1 /* Source-language-related definitions for GDB.
3 Copyright (C) 1991-2020 Free Software Foundation, Inc.
5 Contributed by the Department of Computer Science at the State University
6 of New York at Buffalo.
8 This file is part of GDB.
10 This program is free software; you can redistribute it and/or modify
11 it under the terms of the GNU General Public License as published by
12 the Free Software Foundation; either version 3 of the License, or
13 (at your option) any later version.
15 This program is distributed in the hope that it will be useful,
16 but WITHOUT ANY WARRANTY; without even the implied warranty of
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 GNU General Public License for more details.
20 You should have received a copy of the GNU General Public License
21 along with this program. If not, see <http://www.gnu.org/licenses/>. */
23 #if !defined (LANGUAGE_H)
27 #include "gdbsupport/function-view.h"
28 #include "expression.h"
30 /* Forward decls for prototypes. */
35 struct value_print_options
;
36 struct type_print_options
;
37 struct lang_varobj_ops
;
39 class compile_instance
;
40 struct completion_match_for_lcd
;
41 class innermost_block_tracker
;
43 #define MAX_FORTRAN_DIMS 7 /* Maximum number of F77 array dims. */
46 range_mode_auto: range_check set automatically to default of language.
47 range_mode_manual: range_check set manually by user. */
49 extern enum range_mode
51 range_mode_auto
, range_mode_manual
56 range_check_on: Ranges are checked in GDB expressions, producing errors.
57 range_check_warn: Ranges are checked, producing warnings.
58 range_check_off: Ranges are not checked in GDB expressions. */
60 extern enum range_check
62 range_check_off
, range_check_warn
, range_check_on
67 case_mode_auto: case_sensitivity set upon selection of scope.
68 case_mode_manual: case_sensitivity set only by user. */
72 case_mode_auto
, case_mode_manual
77 array_row_major: Arrays are in row major order.
78 array_column_major: Arrays are in column major order. */
80 extern enum array_ordering
82 array_row_major
, array_column_major
87 /* case_sensitivity ==
88 case_sensitive_on: Case sensitivity in name matching is used.
89 case_sensitive_off: Case sensitivity in name matching is not used. */
91 extern enum case_sensitivity
93 case_sensitive_on
, case_sensitive_off
99 macro_expansion_no: No macro expansion is available.
100 macro_expansion_c: C-like macro expansion is available. */
104 macro_expansion_no
, macro_expansion_c
108 /* Per architecture (OS/ABI) language information. */
110 struct language_arch_info
112 /* Its primitive types. This is a vector ended by a NULL pointer.
113 These types can be specified by name in parsing types in
114 expressions, regardless of whether the program being debugged
115 actually defines such a type. */
116 struct type
**primitive_type_vector
;
118 /* Symbol wrappers around primitive_type_vector, so that the symbol lookup
119 machinery can return them. */
120 struct symbol
**primitive_type_symbols
;
122 /* Type of elements of strings. */
123 struct type
*string_char_type
;
125 /* Symbol name of type to use as boolean type, if defined. */
126 const char *bool_type_symbol
;
127 /* Otherwise, this is the default boolean builtin type. */
128 struct type
*bool_type_default
;
131 /* In a language (particularly C++) a function argument of an aggregate
132 type (i.e. class/struct/union) may be implicitly passed by reference
133 even though it is declared a call-by-value argument in the source.
134 The struct below puts together necessary information for GDB to be
135 able to detect and carry out pass-by-reference semantics for a
136 particular type. This type is referred as T in the inlined comments
139 The default values of the fields are chosen to give correct semantics
140 for primitive types and for simple aggregate types, such as
146 struct language_pass_by_ref_info
148 /* True if an argument of type T can be passed to a function by value
149 (i.e. not through an implicit reference). False, otherwise. */
150 bool trivially_copyable
= true;
152 /* True if a copy of a value of type T can be initialized by
153 memcpy'ing the value bit-by-bit. False, otherwise.
154 E.g. If T has a user-defined copy ctor, this should be false. */
155 bool trivially_copy_constructible
= true;
157 /* True if a value of type T can be destructed simply by reclaiming
158 the memory area occupied by the value. False, otherwise.
159 E.g. If T has a user-defined destructor, this should be false. */
160 bool trivially_destructible
= true;
162 /* True if it is allowed to create a copy of a value of type T.
164 E.g. If T has a deleted copy ctor, this should be false. */
165 bool copy_constructible
= true;
167 /* True if a value of type T can be destructed. False, otherwise.
168 E.g. If T has a deleted destructor, this should be false. */
169 bool destructible
= true;
172 /* Structure tying together assorted information about a language. */
176 /* Name of the language. */
180 /* Natural or official name of the language. */
182 const char *la_natural_name
;
184 /* its symtab language-enum (defs.h). */
186 enum language la_language
;
188 /* Default range checking. */
190 enum range_check la_range_check
;
192 /* Default case sensitivity. */
193 enum case_sensitivity la_case_sensitivity
;
195 /* Multi-dimensional array ordering. */
196 enum array_ordering la_array_ordering
;
198 /* Style of macro expansion, if any, supported by this language. */
199 enum macro_expansion la_macro_expansion
;
201 /* A NULL-terminated array of file extensions for this language.
202 The extension must include the ".", like ".c". If this
203 language doesn't need to provide any filename extensions, this
206 const char *const *la_filename_extensions
;
208 /* Definitions related to expression printing, prefixifying, and
211 const struct exp_descriptor
*la_exp_desc
;
213 /* Parser function. */
215 int (*la_parser
) (struct parser_state
*);
217 /* Given an expression *EXPP created by prefixifying the result of
218 la_parser, perform any remaining processing necessary to complete
219 its translation. *EXPP may change; la_post_parser is responsible
220 for releasing its previous contents, if necessary. If
221 VOID_CONTEXT_P, then no value is expected from the expression.
222 If COMPLETING is non-zero, then the expression has been parsed
223 for completion, not evaluation. */
225 void (*la_post_parser
) (expression_up
*expp
, int void_context_p
,
226 int completing
, innermost_block_tracker
*tracker
);
228 void (*la_printchar
) (int ch
, struct type
*chtype
,
229 struct ui_file
* stream
);
231 void (*la_printstr
) (struct ui_file
* stream
, struct type
*elttype
,
232 const gdb_byte
*string
, unsigned int length
,
233 const char *encoding
, int force_ellipses
,
234 const struct value_print_options
*);
236 void (*la_emitchar
) (int ch
, struct type
*chtype
,
237 struct ui_file
* stream
, int quoter
);
239 /* Print a type using syntax appropriate for this language. */
241 void (*la_print_type
) (struct type
*, const char *, struct ui_file
*, int,
242 int, const struct type_print_options
*);
244 /* Print a typedef using syntax appropriate for this language.
245 TYPE is the underlying type. NEW_SYMBOL is the symbol naming
246 the type. STREAM is the output stream on which to print. */
248 void (*la_print_typedef
) (struct type
*type
, struct symbol
*new_symbol
,
249 struct ui_file
*stream
);
251 /* Print a value using syntax appropriate for this language.
252 RECURSE is the recursion depth. It is zero-based. */
254 void (*la_value_print_inner
) (struct value
*, struct ui_file
*,
256 const struct value_print_options
*);
258 /* Print a top-level value using syntax appropriate for this language. */
260 void (*la_value_print
) (struct value
*, struct ui_file
*,
261 const struct value_print_options
*);
263 /* Given a symbol VAR, the corresponding block VAR_BLOCK (if any) and a
264 stack frame id FRAME, read the value of the variable and return (pointer
265 to a) struct value containing the value.
267 VAR_BLOCK is needed if there's a possibility for VAR to be outside
268 FRAME. This is what happens if FRAME correspond to a nested function
269 and VAR is defined in the outer function. If callers know that VAR is
270 located in FRAME or is global/static, NULL can be passed as VAR_BLOCK.
272 Throw an error if the variable cannot be found. */
274 struct value
*(*la_read_var_value
) (struct symbol
*var
,
275 const struct block
*var_block
,
276 struct frame_info
*frame
);
278 /* PC is possibly an unknown languages trampoline.
279 If that PC falls in a trampoline belonging to this language,
280 return the address of the first pc in the real function, or 0
281 if it isn't a language tramp for this language. */
282 CORE_ADDR (*skip_trampoline
) (struct frame_info
*, CORE_ADDR
);
284 /* Now come some hooks for lookup_symbol. */
286 /* If this is non-NULL, specifies the name that of the implicit
287 local variable that refers to the current object instance. */
289 const char *la_name_of_this
;
291 /* True if the symbols names should be stored in GDB's data structures
292 for minimal/partial/full symbols using their linkage (aka mangled)
293 form; false if the symbol names should be demangled first.
295 Most languages implement symbol lookup by comparing the demangled
296 names, in which case it is advantageous to store that information
297 already demangled, and so would set this field to false.
299 On the other hand, some languages have opted for doing symbol
300 lookups by comparing mangled names instead, for reasons usually
301 specific to the language. Those languages should set this field
304 And finally, other languages such as C or Asm do not have
305 the concept of mangled vs demangled name, so those languages
306 should set this field to true as well, to prevent any accidental
307 demangling through an unrelated language's demangler. */
309 const bool la_store_sym_names_in_linkage_form_p
;
311 /* This is a function that lookup_symbol will call when it gets to
312 the part of symbol lookup where C looks up static and global
315 struct block_symbol (*la_lookup_symbol_nonlocal
)
316 (const struct language_defn
*,
318 const struct block
*,
321 /* Find the definition of the type with the given name. */
322 struct type
*(*la_lookup_transparent_type
) (const char *);
324 /* Return demangled language symbol, or NULL. */
325 char *(*la_demangle
) (const char *mangled
, int options
);
327 /* Demangle a symbol according to this language's rules. Unlike
328 la_demangle, this does not take any options.
330 *DEMANGLED will be set by this function.
332 If this function returns 0, then *DEMANGLED must always be set
335 If this function returns 1, the implementation may set this to
336 a xmalloc'd string holding the demangled form. However, it is
337 not required to. The string, if any, is owned by the caller.
339 The resulting string should be of the form that will be
340 installed into a symbol. */
341 int (*la_sniff_from_mangled_name
) (const char *mangled
, char **demangled
);
343 /* Return class name of a mangled method name or NULL. */
344 char *(*la_class_name_from_physname
) (const char *physname
);
346 /* Table for printing expressions. */
348 const struct op_print
*la_op_print_tab
;
350 /* Zero if the language has first-class arrays. True if there are no
351 array values, and array objects decay to pointers, as in C. */
355 /* Index to use for extracting the first element of a string. */
356 char string_lower_bound
;
358 /* The list of characters forming word boundaries. */
359 const char *(*la_word_break_characters
) (void);
361 /* Add to the completion tracker all symbols which are possible
362 completions for TEXT. WORD is the entire command on which the
363 completion is being made. If CODE is TYPE_CODE_UNDEF, then all
364 symbols should be examined; otherwise, only STRUCT_DOMAIN
365 symbols whose type has a code of CODE should be matched. */
366 void (*la_collect_symbol_completion_matches
)
367 (completion_tracker
&tracker
,
368 complete_symbol_mode mode
,
369 symbol_name_match_type match_type
,
372 enum type_code code
);
374 /* The per-architecture (OS/ABI) language information. */
375 void (*la_language_arch_info
) (struct gdbarch
*,
376 struct language_arch_info
*);
378 /* Print the index of an element of an array. */
379 void (*la_print_array_index
) (struct value
*index_value
,
380 struct ui_file
*stream
,
381 const struct value_print_options
*options
);
383 /* Return information about whether TYPE should be passed
384 (and returned) by reference at the language level. */
385 struct language_pass_by_ref_info (*la_pass_by_reference
)
388 /* Return an expression that can be used for a location
389 watchpoint. TYPE is a pointer type that points to the memory
390 to watch, and ADDR is the address of the watched memory. */
391 gdb::unique_xmalloc_ptr
<char> (*la_watch_location_expression
)
392 (struct type
*type
, CORE_ADDR addr
);
394 /* Return a pointer to the function that should be used to match a
395 symbol name against LOOKUP_NAME, according to this language's
396 rules. The matching algorithm depends on LOOKUP_NAME. For
397 example, on Ada, the matching algorithm depends on the symbol
398 name (wild/full/verbatim matching), and on whether we're doing
399 a normal lookup or a completion match lookup.
401 This field may be NULL, in which case
402 default_symbol_name_matcher is used to perform the
404 symbol_name_matcher_ftype
*(*la_get_symbol_name_matcher
)
405 (const lookup_name_info
&);
407 /* Find all symbols in the current program space matching NAME in
408 DOMAIN, according to this language's rules.
410 The search is done in BLOCK only.
411 The caller is responsible for iterating up through superblocks
414 For each one, call CALLBACK with the symbol. If CALLBACK
415 returns false, the iteration ends at that point.
417 This field may not be NULL. If the language does not need any
418 special processing here, 'iterate_over_symbols' should be
419 used as the definition. */
420 bool (*la_iterate_over_symbols
)
421 (const struct block
*block
, const lookup_name_info
&name
,
423 gdb::function_view
<symbol_found_callback_ftype
> callback
);
425 /* Hash the given symbol search name. Use
426 default_search_name_hash if no special treatment is
428 unsigned int (*la_search_name_hash
) (const char *name
);
430 /* Various operations on varobj. */
431 const struct lang_varobj_ops
*la_varobj_ops
;
433 /* If this language allows compilation from the gdb command line,
434 this method should be non-NULL. When called it should return
435 an instance of struct gcc_context appropriate to the language.
436 When defined this method must never return NULL; instead it
437 should throw an exception on failure. The returned compiler
438 instance is owned by its caller and must be deallocated by
439 calling its 'destroy' method. */
441 compile_instance
*(*la_get_compile_instance
) (void);
443 /* This method must be defined if 'la_get_gcc_context' is defined.
444 If 'la_get_gcc_context' is not defined, then this method is
447 This takes the user-supplied text and returns a new bit of code
450 INST is the compiler instance being used.
451 INPUT is the user's input text.
452 GDBARCH is the architecture to use.
453 EXPR_BLOCK is the block in which the expression is being
455 EXPR_PC is the PC at which the expression is being parsed. */
457 std::string (*la_compute_program
) (compile_instance
*inst
,
459 struct gdbarch
*gdbarch
,
460 const struct block
*expr_block
,
463 /* Return true if TYPE is a string type. */
464 bool (*la_is_string_type_p
) (struct type
*type
);
466 /* This string is used by the 'set print max-depth' setting. When GDB
467 replaces a struct or union (during value printing) that is "too
468 deep" this string is displayed instead. */
469 const char *la_struct_too_deep_ellipsis
;
473 /* Pointer to the language_defn for our current language. This pointer
474 always points to *some* valid struct; it can be used without checking
477 The current language affects expression parsing and evaluation
478 (FIXME: it might be cleaner to make the evaluation-related stuff
479 separate exp_opcodes for each different set of semantics. We
480 should at least think this through more clearly with respect to
481 what happens if the language is changed between parsing and
482 evaluation) and printing of things like types and arrays. It does
483 *not* affect symbol-reading-- each source file in a symbol-file has
484 its own language and we should keep track of that regardless of the
485 language when symbols are read. If we want some manual setting for
486 the language of symbol files (e.g. detecting when ".c" files are
487 C++), it should be a separate setting from the current_language. */
489 extern const struct language_defn
*current_language
;
491 /* Pointer to the language_defn expected by the user, e.g. the language
492 of main(), or the language we last mentioned in a message, or C. */
494 extern const struct language_defn
*expected_language
;
496 /* Warning issued when current_language and the language of the current
497 frame do not match. */
499 extern const char lang_frame_mismatch_warn
[];
502 language_mode_auto: current_language automatically set upon selection
503 of scope (e.g. stack frame)
504 language_mode_manual: current_language set only by user. */
506 extern enum language_mode
508 language_mode_auto
, language_mode_manual
512 struct type
*language_bool_type (const struct language_defn
*l
,
513 struct gdbarch
*gdbarch
);
515 struct type
*language_string_char_type (const struct language_defn
*l
,
516 struct gdbarch
*gdbarch
);
518 /* Look up type NAME in language L, and return its definition for architecture
519 GDBARCH. Returns NULL if not found. */
521 struct type
*language_lookup_primitive_type (const struct language_defn
*l
,
522 struct gdbarch
*gdbarch
,
525 /* Wrapper around language_lookup_primitive_type to return the
526 corresponding symbol. */
529 language_lookup_primitive_type_as_symbol (const struct language_defn
*l
,
530 struct gdbarch
*gdbarch
,
534 /* These macros define the behaviour of the expression
537 /* Should we range check values against the domain of their type? */
538 #define RANGE_CHECK (range_check != range_check_off)
540 /* "cast" really means conversion. */
541 /* FIXME -- should be a setting in language_defn. */
542 #define CAST_IS_CONVERSION(LANG) ((LANG)->la_language == language_c || \
543 (LANG)->la_language == language_cplus || \
544 (LANG)->la_language == language_objc)
546 extern void language_info (int);
548 extern enum language
set_language (enum language
);
551 /* This page contains functions that return things that are
552 specific to languages. Each of these functions is based on
553 the current setting of working_lang, which the user sets
554 with the "set language" command. */
556 #define LA_PRINT_TYPE(type,varstring,stream,show,level,flags) \
557 (current_language->la_print_type(type,varstring,stream,show,level,flags))
559 #define LA_PRINT_TYPEDEF(type,new_symbol,stream) \
560 (current_language->la_print_typedef(type,new_symbol,stream))
562 #define LA_VALUE_PRINT(val,stream,options) \
563 (current_language->la_value_print(val,stream,options))
565 #define LA_PRINT_CHAR(ch, type, stream) \
566 (current_language->la_printchar(ch, type, stream))
567 #define LA_PRINT_STRING(stream, elttype, string, length, encoding, force_ellipses, options) \
568 (current_language->la_printstr(stream, elttype, string, length, \
569 encoding, force_ellipses,options))
570 #define LA_EMIT_CHAR(ch, type, stream, quoter) \
571 (current_language->la_emitchar(ch, type, stream, quoter))
573 #define LA_PRINT_ARRAY_INDEX(index_value, stream, options) \
574 (current_language->la_print_array_index(index_value, stream, options))
576 #define LA_ITERATE_OVER_SYMBOLS(BLOCK, NAME, DOMAIN, CALLBACK) \
577 (current_language->la_iterate_over_symbols (BLOCK, NAME, DOMAIN, CALLBACK))
579 /* Test a character to decide whether it can be printed in literal form
580 or needs to be printed in another representation. For example,
581 in C the literal form of the character with octal value 141 is 'a'
582 and the "other representation" is '\141'. The "other representation"
583 is program language dependent. */
585 #define PRINT_LITERAL_FORM(c) \
587 && ((c) < 0x7F || (c) >= 0xA0) \
588 && (!sevenbit_strings || (c) < 0x80))
590 /* Type predicates */
592 extern int pointer_type (struct type
*);
594 /* Return true if TYPE is a string type, otherwise return false. This
595 default implementation only detects TYPE_CODE_STRING. */
596 extern bool default_is_string_type_p (struct type
*type
);
600 extern void range_error (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
602 /* Data: Does this value represent "truth" to the current language? */
604 extern int value_true (struct value
*);
606 /* Misc: The string representing a particular enum language. */
608 extern enum language
language_enum (const char *str
);
610 extern const struct language_defn
*language_def (enum language
);
612 extern const char *language_str (enum language
);
614 /* Check for a language-specific trampoline. */
616 extern CORE_ADDR
skip_language_trampoline (struct frame_info
*, CORE_ADDR pc
);
618 /* Return demangled language symbol, or NULL. */
619 extern char *language_demangle (const struct language_defn
*current_language
,
620 const char *mangled
, int options
);
622 /* A wrapper for la_sniff_from_mangled_name. The arguments and result
623 are as for the method. */
625 extern int language_sniff_from_mangled_name (const struct language_defn
*lang
,
629 /* Return class name from physname, or NULL. */
630 extern char *language_class_name_from_physname (const struct language_defn
*,
631 const char *physname
);
633 /* Splitting strings into words. */
634 extern const char *default_word_break_characters (void);
636 /* Print the index of an array element using the C99 syntax. */
637 extern void default_print_array_index (struct value
*index_value
,
638 struct ui_file
*stream
,
639 const struct value_print_options
*options
);
641 /* Return information about whether TYPE should be passed
642 (and returned) by reference at the language level. */
643 struct language_pass_by_ref_info
language_pass_by_reference (struct type
*type
);
645 /* Return a default struct that provides pass-by-reference information
646 about the given TYPE. Languages should update the default values
648 struct language_pass_by_ref_info
default_pass_by_reference (struct type
*type
);
650 /* The default implementation of la_print_typedef. */
651 void default_print_typedef (struct type
*type
, struct symbol
*new_symbol
,
652 struct ui_file
*stream
);
654 /* Default name hashing function. */
656 /* Produce an unsigned hash value from SEARCH_NAME that is consistent
657 with strcmp_iw, strcmp, and, at least on Ada symbols, wild_match.
658 That is, two identifiers equivalent according to any of those three
659 comparison operators hash to the same value. */
660 extern unsigned int default_search_name_hash (const char *search_name
);
662 void c_get_string (struct value
*value
,
663 gdb::unique_xmalloc_ptr
<gdb_byte
> *buffer
,
664 int *length
, struct type
**char_type
,
665 const char **charset
);
667 /* The default implementation of la_symbol_name_matcher. Matches with
669 extern bool default_symbol_name_matcher
670 (const char *symbol_search_name
,
671 const lookup_name_info
&lookup_name
,
672 completion_match_result
*comp_match_res
);
674 /* Get LANG's symbol_name_matcher method for LOOKUP_NAME. Returns
675 default_symbol_name_matcher if not set. LANG is used as a hint;
676 the function may ignore it depending on the current language and
677 LOOKUP_NAME. Specifically, if the current language is Ada, this
678 may return an Ada matcher regardless of LANG. */
679 symbol_name_matcher_ftype
*get_symbol_name_matcher
680 (const language_defn
*lang
, const lookup_name_info
&lookup_name
);
682 /* The languages supported by GDB. */
684 extern const struct language_defn auto_language_defn
;
685 extern const struct language_defn unknown_language_defn
;
686 extern const struct language_defn minimal_language_defn
;
688 extern const struct language_defn ada_language_defn
;
689 extern const struct language_defn asm_language_defn
;
690 extern const struct language_defn c_language_defn
;
691 extern const struct language_defn cplus_language_defn
;
692 extern const struct language_defn d_language_defn
;
693 extern const struct language_defn f_language_defn
;
694 extern const struct language_defn go_language_defn
;
695 extern const struct language_defn m2_language_defn
;
696 extern const struct language_defn objc_language_defn
;
697 extern const struct language_defn opencl_language_defn
;
698 extern const struct language_defn pascal_language_defn
;
699 extern const struct language_defn rust_language_defn
;
701 /* Save the current language and restore it upon destruction. */
703 class scoped_restore_current_language
707 explicit scoped_restore_current_language ()
708 : m_lang (current_language
->la_language
)
712 ~scoped_restore_current_language ()
714 set_language (m_lang
);
717 scoped_restore_current_language (const scoped_restore_current_language
&)
719 scoped_restore_current_language
&operator=
720 (const scoped_restore_current_language
&) = delete;
724 enum language m_lang
;
727 /* If language_mode is language_mode_auto,
728 then switch current language to the language of SYM
729 and restore current language upon destruction.
733 class scoped_switch_to_sym_language_if_auto
737 explicit scoped_switch_to_sym_language_if_auto (const struct symbol
*sym
)
739 if (language_mode
== language_mode_auto
)
741 m_lang
= current_language
->la_language
;
743 set_language (sym
->language ());
748 /* Assign to m_lang to silence a GCC warning. See
749 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80635. */
750 m_lang
= language_unknown
;
754 ~scoped_switch_to_sym_language_if_auto ()
757 set_language (m_lang
);
760 DISABLE_COPY_AND_ASSIGN (scoped_switch_to_sym_language_if_auto
);
764 enum language m_lang
;
767 #endif /* defined (LANGUAGE_H) */