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 /* its symtab language-enum (defs.h). */
193 enum language la_language
;
195 /* Default range checking. */
197 enum range_check la_range_check
;
199 /* Default case sensitivity. */
200 enum case_sensitivity la_case_sensitivity
;
202 /* Multi-dimensional array ordering. */
203 enum array_ordering la_array_ordering
;
205 /* Style of macro expansion, if any, supported by this language. */
206 enum macro_expansion la_macro_expansion
;
208 /* A NULL-terminated array of file extensions for this language.
209 The extension must include the ".", like ".c". If this
210 language doesn't need to provide any filename extensions, this
213 const char *const *la_filename_extensions
;
215 /* Definitions related to expression printing, prefixifying, and
218 const struct exp_descriptor
*la_exp_desc
;
220 /* Now come some hooks for lookup_symbol. */
222 /* True if the symbols names should be stored in GDB's data structures
223 for minimal/partial/full symbols using their linkage (aka mangled)
224 form; false if the symbol names should be demangled first.
226 Most languages implement symbol lookup by comparing the demangled
227 names, in which case it is advantageous to store that information
228 already demangled, and so would set this field to false.
230 On the other hand, some languages have opted for doing symbol
231 lookups by comparing mangled names instead, for reasons usually
232 specific to the language. Those languages should set this field
235 And finally, other languages such as C or Asm do not have
236 the concept of mangled vs demangled name, so those languages
237 should set this field to true as well, to prevent any accidental
238 demangling through an unrelated language's demangler. */
240 const bool la_store_sym_names_in_linkage_form_p
;
242 /* Table for printing expressions. */
244 const struct op_print
*la_op_print_tab
;
246 /* Zero if the language has first-class arrays. True if there are no
247 array values, and array objects decay to pointers, as in C. */
251 /* Index to use for extracting the first element of a string. */
252 char string_lower_bound
;
254 /* Various operations on varobj. */
255 const struct lang_varobj_ops
*la_varobj_ops
;
258 /* Base class from which all other language classes derive. */
260 struct language_defn
: language_data
262 language_defn (enum language lang
, const language_data
&init_data
)
263 : language_data (init_data
)
265 /* We should only ever create one instance of each language. */
266 gdb_assert (languages
[lang
] == nullptr);
267 languages
[lang
] = this;
270 /* Name of the language. */
272 virtual const char *name () const = 0;
274 /* Natural or official name of the language. */
276 virtual const char *natural_name () const = 0;
278 /* Print the index of an element of an array. This default
279 implementation prints using C99 syntax. */
281 virtual void print_array_index (struct type
*index_type
,
283 struct ui_file
*stream
,
284 const value_print_options
*options
) const;
286 /* Given a symbol VAR, the corresponding block VAR_BLOCK (if any) and a
287 stack frame id FRAME, read the value of the variable and return (pointer
288 to a) struct value containing the value.
290 VAR_BLOCK is needed if there's a possibility for VAR to be outside
291 FRAME. This is what happens if FRAME correspond to a nested function
292 and VAR is defined in the outer function. If callers know that VAR is
293 located in FRAME or is global/static, NULL can be passed as VAR_BLOCK.
295 Throw an error if the variable cannot be found. */
297 virtual struct value
*read_var_value (struct symbol
*var
,
298 const struct block
*var_block
,
299 struct frame_info
*frame
) const;
301 /* Return information about whether TYPE should be passed
302 (and returned) by reference at the language level. The default
303 implementation returns a LANGUAGE_PASS_BY_REF_INFO initialised in its
306 virtual struct language_pass_by_ref_info pass_by_reference_info
307 (struct type
*type
) const
312 /* The per-architecture (OS/ABI) language information. */
314 virtual void language_arch_info (struct gdbarch
*,
315 struct language_arch_info
*) const = 0;
317 /* Find the definition of the type with the given name. */
319 virtual struct type
*lookup_transparent_type (const char *name
) const
321 return basic_lookup_transparent_type (name
);
324 /* Find all symbols in the current program space matching NAME in
325 DOMAIN, according to this language's rules.
327 The search is done in BLOCK only.
328 The caller is responsible for iterating up through superblocks
331 For each one, call CALLBACK with the symbol. If CALLBACK
332 returns false, the iteration ends at that point.
334 This field may not be NULL. If the language does not need any
335 special processing here, 'iterate_over_symbols' should be
336 used as the definition. */
337 virtual bool iterate_over_symbols
338 (const struct block
*block
, const lookup_name_info
&name
,
340 gdb::function_view
<symbol_found_callback_ftype
> callback
) const
342 return ::iterate_over_symbols (block
, name
, domain
, callback
);
345 /* Return a pointer to the function that should be used to match a
346 symbol name against LOOKUP_NAME, according to this language's
347 rules. The matching algorithm depends on LOOKUP_NAME. For
348 example, on Ada, the matching algorithm depends on the symbol
349 name (wild/full/verbatim matching), and on whether we're doing
350 a normal lookup or a completion match lookup.
352 As Ada wants to capture symbol matching for all languages in some
353 cases, then this method is a non-overridable interface. Languages
354 should override GET_SYMBOL_NAME_MATCHER_INNER if they need to. */
356 symbol_name_matcher_ftype
*get_symbol_name_matcher
357 (const lookup_name_info
&lookup_name
) const;
359 /* If this language allows compilation from the gdb command line, then
360 this method will return an instance of struct gcc_context appropriate
361 to the language. If compilation for this language is generally
362 supported, but something goes wrong then an exception is thrown. The
363 returned compiler instance is owned by its caller and must be
364 deallocated by the caller. If compilation is not supported for this
365 language then this method returns NULL. */
367 virtual compile_instance
*get_compile_instance () const
372 /* This method must be overridden if 'get_compile_instance' is
375 This takes the user-supplied text and returns a new bit of code
378 INST is the compiler instance being used.
379 INPUT is the user's input text.
380 GDBARCH is the architecture to use.
381 EXPR_BLOCK is the block in which the expression is being
383 EXPR_PC is the PC at which the expression is being parsed. */
385 virtual std::string
compute_program (compile_instance
*inst
,
387 struct gdbarch
*gdbarch
,
388 const struct block
*expr_block
,
389 CORE_ADDR expr_pc
) const
391 gdb_assert_not_reached ("language_defn::compute_program");
394 /* Hash the given symbol search name. */
395 virtual unsigned int search_name_hash (const char *name
) const;
397 /* Demangle a symbol according to this language's rules. Unlike
398 la_demangle, this does not take any options.
400 *DEMANGLED will be set by this function.
402 If this function returns false, then *DEMANGLED must always be set
405 If this function returns true, the implementation may set this to
406 a xmalloc'd string holding the demangled form. However, it is
407 not required to. The string, if any, is owned by the caller.
409 The resulting string should be of the form that will be
410 installed into a symbol. */
411 virtual bool sniff_from_mangled_name (const char *mangled
,
412 char **demangled
) const
414 *demangled
= nullptr;
418 /* Return demangled language symbol version of MANGLED, or NULL. */
419 virtual char *demangle (const char *mangled
, int options
) const
424 /* Print a type using syntax appropriate for this language. */
426 virtual void print_type (struct type
*, const char *, struct ui_file
*, int,
427 int, const struct type_print_options
*) const = 0;
429 /* PC is possibly an unknown languages trampoline.
430 If that PC falls in a trampoline belonging to this language, return
431 the address of the first pc in the real function, or 0 if it isn't a
432 language tramp for this language. */
433 virtual CORE_ADDR
skip_trampoline (struct frame_info
*fi
, CORE_ADDR pc
) const
435 return (CORE_ADDR
) 0;
438 /* Return class name of a mangled method name or NULL. */
439 virtual char *class_name_from_physname (const char *physname
) const
444 /* The list of characters forming word boundaries. */
445 virtual const char *word_break_characters (void) const
447 return default_word_break_characters ();
450 /* Add to the completion tracker all symbols which are possible
451 completions for TEXT. WORD is the entire command on which the
452 completion is being made. If CODE is TYPE_CODE_UNDEF, then all
453 symbols should be examined; otherwise, only STRUCT_DOMAIN symbols
454 whose type has a code of CODE should be matched. */
456 virtual void collect_symbol_completion_matches
457 (completion_tracker
&tracker
,
458 complete_symbol_mode mode
,
459 symbol_name_match_type name_match_type
,
462 enum type_code code
) const
464 return default_collect_symbol_completion_matches_break_on
465 (tracker
, mode
, name_match_type
, text
, word
, "", code
);
468 /* This is a function that lookup_symbol will call when it gets to
469 the part of symbol lookup where C looks up static and global
470 variables. This default implements the basic C lookup rules. */
472 virtual struct block_symbol lookup_symbol_nonlocal
474 const struct block
*block
,
475 const domain_enum domain
) const;
477 /* Return an expression that can be used for a location
478 watchpoint. TYPE is a pointer type that points to the memory
479 to watch, and ADDR is the address of the watched memory. */
480 virtual gdb::unique_xmalloc_ptr
<char> watch_location_expression
481 (struct type
*type
, CORE_ADDR addr
) const;
483 /* List of all known languages. */
484 static const struct language_defn
*languages
[nr_languages
];
486 /* Print a top-level value using syntax appropriate for this language. */
487 virtual void value_print (struct value
*val
, struct ui_file
*stream
,
488 const struct value_print_options
*options
) const;
490 /* Print a value using syntax appropriate for this language. RECURSE is
491 the recursion depth. It is zero-based. */
492 virtual void value_print_inner
493 (struct value
*val
, struct ui_file
*stream
, int recurse
,
494 const struct value_print_options
*options
) const;
496 /* Parser function. */
498 virtual int parser (struct parser_state
*ps
) const;
500 /* Given an expression *EXPP created by prefixifying the result of
501 la_parser, perform any remaining processing necessary to complete its
502 translation. *EXPP may change; la_post_parser is responsible for
503 releasing its previous contents, if necessary. If VOID_CONTEXT_P,
504 then no value is expected from the expression. If COMPLETING is
505 non-zero, then the expression has been parsed for completion, not
508 virtual void post_parser (expression_up
*expp
, int void_context_p
,
510 innermost_block_tracker
*tracker
) const
512 /* By default the post-parser does nothing. */
515 /* Print the character CH (of type CHTYPE) on STREAM as part of the
516 contents of a literal string whose delimiter is QUOTER. */
518 virtual void emitchar (int ch
, struct type
*chtype
,
519 struct ui_file
*stream
, int quoter
) const;
521 virtual void printchar (int ch
, struct type
*chtype
,
522 struct ui_file
* stream
) const;
524 /* Print the character string STRING, printing at most LENGTH characters.
525 Printing stops early if the number hits print_max; repeat counts
526 are printed as appropriate. Print ellipses at the end if we
527 had to stop before printing LENGTH characters, or if FORCE_ELLIPSES. */
529 virtual void printstr (struct ui_file
*stream
, struct type
*elttype
,
530 const gdb_byte
*string
, unsigned int length
,
531 const char *encoding
, int force_ellipses
,
532 const struct value_print_options
*options
) const;
535 /* Print a typedef using syntax appropriate for this language.
536 TYPE is the underlying type. NEW_SYMBOL is the symbol naming
537 the type. STREAM is the output stream on which to print. */
539 virtual void print_typedef (struct type
*type
, struct symbol
*new_symbol
,
540 struct ui_file
*stream
) const;
542 /* Return true if TYPE is a string type. */
543 virtual bool is_string_type_p (struct type
*type
) const;
545 /* Return a string that is used by the 'set print max-depth' setting.
546 When GDB replaces a struct or union (during value printing) that is
547 "too deep" this string is displayed instead. The default value here
548 suits most languages. If overriding then the string here should
549 ideally be similar in style to the default; an opener, three '.', and
552 virtual const char *struct_too_deep_ellipsis () const
555 /* If this returns non-NULL then the string returned specifies the name
556 of the implicit local variable that refers to the current object
557 instance. Return NULL (the default) for languages that have no name
558 for the current object instance. */
560 virtual const char *name_of_this () const
565 /* This is the overridable part of the GET_SYMBOL_NAME_MATCHER method.
566 See that method for a description of the arguments. */
568 virtual symbol_name_matcher_ftype
*get_symbol_name_matcher_inner
569 (const lookup_name_info
&lookup_name
) const;
572 /* Pointer to the language_defn for our current language. This pointer
573 always points to *some* valid struct; it can be used without checking
576 The current language affects expression parsing and evaluation
577 (FIXME: it might be cleaner to make the evaluation-related stuff
578 separate exp_opcodes for each different set of semantics. We
579 should at least think this through more clearly with respect to
580 what happens if the language is changed between parsing and
581 evaluation) and printing of things like types and arrays. It does
582 *not* affect symbol-reading-- each source file in a symbol-file has
583 its own language and we should keep track of that regardless of the
584 language when symbols are read. If we want some manual setting for
585 the language of symbol files (e.g. detecting when ".c" files are
586 C++), it should be a separate setting from the current_language. */
588 extern const struct language_defn
*current_language
;
590 /* Pointer to the language_defn expected by the user, e.g. the language
591 of main(), or the language we last mentioned in a message, or C. */
593 extern const struct language_defn
*expected_language
;
595 /* Warning issued when current_language and the language of the current
596 frame do not match. */
598 extern const char lang_frame_mismatch_warn
[];
601 language_mode_auto: current_language automatically set upon selection
602 of scope (e.g. stack frame)
603 language_mode_manual: current_language set only by user. */
605 extern enum language_mode
607 language_mode_auto
, language_mode_manual
611 struct type
*language_bool_type (const struct language_defn
*l
,
612 struct gdbarch
*gdbarch
);
614 struct type
*language_string_char_type (const struct language_defn
*l
,
615 struct gdbarch
*gdbarch
);
617 /* Look up type NAME in language L, and return its definition for architecture
618 GDBARCH. Returns NULL if not found. */
620 struct type
*language_lookup_primitive_type (const struct language_defn
*l
,
621 struct gdbarch
*gdbarch
,
624 /* Wrapper around language_lookup_primitive_type to return the
625 corresponding symbol. */
628 language_lookup_primitive_type_as_symbol (const struct language_defn
*l
,
629 struct gdbarch
*gdbarch
,
633 /* These macros define the behaviour of the expression
636 /* Should we range check values against the domain of their type? */
637 #define RANGE_CHECK (range_check != range_check_off)
639 /* "cast" really means conversion. */
640 /* FIXME -- should be a setting in language_defn. */
641 #define CAST_IS_CONVERSION(LANG) ((LANG)->la_language == language_c || \
642 (LANG)->la_language == language_cplus || \
643 (LANG)->la_language == language_objc)
645 extern void language_info (int);
647 extern enum language
set_language (enum language
);
650 /* This page contains functions that return things that are
651 specific to languages. Each of these functions is based on
652 the current setting of working_lang, which the user sets
653 with the "set language" command. */
655 #define LA_PRINT_TYPE(type,varstring,stream,show,level,flags) \
656 (current_language->print_type(type,varstring,stream,show,level,flags))
658 #define LA_PRINT_TYPEDEF(type,new_symbol,stream) \
659 (current_language->print_typedef (type,new_symbol,stream))
661 #define LA_VALUE_PRINT(val,stream,options) \
662 (current_language->value_print (val,stream,options))
664 #define LA_PRINT_CHAR(ch, type, stream) \
665 (current_language->printchar (ch, type, stream))
666 #define LA_PRINT_STRING(stream, elttype, string, length, encoding, force_ellipses, options) \
667 (current_language->printstr (stream, elttype, string, length, \
668 encoding, force_ellipses,options))
669 #define LA_EMIT_CHAR(ch, type, stream, quoter) \
670 (current_language->emitchar (ch, type, stream, quoter))
672 #define LA_PRINT_ARRAY_INDEX(index_type, index_value, stream, options) \
673 (current_language->print_array_index(index_type, index_value, stream, \
676 #define LA_ITERATE_OVER_SYMBOLS(BLOCK, NAME, DOMAIN, CALLBACK) \
677 (current_language->iterate_over_symbols (BLOCK, NAME, DOMAIN, CALLBACK))
679 /* Test a character to decide whether it can be printed in literal form
680 or needs to be printed in another representation. For example,
681 in C the literal form of the character with octal value 141 is 'a'
682 and the "other representation" is '\141'. The "other representation"
683 is program language dependent. */
685 #define PRINT_LITERAL_FORM(c) \
687 && ((c) < 0x7F || (c) >= 0xA0) \
688 && (!sevenbit_strings || (c) < 0x80))
690 /* Type predicates */
692 extern int pointer_type (struct type
*);
696 extern void range_error (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
698 /* Data: Does this value represent "truth" to the current language? */
700 extern int value_true (struct value
*);
702 /* Misc: The string representing a particular enum language. */
704 extern enum language
language_enum (const char *str
);
706 extern const struct language_defn
*language_def (enum language
);
708 extern const char *language_str (enum language
);
710 /* Check for a language-specific trampoline. */
712 extern CORE_ADDR
skip_language_trampoline (struct frame_info
*, CORE_ADDR pc
);
714 /* Return demangled language symbol, or NULL. */
715 extern char *language_demangle (const struct language_defn
*current_language
,
716 const char *mangled
, int options
);
718 /* Return information about whether TYPE should be passed
719 (and returned) by reference at the language level. */
720 struct language_pass_by_ref_info
language_pass_by_reference (struct type
*type
);
722 void c_get_string (struct value
*value
,
723 gdb::unique_xmalloc_ptr
<gdb_byte
> *buffer
,
724 int *length
, struct type
**char_type
,
725 const char **charset
);
727 /* Get LANG's symbol_name_matcher method for LOOKUP_NAME. Returns
728 default_symbol_name_matcher if not set. LANG is used as a hint;
729 the function may ignore it depending on the current language and
730 LOOKUP_NAME. Specifically, if the current language is Ada, this
731 may return an Ada matcher regardless of LANG. */
732 symbol_name_matcher_ftype
*get_symbol_name_matcher
733 (const language_defn
*lang
, const lookup_name_info
&lookup_name
);
735 /* Save the current language and restore it upon destruction. */
737 class scoped_restore_current_language
741 explicit scoped_restore_current_language ()
742 : m_lang (current_language
->la_language
)
746 ~scoped_restore_current_language ()
748 set_language (m_lang
);
751 scoped_restore_current_language (const scoped_restore_current_language
&)
753 scoped_restore_current_language
&operator=
754 (const scoped_restore_current_language
&) = delete;
758 enum language m_lang
;
761 /* If language_mode is language_mode_auto,
762 then switch current language to the language of SYM
763 and restore current language upon destruction.
767 class scoped_switch_to_sym_language_if_auto
771 explicit scoped_switch_to_sym_language_if_auto (const struct symbol
*sym
)
773 if (language_mode
== language_mode_auto
)
775 m_lang
= current_language
->la_language
;
777 set_language (sym
->language ());
782 /* Assign to m_lang to silence a GCC warning. See
783 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80635. */
784 m_lang
= language_unknown
;
788 ~scoped_switch_to_sym_language_if_auto ()
791 set_language (m_lang
);
794 DISABLE_COPY_AND_ASSIGN (scoped_switch_to_sym_language_if_auto
);
798 enum language m_lang
;
801 #endif /* defined (LANGUAGE_H) */