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 /* Given a symbol VAR, the corresponding block VAR_BLOCK (if any) and a
276 stack frame id FRAME, read the value of the variable and return (pointer
277 to a) struct value containing the value.
279 VAR_BLOCK is needed if there's a possibility for VAR to be outside
280 FRAME. This is what happens if FRAME correspond to a nested function
281 and VAR is defined in the outer function. If callers know that VAR is
282 located in FRAME or is global/static, NULL can be passed as VAR_BLOCK.
284 Throw an error if the variable cannot be found. */
286 struct value
*(*la_read_var_value
) (struct symbol
*var
,
287 const struct block
*var_block
,
288 struct frame_info
*frame
);
290 /* PC is possibly an unknown languages trampoline.
291 If that PC falls in a trampoline belonging to this language,
292 return the address of the first pc in the real function, or 0
293 if it isn't a language tramp for this language. */
294 CORE_ADDR (*skip_trampoline
) (struct frame_info
*, CORE_ADDR
);
296 /* Now come some hooks for lookup_symbol. */
298 /* If this is non-NULL, specifies the name that of the implicit
299 local variable that refers to the current object instance. */
301 const char *la_name_of_this
;
303 /* True if the symbols names should be stored in GDB's data structures
304 for minimal/partial/full symbols using their linkage (aka mangled)
305 form; false if the symbol names should be demangled first.
307 Most languages implement symbol lookup by comparing the demangled
308 names, in which case it is advantageous to store that information
309 already demangled, and so would set this field to false.
311 On the other hand, some languages have opted for doing symbol
312 lookups by comparing mangled names instead, for reasons usually
313 specific to the language. Those languages should set this field
316 And finally, other languages such as C or Asm do not have
317 the concept of mangled vs demangled name, so those languages
318 should set this field to true as well, to prevent any accidental
319 demangling through an unrelated language's demangler. */
321 const bool la_store_sym_names_in_linkage_form_p
;
323 /* This is a function that lookup_symbol will call when it gets to
324 the part of symbol lookup where C looks up static and global
327 struct block_symbol (*la_lookup_symbol_nonlocal
)
328 (const struct language_defn
*,
330 const struct block
*,
333 /* Find the definition of the type with the given name. */
334 struct type
*(*la_lookup_transparent_type
) (const char *);
336 /* Return demangled language symbol, or NULL. */
337 char *(*la_demangle
) (const char *mangled
, int options
);
339 /* Demangle a symbol according to this language's rules. Unlike
340 la_demangle, this does not take any options.
342 *DEMANGLED will be set by this function.
344 If this function returns 0, then *DEMANGLED must always be set
347 If this function returns 1, the implementation may set this to
348 a xmalloc'd string holding the demangled form. However, it is
349 not required to. The string, if any, is owned by the caller.
351 The resulting string should be of the form that will be
352 installed into a symbol. */
353 int (*la_sniff_from_mangled_name
) (const char *mangled
, char **demangled
);
355 /* Return class name of a mangled method name or NULL. */
356 char *(*la_class_name_from_physname
) (const char *physname
);
358 /* Table for printing expressions. */
360 const struct op_print
*la_op_print_tab
;
362 /* Zero if the language has first-class arrays. True if there are no
363 array values, and array objects decay to pointers, as in C. */
367 /* Index to use for extracting the first element of a string. */
368 char string_lower_bound
;
370 /* The list of characters forming word boundaries. */
371 const char *(*la_word_break_characters
) (void);
373 /* Add to the completion tracker all symbols which are possible
374 completions for TEXT. WORD is the entire command on which the
375 completion is being made. If CODE is TYPE_CODE_UNDEF, then all
376 symbols should be examined; otherwise, only STRUCT_DOMAIN
377 symbols whose type has a code of CODE should be matched. */
378 void (*la_collect_symbol_completion_matches
)
379 (completion_tracker
&tracker
,
380 complete_symbol_mode mode
,
381 symbol_name_match_type match_type
,
384 enum type_code code
);
386 /* The per-architecture (OS/ABI) language information. */
387 void (*la_language_arch_info
) (struct gdbarch
*,
388 struct language_arch_info
*);
390 /* Print the index of an element of an array. */
391 void (*la_print_array_index
) (struct type
*index_type
,
393 struct ui_file
*stream
,
394 const struct value_print_options
*options
);
396 /* Return information about whether TYPE should be passed
397 (and returned) by reference at the language level. */
398 struct language_pass_by_ref_info (*la_pass_by_reference
)
401 /* Return an expression that can be used for a location
402 watchpoint. TYPE is a pointer type that points to the memory
403 to watch, and ADDR is the address of the watched memory. */
404 gdb::unique_xmalloc_ptr
<char> (*la_watch_location_expression
)
405 (struct type
*type
, CORE_ADDR addr
);
407 /* Return a pointer to the function that should be used to match a
408 symbol name against LOOKUP_NAME, according to this language's
409 rules. The matching algorithm depends on LOOKUP_NAME. For
410 example, on Ada, the matching algorithm depends on the symbol
411 name (wild/full/verbatim matching), and on whether we're doing
412 a normal lookup or a completion match lookup.
414 This field may be NULL, in which case
415 default_symbol_name_matcher is used to perform the
417 symbol_name_matcher_ftype
*(*la_get_symbol_name_matcher
)
418 (const lookup_name_info
&);
420 /* Find all symbols in the current program space matching NAME in
421 DOMAIN, according to this language's rules.
423 The search is done in BLOCK only.
424 The caller is responsible for iterating up through superblocks
427 For each one, call CALLBACK with the symbol. If CALLBACK
428 returns false, the iteration ends at that point.
430 This field may not be NULL. If the language does not need any
431 special processing here, 'iterate_over_symbols' should be
432 used as the definition. */
433 bool (*la_iterate_over_symbols
)
434 (const struct block
*block
, const lookup_name_info
&name
,
436 gdb::function_view
<symbol_found_callback_ftype
> callback
);
438 /* Hash the given symbol search name. Use
439 default_search_name_hash if no special treatment is
441 unsigned int (*la_search_name_hash
) (const char *name
);
443 /* Various operations on varobj. */
444 const struct lang_varobj_ops
*la_varobj_ops
;
446 /* If this language allows compilation from the gdb command line,
447 this method should be non-NULL. When called it should return
448 an instance of struct gcc_context appropriate to the language.
449 When defined this method must never return NULL; instead it
450 should throw an exception on failure. The returned compiler
451 instance is owned by its caller and must be deallocated by
452 calling its 'destroy' method. */
454 compile_instance
*(*la_get_compile_instance
) (void);
456 /* This method must be defined if 'la_get_gcc_context' is defined.
457 If 'la_get_gcc_context' is not defined, then this method is
460 This takes the user-supplied text and returns a new bit of code
463 INST is the compiler instance being used.
464 INPUT is the user's input text.
465 GDBARCH is the architecture to use.
466 EXPR_BLOCK is the block in which the expression is being
468 EXPR_PC is the PC at which the expression is being parsed. */
470 std::string (*la_compute_program
) (compile_instance
*inst
,
472 struct gdbarch
*gdbarch
,
473 const struct block
*expr_block
,
476 /* Return true if TYPE is a string type. */
477 bool (*la_is_string_type_p
) (struct type
*type
);
479 /* This string is used by the 'set print max-depth' setting. When GDB
480 replaces a struct or union (during value printing) that is "too
481 deep" this string is displayed instead. */
482 const char *la_struct_too_deep_ellipsis
;
486 /* Base class from which all other language classes derive. */
488 struct language_defn
: language_data
490 language_defn (enum language lang
, const language_data
&init_data
)
491 : language_data (init_data
)
493 /* We should only ever create one instance of each language. */
494 gdb_assert (languages
[lang
] == nullptr);
495 languages
[lang
] = this;
498 /* List of all known languages. */
499 static const struct language_defn
*languages
[nr_languages
];
502 /* Pointer to the language_defn for our current language. This pointer
503 always points to *some* valid struct; it can be used without checking
506 The current language affects expression parsing and evaluation
507 (FIXME: it might be cleaner to make the evaluation-related stuff
508 separate exp_opcodes for each different set of semantics. We
509 should at least think this through more clearly with respect to
510 what happens if the language is changed between parsing and
511 evaluation) and printing of things like types and arrays. It does
512 *not* affect symbol-reading-- each source file in a symbol-file has
513 its own language and we should keep track of that regardless of the
514 language when symbols are read. If we want some manual setting for
515 the language of symbol files (e.g. detecting when ".c" files are
516 C++), it should be a separate setting from the current_language. */
518 extern const struct language_defn
*current_language
;
520 /* Pointer to the language_defn expected by the user, e.g. the language
521 of main(), or the language we last mentioned in a message, or C. */
523 extern const struct language_defn
*expected_language
;
525 /* Warning issued when current_language and the language of the current
526 frame do not match. */
528 extern const char lang_frame_mismatch_warn
[];
531 language_mode_auto: current_language automatically set upon selection
532 of scope (e.g. stack frame)
533 language_mode_manual: current_language set only by user. */
535 extern enum language_mode
537 language_mode_auto
, language_mode_manual
541 struct type
*language_bool_type (const struct language_defn
*l
,
542 struct gdbarch
*gdbarch
);
544 struct type
*language_string_char_type (const struct language_defn
*l
,
545 struct gdbarch
*gdbarch
);
547 /* Look up type NAME in language L, and return its definition for architecture
548 GDBARCH. Returns NULL if not found. */
550 struct type
*language_lookup_primitive_type (const struct language_defn
*l
,
551 struct gdbarch
*gdbarch
,
554 /* Wrapper around language_lookup_primitive_type to return the
555 corresponding symbol. */
558 language_lookup_primitive_type_as_symbol (const struct language_defn
*l
,
559 struct gdbarch
*gdbarch
,
563 /* These macros define the behaviour of the expression
566 /* Should we range check values against the domain of their type? */
567 #define RANGE_CHECK (range_check != range_check_off)
569 /* "cast" really means conversion. */
570 /* FIXME -- should be a setting in language_defn. */
571 #define CAST_IS_CONVERSION(LANG) ((LANG)->la_language == language_c || \
572 (LANG)->la_language == language_cplus || \
573 (LANG)->la_language == language_objc)
575 extern void language_info (int);
577 extern enum language
set_language (enum language
);
580 /* This page contains functions that return things that are
581 specific to languages. Each of these functions is based on
582 the current setting of working_lang, which the user sets
583 with the "set language" command. */
585 #define LA_PRINT_TYPE(type,varstring,stream,show,level,flags) \
586 (current_language->la_print_type(type,varstring,stream,show,level,flags))
588 #define LA_PRINT_TYPEDEF(type,new_symbol,stream) \
589 (current_language->la_print_typedef(type,new_symbol,stream))
591 #define LA_VALUE_PRINT(val,stream,options) \
592 (current_language->la_value_print(val,stream,options))
594 #define LA_PRINT_CHAR(ch, type, stream) \
595 (current_language->la_printchar(ch, type, stream))
596 #define LA_PRINT_STRING(stream, elttype, string, length, encoding, force_ellipses, options) \
597 (current_language->la_printstr(stream, elttype, string, length, \
598 encoding, force_ellipses,options))
599 #define LA_EMIT_CHAR(ch, type, stream, quoter) \
600 (current_language->la_emitchar(ch, type, stream, quoter))
602 #define LA_PRINT_ARRAY_INDEX(index_type, index_value, stream, options) \
603 (current_language->la_print_array_index(index_type, index_value, stream, \
606 #define LA_ITERATE_OVER_SYMBOLS(BLOCK, NAME, DOMAIN, CALLBACK) \
607 (current_language->la_iterate_over_symbols (BLOCK, NAME, DOMAIN, CALLBACK))
609 /* Test a character to decide whether it can be printed in literal form
610 or needs to be printed in another representation. For example,
611 in C the literal form of the character with octal value 141 is 'a'
612 and the "other representation" is '\141'. The "other representation"
613 is program language dependent. */
615 #define PRINT_LITERAL_FORM(c) \
617 && ((c) < 0x7F || (c) >= 0xA0) \
618 && (!sevenbit_strings || (c) < 0x80))
620 /* Type predicates */
622 extern int pointer_type (struct type
*);
624 /* Return true if TYPE is a string type, otherwise return false. This
625 default implementation only detects TYPE_CODE_STRING. */
626 extern bool default_is_string_type_p (struct type
*type
);
630 extern void range_error (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
632 /* Data: Does this value represent "truth" to the current language? */
634 extern int value_true (struct value
*);
636 /* Misc: The string representing a particular enum language. */
638 extern enum language
language_enum (const char *str
);
640 extern const struct language_defn
*language_def (enum language
);
642 extern const char *language_str (enum language
);
644 /* Check for a language-specific trampoline. */
646 extern CORE_ADDR
skip_language_trampoline (struct frame_info
*, CORE_ADDR pc
);
648 /* Return demangled language symbol, or NULL. */
649 extern char *language_demangle (const struct language_defn
*current_language
,
650 const char *mangled
, int options
);
652 /* A wrapper for la_sniff_from_mangled_name. The arguments and result
653 are as for the method. */
655 extern int language_sniff_from_mangled_name (const struct language_defn
*lang
,
659 /* Return class name from physname, or NULL. */
660 extern char *language_class_name_from_physname (const struct language_defn
*,
661 const char *physname
);
663 /* Splitting strings into words. */
664 extern const char *default_word_break_characters (void);
666 /* Print the index of an array element using the C99 syntax. */
667 extern void default_print_array_index (struct type
*index_type
, LONGEST index
,
668 struct ui_file
*stream
,
669 const struct value_print_options
*options
);
671 /* Return information about whether TYPE should be passed
672 (and returned) by reference at the language level. */
673 struct language_pass_by_ref_info
language_pass_by_reference (struct type
*type
);
675 /* Return a default struct that provides pass-by-reference information
676 about the given TYPE. Languages should update the default values
678 struct language_pass_by_ref_info
default_pass_by_reference (struct type
*type
);
680 /* The default implementation of la_print_typedef. */
681 void default_print_typedef (struct type
*type
, struct symbol
*new_symbol
,
682 struct ui_file
*stream
);
684 /* Default name hashing function. */
686 /* Produce an unsigned hash value from SEARCH_NAME that is consistent
687 with strcmp_iw, strcmp, and, at least on Ada symbols, wild_match.
688 That is, two identifiers equivalent according to any of those three
689 comparison operators hash to the same value. */
690 extern unsigned int default_search_name_hash (const char *search_name
);
692 void c_get_string (struct value
*value
,
693 gdb::unique_xmalloc_ptr
<gdb_byte
> *buffer
,
694 int *length
, struct type
**char_type
,
695 const char **charset
);
697 /* The default implementation of la_symbol_name_matcher. Matches with
699 extern bool default_symbol_name_matcher
700 (const char *symbol_search_name
,
701 const lookup_name_info
&lookup_name
,
702 completion_match_result
*comp_match_res
);
704 /* Get LANG's symbol_name_matcher method for LOOKUP_NAME. Returns
705 default_symbol_name_matcher if not set. LANG is used as a hint;
706 the function may ignore it depending on the current language and
707 LOOKUP_NAME. Specifically, if the current language is Ada, this
708 may return an Ada matcher regardless of LANG. */
709 symbol_name_matcher_ftype
*get_symbol_name_matcher
710 (const language_defn
*lang
, const lookup_name_info
&lookup_name
);
712 /* Save the current language and restore it upon destruction. */
714 class scoped_restore_current_language
718 explicit scoped_restore_current_language ()
719 : m_lang (current_language
->la_language
)
723 ~scoped_restore_current_language ()
725 set_language (m_lang
);
728 scoped_restore_current_language (const scoped_restore_current_language
&)
730 scoped_restore_current_language
&operator=
731 (const scoped_restore_current_language
&) = delete;
735 enum language m_lang
;
738 /* If language_mode is language_mode_auto,
739 then switch current language to the language of SYM
740 and restore current language upon destruction.
744 class scoped_switch_to_sym_language_if_auto
748 explicit scoped_switch_to_sym_language_if_auto (const struct symbol
*sym
)
750 if (language_mode
== language_mode_auto
)
752 m_lang
= current_language
->la_language
;
754 set_language (sym
->language ());
759 /* Assign to m_lang to silence a GCC warning. See
760 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80635. */
761 m_lang
= language_unknown
;
765 ~scoped_switch_to_sym_language_if_auto ()
768 set_language (m_lang
);
771 DISABLE_COPY_AND_ASSIGN (scoped_switch_to_sym_language_if_auto
);
775 enum language m_lang
;
778 #endif /* defined (LANGUAGE_H) */