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 typedef using syntax appropriate for this language.
252 TYPE is the underlying type. NEW_SYMBOL is the symbol naming
253 the type. STREAM is the output stream on which to print. */
255 void (*la_print_typedef
) (struct type
*type
, struct symbol
*new_symbol
,
256 struct ui_file
*stream
);
258 /* Print a value using syntax appropriate for this language.
259 RECURSE is the recursion depth. It is zero-based. */
261 void (*la_value_print_inner
) (struct value
*, struct ui_file
*,
263 const struct value_print_options
*);
265 /* Print a top-level value using syntax appropriate for this language. */
267 void (*la_value_print
) (struct value
*, struct ui_file
*,
268 const struct value_print_options
*);
270 /* Now come some hooks for lookup_symbol. */
272 /* If this is non-NULL, specifies the name that of the implicit
273 local variable that refers to the current object instance. */
275 const char *la_name_of_this
;
277 /* True if the symbols names should be stored in GDB's data structures
278 for minimal/partial/full symbols using their linkage (aka mangled)
279 form; false if the symbol names should be demangled first.
281 Most languages implement symbol lookup by comparing the demangled
282 names, in which case it is advantageous to store that information
283 already demangled, and so would set this field to false.
285 On the other hand, some languages have opted for doing symbol
286 lookups by comparing mangled names instead, for reasons usually
287 specific to the language. Those languages should set this field
290 And finally, other languages such as C or Asm do not have
291 the concept of mangled vs demangled name, so those languages
292 should set this field to true as well, to prevent any accidental
293 demangling through an unrelated language's demangler. */
295 const bool la_store_sym_names_in_linkage_form_p
;
297 /* This is a function that lookup_symbol will call when it gets to
298 the part of symbol lookup where C looks up static and global
301 struct block_symbol (*la_lookup_symbol_nonlocal
)
302 (const struct language_defn
*,
304 const struct block
*,
307 /* Table for printing expressions. */
309 const struct op_print
*la_op_print_tab
;
311 /* Zero if the language has first-class arrays. True if there are no
312 array values, and array objects decay to pointers, as in C. */
316 /* Index to use for extracting the first element of a string. */
317 char string_lower_bound
;
319 /* The list of characters forming word boundaries. */
320 const char *(*la_word_break_characters
) (void);
322 /* Add to the completion tracker all symbols which are possible
323 completions for TEXT. WORD is the entire command on which the
324 completion is being made. If CODE is TYPE_CODE_UNDEF, then all
325 symbols should be examined; otherwise, only STRUCT_DOMAIN
326 symbols whose type has a code of CODE should be matched. */
327 void (*la_collect_symbol_completion_matches
)
328 (completion_tracker
&tracker
,
329 complete_symbol_mode mode
,
330 symbol_name_match_type match_type
,
333 enum type_code code
);
335 /* Return an expression that can be used for a location
336 watchpoint. TYPE is a pointer type that points to the memory
337 to watch, and ADDR is the address of the watched memory. */
338 gdb::unique_xmalloc_ptr
<char> (*la_watch_location_expression
)
339 (struct type
*type
, CORE_ADDR addr
);
341 /* Various operations on varobj. */
342 const struct lang_varobj_ops
*la_varobj_ops
;
344 /* Return true if TYPE is a string type. */
345 bool (*la_is_string_type_p
) (struct type
*type
);
347 /* This string is used by the 'set print max-depth' setting. When GDB
348 replaces a struct or union (during value printing) that is "too
349 deep" this string is displayed instead. */
350 const char *la_struct_too_deep_ellipsis
;
354 /* Base class from which all other language classes derive. */
356 struct language_defn
: language_data
358 language_defn (enum language lang
, const language_data
&init_data
)
359 : language_data (init_data
)
361 /* We should only ever create one instance of each language. */
362 gdb_assert (languages
[lang
] == nullptr);
363 languages
[lang
] = this;
366 /* Print the index of an element of an array. This default
367 implementation prints using C99 syntax. */
369 virtual void print_array_index (struct type
*index_type
,
371 struct ui_file
*stream
,
372 const value_print_options
*options
) const;
374 /* Given a symbol VAR, the corresponding block VAR_BLOCK (if any) and a
375 stack frame id FRAME, read the value of the variable and return (pointer
376 to a) struct value containing the value.
378 VAR_BLOCK is needed if there's a possibility for VAR to be outside
379 FRAME. This is what happens if FRAME correspond to a nested function
380 and VAR is defined in the outer function. If callers know that VAR is
381 located in FRAME or is global/static, NULL can be passed as VAR_BLOCK.
383 Throw an error if the variable cannot be found. */
385 virtual struct value
*read_var_value (struct symbol
*var
,
386 const struct block
*var_block
,
387 struct frame_info
*frame
) const;
389 /* Return information about whether TYPE should be passed
390 (and returned) by reference at the language level. The default
391 implementation returns a LANGUAGE_PASS_BY_REF_INFO initialised in its
394 virtual struct language_pass_by_ref_info pass_by_reference_info
395 (struct type
*type
) const
400 /* The per-architecture (OS/ABI) language information. */
402 virtual void language_arch_info (struct gdbarch
*,
403 struct language_arch_info
*) const = 0;
405 /* Find the definition of the type with the given name. */
407 virtual struct type
*lookup_transparent_type (const char *name
) const
409 return basic_lookup_transparent_type (name
);
412 /* Find all symbols in the current program space matching NAME in
413 DOMAIN, according to this language's rules.
415 The search is done in BLOCK only.
416 The caller is responsible for iterating up through superblocks
419 For each one, call CALLBACK with the symbol. If CALLBACK
420 returns false, the iteration ends at that point.
422 This field may not be NULL. If the language does not need any
423 special processing here, 'iterate_over_symbols' should be
424 used as the definition. */
425 virtual bool iterate_over_symbols
426 (const struct block
*block
, const lookup_name_info
&name
,
428 gdb::function_view
<symbol_found_callback_ftype
> callback
) const
430 return ::iterate_over_symbols (block
, name
, domain
, callback
);
433 /* Return a pointer to the function that should be used to match a
434 symbol name against LOOKUP_NAME, according to this language's
435 rules. The matching algorithm depends on LOOKUP_NAME. For
436 example, on Ada, the matching algorithm depends on the symbol
437 name (wild/full/verbatim matching), and on whether we're doing
438 a normal lookup or a completion match lookup.
440 As Ada wants to capture symbol matching for all languages in some
441 cases, then this method is a non-overridable interface. Languages
442 should override GET_SYMBOL_NAME_MATCHER_INNER if they need to. */
444 symbol_name_matcher_ftype
*get_symbol_name_matcher
445 (const lookup_name_info
&lookup_name
) const;
447 /* If this language allows compilation from the gdb command line, then
448 this method will return an instance of struct gcc_context appropriate
449 to the language. If compilation for this language is generally
450 supported, but something goes wrong then an exception is thrown. The
451 returned compiler instance is owned by its caller and must be
452 deallocated by the caller. If compilation is not supported for this
453 language then this method returns NULL. */
455 virtual compile_instance
*get_compile_instance () const
460 /* This method must be overridden if 'get_compile_instance' is
463 This takes the user-supplied text and returns a new bit of code
466 INST is the compiler instance being used.
467 INPUT is the user's input text.
468 GDBARCH is the architecture to use.
469 EXPR_BLOCK is the block in which the expression is being
471 EXPR_PC is the PC at which the expression is being parsed. */
473 virtual std::string
compute_program (compile_instance
*inst
,
475 struct gdbarch
*gdbarch
,
476 const struct block
*expr_block
,
477 CORE_ADDR expr_pc
) const
479 gdb_assert_not_reached ("language_defn::compute_program");
482 /* Hash the given symbol search name. */
483 virtual unsigned int search_name_hash (const char *name
) const;
485 /* Demangle a symbol according to this language's rules. Unlike
486 la_demangle, this does not take any options.
488 *DEMANGLED will be set by this function.
490 If this function returns false, then *DEMANGLED must always be set
493 If this function returns true, the implementation may set this to
494 a xmalloc'd string holding the demangled form. However, it is
495 not required to. The string, if any, is owned by the caller.
497 The resulting string should be of the form that will be
498 installed into a symbol. */
499 virtual bool sniff_from_mangled_name (const char *mangled
,
500 char **demangled
) const
502 *demangled
= nullptr;
506 /* Return demangled language symbol version of MANGLED, or NULL. */
507 virtual char *demangle (const char *mangled
, int options
) const
512 /* Print a type using syntax appropriate for this language. */
514 virtual void print_type (struct type
*, const char *, struct ui_file
*, int,
515 int, const struct type_print_options
*) const = 0;
517 /* PC is possibly an unknown languages trampoline.
518 If that PC falls in a trampoline belonging to this language, return
519 the address of the first pc in the real function, or 0 if it isn't a
520 language tramp for this language. */
521 virtual CORE_ADDR
skip_trampoline (struct frame_info
*fi
, CORE_ADDR pc
) const
523 return (CORE_ADDR
) 0;
526 /* Return class name of a mangled method name or NULL. */
527 virtual char *class_name_from_physname (const char *physname
) const
532 /* List of all known languages. */
533 static const struct language_defn
*languages
[nr_languages
];
537 /* This is the overridable part of the GET_SYMBOL_NAME_MATCHER method.
538 See that method for a description of the arguments. */
540 virtual symbol_name_matcher_ftype
*get_symbol_name_matcher_inner
541 (const lookup_name_info
&lookup_name
) const;
544 /* Pointer to the language_defn for our current language. This pointer
545 always points to *some* valid struct; it can be used without checking
548 The current language affects expression parsing and evaluation
549 (FIXME: it might be cleaner to make the evaluation-related stuff
550 separate exp_opcodes for each different set of semantics. We
551 should at least think this through more clearly with respect to
552 what happens if the language is changed between parsing and
553 evaluation) and printing of things like types and arrays. It does
554 *not* affect symbol-reading-- each source file in a symbol-file has
555 its own language and we should keep track of that regardless of the
556 language when symbols are read. If we want some manual setting for
557 the language of symbol files (e.g. detecting when ".c" files are
558 C++), it should be a separate setting from the current_language. */
560 extern const struct language_defn
*current_language
;
562 /* Pointer to the language_defn expected by the user, e.g. the language
563 of main(), or the language we last mentioned in a message, or C. */
565 extern const struct language_defn
*expected_language
;
567 /* Warning issued when current_language and the language of the current
568 frame do not match. */
570 extern const char lang_frame_mismatch_warn
[];
573 language_mode_auto: current_language automatically set upon selection
574 of scope (e.g. stack frame)
575 language_mode_manual: current_language set only by user. */
577 extern enum language_mode
579 language_mode_auto
, language_mode_manual
583 struct type
*language_bool_type (const struct language_defn
*l
,
584 struct gdbarch
*gdbarch
);
586 struct type
*language_string_char_type (const struct language_defn
*l
,
587 struct gdbarch
*gdbarch
);
589 /* Look up type NAME in language L, and return its definition for architecture
590 GDBARCH. Returns NULL if not found. */
592 struct type
*language_lookup_primitive_type (const struct language_defn
*l
,
593 struct gdbarch
*gdbarch
,
596 /* Wrapper around language_lookup_primitive_type to return the
597 corresponding symbol. */
600 language_lookup_primitive_type_as_symbol (const struct language_defn
*l
,
601 struct gdbarch
*gdbarch
,
605 /* These macros define the behaviour of the expression
608 /* Should we range check values against the domain of their type? */
609 #define RANGE_CHECK (range_check != range_check_off)
611 /* "cast" really means conversion. */
612 /* FIXME -- should be a setting in language_defn. */
613 #define CAST_IS_CONVERSION(LANG) ((LANG)->la_language == language_c || \
614 (LANG)->la_language == language_cplus || \
615 (LANG)->la_language == language_objc)
617 extern void language_info (int);
619 extern enum language
set_language (enum language
);
622 /* This page contains functions that return things that are
623 specific to languages. Each of these functions is based on
624 the current setting of working_lang, which the user sets
625 with the "set language" command. */
627 #define LA_PRINT_TYPE(type,varstring,stream,show,level,flags) \
628 (current_language->print_type(type,varstring,stream,show,level,flags))
630 #define LA_PRINT_TYPEDEF(type,new_symbol,stream) \
631 (current_language->la_print_typedef(type,new_symbol,stream))
633 #define LA_VALUE_PRINT(val,stream,options) \
634 (current_language->la_value_print(val,stream,options))
636 #define LA_PRINT_CHAR(ch, type, stream) \
637 (current_language->la_printchar(ch, type, stream))
638 #define LA_PRINT_STRING(stream, elttype, string, length, encoding, force_ellipses, options) \
639 (current_language->la_printstr(stream, elttype, string, length, \
640 encoding, force_ellipses,options))
641 #define LA_EMIT_CHAR(ch, type, stream, quoter) \
642 (current_language->la_emitchar(ch, type, stream, quoter))
644 #define LA_PRINT_ARRAY_INDEX(index_type, index_value, stream, options) \
645 (current_language->print_array_index(index_type, index_value, stream, \
648 #define LA_ITERATE_OVER_SYMBOLS(BLOCK, NAME, DOMAIN, CALLBACK) \
649 (current_language->iterate_over_symbols (BLOCK, NAME, DOMAIN, CALLBACK))
651 /* Test a character to decide whether it can be printed in literal form
652 or needs to be printed in another representation. For example,
653 in C the literal form of the character with octal value 141 is 'a'
654 and the "other representation" is '\141'. The "other representation"
655 is program language dependent. */
657 #define PRINT_LITERAL_FORM(c) \
659 && ((c) < 0x7F || (c) >= 0xA0) \
660 && (!sevenbit_strings || (c) < 0x80))
662 /* Type predicates */
664 extern int pointer_type (struct type
*);
666 /* Return true if TYPE is a string type, otherwise return false. This
667 default implementation only detects TYPE_CODE_STRING. */
668 extern bool default_is_string_type_p (struct type
*type
);
672 extern void range_error (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
674 /* Data: Does this value represent "truth" to the current language? */
676 extern int value_true (struct value
*);
678 /* Misc: The string representing a particular enum language. */
680 extern enum language
language_enum (const char *str
);
682 extern const struct language_defn
*language_def (enum language
);
684 extern const char *language_str (enum language
);
686 /* Check for a language-specific trampoline. */
688 extern CORE_ADDR
skip_language_trampoline (struct frame_info
*, CORE_ADDR pc
);
690 /* Return demangled language symbol, or NULL. */
691 extern char *language_demangle (const struct language_defn
*current_language
,
692 const char *mangled
, int options
);
694 /* Splitting strings into words. */
695 extern const char *default_word_break_characters (void);
697 /* Return information about whether TYPE should be passed
698 (and returned) by reference at the language level. */
699 struct language_pass_by_ref_info
language_pass_by_reference (struct type
*type
);
701 /* The default implementation of la_print_typedef. */
702 void default_print_typedef (struct type
*type
, struct symbol
*new_symbol
,
703 struct ui_file
*stream
);
705 void c_get_string (struct value
*value
,
706 gdb::unique_xmalloc_ptr
<gdb_byte
> *buffer
,
707 int *length
, struct type
**char_type
,
708 const char **charset
);
710 /* Get LANG's symbol_name_matcher method for LOOKUP_NAME. Returns
711 default_symbol_name_matcher if not set. LANG is used as a hint;
712 the function may ignore it depending on the current language and
713 LOOKUP_NAME. Specifically, if the current language is Ada, this
714 may return an Ada matcher regardless of LANG. */
715 symbol_name_matcher_ftype
*get_symbol_name_matcher
716 (const language_defn
*lang
, const lookup_name_info
&lookup_name
);
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) */