644 default implementation only detects TYPE_CODE_STRING. */
645 extern bool default_is_string_type_p (struct type
*type
);
649 extern void range_error (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
651 /* Data: Does this value represent "truth" to the current language? */
653 extern int value_true (struct value
*);
655 /* Misc: The string representing a particular enum language. */
657 extern enum language
language_enum (const char *str
);
659 extern const struct language_defn
*language_def (enum language
);
661 extern const char *language_str (enum language
);
663 /* Check for a language-specific trampoline. */
665 extern CORE_ADDR
skip_language_trampoline (struct frame_info
*, CORE_ADDR pc
);
667 /* Return demangled language symbol, or NULL. */
668 extern char *language_demangle (const struct language_defn
*current_language
,
669 const char *mangled
, int options
);
671 /* A wrapper for la_sniff_from_mangled_name. The arguments and result
672 are as for the method. */
674 extern int language_sniff_from_mangled_name (const struct language_defn
*lang
,
678 /* Return class name from physname, or NULL. */
679 extern char *language_class_name_from_physname (const struct language_defn
*,
680 const char *physname
);
682 /* Splitting strings into words. */
683 extern const char *default_word_break_characters (void);
685 /* Return information about whether TYPE should be passed
686 (and returned) by reference at the language level. */
687 struct language_pass_by_ref_info
language_pass_by_reference (struct type
*type
);
689 /* The default implementation of la_print_typedef. */
690 void default_print_typedef (struct type
*type
, struct symbol
*new_symbol
,
691 struct ui_file
*stream
);
693 /* Default name hashing function. */
695 /* Produce an unsigned hash value from SEARCH_NAME that is consistent
696 with strcmp_iw, strcmp, and, at least on Ada symbols, wild_match.
697 That is, two identifiers equivalent according to any of those three
698 comparison operators hash to the same value. */
699 extern unsigned int default_search_name_hash (const char *search_name
);
701 void c_get_string (struct value
*value
,
702 gdb::unique_xmalloc_ptr
<gdb_byte
> *buffer
,
703 int *length
, struct type
**char_type
,
704 const char **charset
);
706 /* The default implementation of la_symbol_name_matcher. Matches with
708 extern bool default_symbol_name_matcher
709 (const char *symbol_search_name
,
710 const lookup_name_info
&lookup_name
,
711 completion_match_result
*comp_match_res
);
713 /* Get LANG's symbol_name_matcher method for LOOKUP_NAME. Returns
714 default_symbol_name_matcher if not set. LANG is used as a hint;
715 the function may ignore it depending on the current language and
716 LOOKUP_NAME. Specifically, if the current language is Ada, this
717 may return an Ada matcher regardless of LANG. */
718 symbol_name_matcher_ftype
*get_symbol_name_matcher
719 (const language_defn
*lang
, const lookup_name_info
&lookup_name
);
721 /* Save the current language and restore it upon destruction. */
723 class scoped_restore_current_language
727 explicit scoped_restore_current_language ()
728 : m_lang (current_language
->la_language
)
732 ~scoped_restore_current_language ()
734 set_language (m_lang
);
737 scoped_restore_current_language (const scoped_restore_current_language
&)
739 scoped_restore_current_language
&operator=
740 (const scoped_restore_current_language
&) = delete;
744 enum language m_lang
;
747 /* If language_mode is language_mode_auto,
748 then switch current language to the language of SYM
749 and restore current language upon destruction.
753 class scoped_switch_to_sym_language_if_auto
757 explicit scoped_switch_to_sym_language_if_auto (const struct symbol
*sym
)
759 if (language_mode
== language_mode_auto
)
761 m_lang
= current_language
->la_language
;
763 set_language (sym
->language ());
768 /* Assign to m_lang to silence a GCC warning. See
769 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80635. */
770 m_lang
= language_unknown
;
774 ~scoped_switch_to_sym_language_if_auto ()
777 set_language (m_lang
);
780 DISABLE_COPY_AND_ASSIGN (scoped_switch_to_sym_language_if_auto
);
784 enum language m_lang
;
787 #endif /* defined (LANGUAGE_H) */