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 /* Splitting strings into words. */
173 extern const char *default_word_break_characters (void);
175 /* Structure tying together assorted information about a language.
177 As we move over from the old structure based languages to a class
178 hierarchy of languages this structure will continue to contain a
179 mixture of both data and function pointers.
181 Once the class hierarchy of languages in place the first task is to
182 remove the function pointers from this structure and convert them into
183 member functions on the different language classes.
185 The current plan it to keep the constant data that describes a language
186 in this structure, and have each language pass in an instance of this
187 structure at construction time. */
191 /* Name of the language. */
195 /* Natural or official name of the language. */
197 const char *la_natural_name
;
199 /* its symtab language-enum (defs.h). */
201 enum language la_language
;
203 /* Default range checking. */
205 enum range_check la_range_check
;
207 /* Default case sensitivity. */
208 enum case_sensitivity la_case_sensitivity
;
210 /* Multi-dimensional array ordering. */
211 enum array_ordering la_array_ordering
;
213 /* Style of macro expansion, if any, supported by this language. */
214 enum macro_expansion la_macro_expansion
;
216 /* A NULL-terminated array of file extensions for this language.
217 The extension must include the ".", like ".c". If this
218 language doesn't need to provide any filename extensions, this
221 const char *const *la_filename_extensions
;
223 /* Definitions related to expression printing, prefixifying, and
226 const struct exp_descriptor
*la_exp_desc
;
228 /* Parser function. */
230 int (*la_parser
) (struct parser_state
*);
232 /* Given an expression *EXPP created by prefixifying the result of
233 la_parser, perform any remaining processing necessary to complete
234 its translation. *EXPP may change; la_post_parser is responsible
235 for releasing its previous contents, if necessary. If
236 VOID_CONTEXT_P, then no value is expected from the expression.
237 If COMPLETING is non-zero, then the expression has been parsed
238 for completion, not evaluation. */
240 void (*la_post_parser
) (expression_up
*expp
, int void_context_p
,
241 int completing
, innermost_block_tracker
*tracker
);
243 void (*la_printchar
) (int ch
, struct type
*chtype
,
244 struct ui_file
* stream
);
246 void (*la_printstr
) (struct ui_file
* stream
, struct type
*elttype
,
247 const gdb_byte
*string
, unsigned int length
,
248 const char *encoding
, int force_ellipses
,
249 const struct value_print_options
*);
251 void (*la_emitchar
) (int ch
, struct type
*chtype
,
252 struct ui_file
* stream
, int quoter
);
254 /* Print a typedef using syntax appropriate for this language.
255 TYPE is the underlying type. NEW_SYMBOL is the symbol naming
256 the type. STREAM is the output stream on which to print. */
258 void (*la_print_typedef
) (struct type
*type
, struct symbol
*new_symbol
,
259 struct ui_file
*stream
);
261 /* Now come some hooks for lookup_symbol. */
263 /* If this is non-NULL, specifies the name that of the implicit
264 local variable that refers to the current object instance. */
266 const char *la_name_of_this
;
268 /* True if the symbols names should be stored in GDB's data structures
269 for minimal/partial/full symbols using their linkage (aka mangled)
270 form; false if the symbol names should be demangled first.
272 Most languages implement symbol lookup by comparing the demangled
273 names, in which case it is advantageous to store that information
274 already demangled, and so would set this field to false.
276 On the other hand, some languages have opted for doing symbol
277 lookups by comparing mangled names instead, for reasons usually
278 specific to the language. Those languages should set this field
281 And finally, other languages such as C or Asm do not have
282 the concept of mangled vs demangled name, so those languages
283 should set this field to true as well, to prevent any accidental
284 demangling through an unrelated language's demangler. */
286 const bool la_store_sym_names_in_linkage_form_p
;
288 /* Table for printing expressions. */
290 const struct op_print
*la_op_print_tab
;
292 /* Zero if the language has first-class arrays. True if there are no
293 array values, and array objects decay to pointers, as in C. */
297 /* Index to use for extracting the first element of a string. */
298 char string_lower_bound
;
300 /* Various operations on varobj. */
301 const struct lang_varobj_ops
*la_varobj_ops
;
303 /* Return true if TYPE is a string type. */
304 bool (*la_is_string_type_p
) (struct type
*type
);
306 /* This string is used by the 'set print max-depth' setting. When GDB
307 replaces a struct or union (during value printing) that is "too
308 deep" this string is displayed instead. */
309 const char *la_struct_too_deep_ellipsis
;
313 /* Base class from which all other language classes derive. */
315 struct language_defn
: language_data
317 language_defn (enum language lang
, const language_data
&init_data
)
318 : language_data (init_data
)
320 /* We should only ever create one instance of each language. */
321 gdb_assert (languages
[lang
] == nullptr);
322 languages
[lang
] = this;
325 /* Print the index of an element of an array. This default
326 implementation prints using C99 syntax. */
328 virtual void print_array_index (struct type
*index_type
,
330 struct ui_file
*stream
,
331 const value_print_options
*options
) const;
333 /* Given a symbol VAR, the corresponding block VAR_BLOCK (if any) and a
334 stack frame id FRAME, read the value of the variable and return (pointer
335 to a) struct value containing the value.
337 VAR_BLOCK is needed if there's a possibility for VAR to be outside
338 FRAME. This is what happens if FRAME correspond to a nested function
339 and VAR is defined in the outer function. If callers know that VAR is
340 located in FRAME or is global/static, NULL can be passed as VAR_BLOCK.
342 Throw an error if the variable cannot be found. */
344 virtual struct value
*read_var_value (struct symbol
*var
,
345 const struct block
*var_block
,
346 struct frame_info
*frame
) const;
348 /* Return information about whether TYPE should be passed
349 (and returned) by reference at the language level. The default
350 implementation returns a LANGUAGE_PASS_BY_REF_INFO initialised in its
353 virtual struct language_pass_by_ref_info pass_by_reference_info
354 (struct type
*type
) const
359 /* The per-architecture (OS/ABI) language information. */
361 virtual void language_arch_info (struct gdbarch
*,
362 struct language_arch_info
*) const = 0;
364 /* Find the definition of the type with the given name. */
366 virtual struct type
*lookup_transparent_type (const char *name
) const
368 return basic_lookup_transparent_type (name
);
371 /* Find all symbols in the current program space matching NAME in
372 DOMAIN, according to this language's rules.
374 The search is done in BLOCK only.
375 The caller is responsible for iterating up through superblocks
378 For each one, call CALLBACK with the symbol. If CALLBACK
379 returns false, the iteration ends at that point.
381 This field may not be NULL. If the language does not need any
382 special processing here, 'iterate_over_symbols' should be
383 used as the definition. */
384 virtual bool iterate_over_symbols
385 (const struct block
*block
, const lookup_name_info
&name
,
387 gdb::function_view
<symbol_found_callback_ftype
> callback
) const
389 return ::iterate_over_symbols (block
, name
, domain
, callback
);
392 /* Return a pointer to the function that should be used to match a
393 symbol name against LOOKUP_NAME, according to this language's
394 rules. The matching algorithm depends on LOOKUP_NAME. For
395 example, on Ada, the matching algorithm depends on the symbol
396 name (wild/full/verbatim matching), and on whether we're doing
397 a normal lookup or a completion match lookup.
399 As Ada wants to capture symbol matching for all languages in some
400 cases, then this method is a non-overridable interface. Languages
401 should override GET_SYMBOL_NAME_MATCHER_INNER if they need to. */
403 symbol_name_matcher_ftype
*get_symbol_name_matcher
404 (const lookup_name_info
&lookup_name
) const;
406 /* If this language allows compilation from the gdb command line, then
407 this method will return an instance of struct gcc_context appropriate
408 to the language. If compilation for this language is generally
409 supported, but something goes wrong then an exception is thrown. The
410 returned compiler instance is owned by its caller and must be
411 deallocated by the caller. If compilation is not supported for this
412 language then this method returns NULL. */
414 virtual compile_instance
*get_compile_instance () const
419 /* This method must be overridden if 'get_compile_instance' is
422 This takes the user-supplied text and returns a new bit of code
425 INST is the compiler instance being used.
426 INPUT is the user's input text.
427 GDBARCH is the architecture to use.
428 EXPR_BLOCK is the block in which the expression is being
430 EXPR_PC is the PC at which the expression is being parsed. */
432 virtual std::string
compute_program (compile_instance
*inst
,
434 struct gdbarch
*gdbarch
,
435 const struct block
*expr_block
,
436 CORE_ADDR expr_pc
) const
438 gdb_assert_not_reached ("language_defn::compute_program");
441 /* Hash the given symbol search name. */
442 virtual unsigned int search_name_hash (const char *name
) const;
444 /* Demangle a symbol according to this language's rules. Unlike
445 la_demangle, this does not take any options.
447 *DEMANGLED will be set by this function.
449 If this function returns false, then *DEMANGLED must always be set
452 If this function returns true, the implementation may set this to
453 a xmalloc'd string holding the demangled form. However, it is
454 not required to. The string, if any, is owned by the caller.
456 The resulting string should be of the form that will be
457 installed into a symbol. */
458 virtual bool sniff_from_mangled_name (const char *mangled
,
459 char **demangled
) const
461 *demangled
= nullptr;
465 /* Return demangled language symbol version of MANGLED, or NULL. */
466 virtual char *demangle (const char *mangled
, int options
) const
471 /* Print a type using syntax appropriate for this language. */
473 virtual void print_type (struct type
*, const char *, struct ui_file
*, int,
474 int, const struct type_print_options
*) const = 0;
476 /* PC is possibly an unknown languages trampoline.
477 If that PC falls in a trampoline belonging to this language, return
478 the address of the first pc in the real function, or 0 if it isn't a
479 language tramp for this language. */
480 virtual CORE_ADDR
skip_trampoline (struct frame_info
*fi
, CORE_ADDR pc
) const
482 return (CORE_ADDR
) 0;
485 /* Return class name of a mangled method name or NULL. */
486 virtual char *class_name_from_physname (const char *physname
) const
491 /* The list of characters forming word boundaries. */
492 virtual const char *word_break_characters (void) const
494 return default_word_break_characters ();
497 /* Add to the completion tracker all symbols which are possible
498 completions for TEXT. WORD is the entire command on which the
499 completion is being made. If CODE is TYPE_CODE_UNDEF, then all
500 symbols should be examined; otherwise, only STRUCT_DOMAIN symbols
501 whose type has a code of CODE should be matched. */
503 virtual void collect_symbol_completion_matches
504 (completion_tracker
&tracker
,
505 complete_symbol_mode mode
,
506 symbol_name_match_type name_match_type
,
509 enum type_code code
) const
511 return default_collect_symbol_completion_matches_break_on
512 (tracker
, mode
, name_match_type
, text
, word
, "", code
);
515 /* This is a function that lookup_symbol will call when it gets to
516 the part of symbol lookup where C looks up static and global
517 variables. This default implements the basic C lookup rules. */
519 virtual struct block_symbol lookup_symbol_nonlocal
521 const struct block
*block
,
522 const domain_enum domain
) const;
524 /* Return an expression that can be used for a location
525 watchpoint. TYPE is a pointer type that points to the memory
526 to watch, and ADDR is the address of the watched memory. */
527 virtual gdb::unique_xmalloc_ptr
<char> watch_location_expression
528 (struct type
*type
, CORE_ADDR addr
) const;
530 /* List of all known languages. */
531 static const struct language_defn
*languages
[nr_languages
];
533 /* Print a top-level value using syntax appropriate for this language. */
534 virtual void value_print (struct value
*val
, struct ui_file
*stream
,
535 const struct value_print_options
*options
) const;
537 /* Print a value using syntax appropriate for this language. RECURSE is
538 the recursion depth. It is zero-based. */
539 virtual void value_print_inner
540 (struct value
*val
, struct ui_file
*stream
, int recurse
,
541 const struct value_print_options
*options
) const;
545 /* This is the overridable part of the GET_SYMBOL_NAME_MATCHER method.
546 See that method for a description of the arguments. */
548 virtual symbol_name_matcher_ftype
*get_symbol_name_matcher_inner
549 (const lookup_name_info
&lookup_name
) const;
552 /* Pointer to the language_defn for our current language. This pointer
553 always points to *some* valid struct; it can be used without checking
556 The current language affects expression parsing and evaluation
557 (FIXME: it might be cleaner to make the evaluation-related stuff
558 separate exp_opcodes for each different set of semantics. We
559 should at least think this through more clearly with respect to
560 what happens if the language is changed between parsing and
561 evaluation) and printing of things like types and arrays. It does
562 *not* affect symbol-reading-- each source file in a symbol-file has
563 its own language and we should keep track of that regardless of the
564 language when symbols are read. If we want some manual setting for
565 the language of symbol files (e.g. detecting when ".c" files are
566 C++), it should be a separate setting from the current_language. */
568 extern const struct language_defn
*current_language
;
570 /* Pointer to the language_defn expected by the user, e.g. the language
571 of main(), or the language we last mentioned in a message, or C. */
573 extern const struct language_defn
*expected_language
;
575 /* Warning issued when current_language and the language of the current
576 frame do not match. */
578 extern const char lang_frame_mismatch_warn
[];
581 language_mode_auto: current_language automatically set upon selection
582 of scope (e.g. stack frame)
583 language_mode_manual: current_language set only by user. */
585 extern enum language_mode
587 language_mode_auto
, language_mode_manual
591 struct type
*language_bool_type (const struct language_defn
*l
,
592 struct gdbarch
*gdbarch
);
594 struct type
*language_string_char_type (const struct language_defn
*l
,
595 struct gdbarch
*gdbarch
);
597 /* Look up type NAME in language L, and return its definition for architecture
598 GDBARCH. Returns NULL if not found. */
600 struct type
*language_lookup_primitive_type (const struct language_defn
*l
,
601 struct gdbarch
*gdbarch
,
604 /* Wrapper around language_lookup_primitive_type to return the
605 corresponding symbol. */
608 language_lookup_primitive_type_as_symbol (const struct language_defn
*l
,
609 struct gdbarch
*gdbarch
,
613 /* These macros define the behaviour of the expression
616 /* Should we range check values against the domain of their type? */
617 #define RANGE_CHECK (range_check != range_check_off)
619 /* "cast" really means conversion. */
620 /* FIXME -- should be a setting in language_defn. */
621 #define CAST_IS_CONVERSION(LANG) ((LANG)->la_language == language_c || \
622 (LANG)->la_language == language_cplus || \
623 (LANG)->la_language == language_objc)
625 extern void language_info (int);
627 extern enum language
set_language (enum language
);
630 /* This page contains functions that return things that are
631 specific to languages. Each of these functions is based on
632 the current setting of working_lang, which the user sets
633 with the "set language" command. */
635 #define LA_PRINT_TYPE(type,varstring,stream,show,level,flags) \
636 (current_language->print_type(type,varstring,stream,show,level,flags))
638 #define LA_PRINT_TYPEDEF(type,new_symbol,stream) \
639 (current_language->la_print_typedef(type,new_symbol,stream))
641 #define LA_VALUE_PRINT(val,stream,options) \
642 (current_language->value_print (val,stream,options))
644 #define LA_PRINT_CHAR(ch, type, stream) \
645 (current_language->la_printchar(ch, type, stream))
646 #define LA_PRINT_STRING(stream, elttype, string, length, encoding, force_ellipses, options) \
647 (current_language->la_printstr(stream, elttype, string, length, \
648 encoding, force_ellipses,options))
649 #define LA_EMIT_CHAR(ch, type, stream, quoter) \
650 (current_language->la_emitchar(ch, type, stream, quoter))
652 #define LA_PRINT_ARRAY_INDEX(index_type, index_value, stream, options) \
653 (current_language->print_array_index(index_type, index_value, stream, \
656 #define LA_ITERATE_OVER_SYMBOLS(BLOCK, NAME, DOMAIN, CALLBACK) \
657 (current_language->iterate_over_symbols (BLOCK, NAME, DOMAIN, CALLBACK))
659 /* Test a character to decide whether it can be printed in literal form
660 or needs to be printed in another representation. For example,
661 in C the literal form of the character with octal value 141 is 'a'
662 and the "other representation" is '\141'. The "other representation"
663 is program language dependent. */
665 #define PRINT_LITERAL_FORM(c) \
667 && ((c) < 0x7F || (c) >= 0xA0) \
668 && (!sevenbit_strings || (c) < 0x80))
670 /* Type predicates */
672 extern int pointer_type (struct type
*);
674 /* Return true if TYPE is a string type, otherwise return false. This
675 default implementation only detects TYPE_CODE_STRING. */
676 extern bool default_is_string_type_p (struct type
*type
);
680 extern void range_error (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
682 /* Data: Does this value represent "truth" to the current language? */
684 extern int value_true (struct value
*);
686 /* Misc: The string representing a particular enum language. */
688 extern enum language
language_enum (const char *str
);
690 extern const struct language_defn
*language_def (enum language
);
692 extern const char *language_str (enum language
);
694 /* Check for a language-specific trampoline. */
696 extern CORE_ADDR
skip_language_trampoline (struct frame_info
*, CORE_ADDR pc
);
698 /* Return demangled language symbol, or NULL. */
699 extern char *language_demangle (const struct language_defn
*current_language
,
700 const char *mangled
, int options
);
702 /* Return information about whether TYPE should be passed
703 (and returned) by reference at the language level. */
704 struct language_pass_by_ref_info
language_pass_by_reference (struct type
*type
);
706 /* The default implementation of la_print_typedef. */
707 void default_print_typedef (struct type
*type
, struct symbol
*new_symbol
,
708 struct ui_file
*stream
);
710 void c_get_string (struct value
*value
,
711 gdb::unique_xmalloc_ptr
<gdb_byte
> *buffer
,
712 int *length
, struct type
**char_type
,
713 const char **charset
);
715 /* Get LANG's symbol_name_matcher method for LOOKUP_NAME. Returns
716 default_symbol_name_matcher if not set. LANG is used as a hint;
717 the function may ignore it depending on the current language and
718 LOOKUP_NAME. Specifically, if the current language is Ada, this
719 may return an Ada matcher regardless of LANG. */
720 symbol_name_matcher_ftype
*get_symbol_name_matcher
721 (const language_defn
*lang
, const lookup_name_info
&lookup_name
);
723 /* Save the current language and restore it upon destruction. */
725 class scoped_restore_current_language
729 explicit scoped_restore_current_language ()
730 : m_lang (current_language
->la_language
)
734 ~scoped_restore_current_language ()
736 set_language (m_lang
);
739 scoped_restore_current_language (const scoped_restore_current_language
&)
741 scoped_restore_current_language
&operator=
742 (const scoped_restore_current_language
&) = delete;
746 enum language m_lang
;
749 /* If language_mode is language_mode_auto,
750 then switch current language to the language of SYM
751 and restore current language upon destruction.
755 class scoped_switch_to_sym_language_if_auto
759 explicit scoped_switch_to_sym_language_if_auto (const struct symbol
*sym
)
761 if (language_mode
== language_mode_auto
)
763 m_lang
= current_language
->la_language
;
765 set_language (sym
->language ());
770 /* Assign to m_lang to silence a GCC warning. See
771 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80635. */
772 m_lang
= language_unknown
;
776 ~scoped_switch_to_sym_language_if_auto ()
779 set_language (m_lang
);
782 DISABLE_COPY_AND_ASSIGN (scoped_switch_to_sym_language_if_auto
);
786 enum language m_lang
;
789 #endif /* defined (LANGUAGE_H) */