gdb/language.h

   1 /* Source-language-related definitions for GDB.
   2
   3    Copyright (C) 1991-2017 Free Software Foundation, Inc.
   4
   5    Contributed by the Department of Computer Science at the State University
   6    of New York at Buffalo.
   7
   8    This file is part of GDB.
   9
  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.
  14
  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.
  19
  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/>.  */
  22
  23 #if !defined (LANGUAGE_H)
  24 #define LANGUAGE_H 1
  25
  26 #include "symtab.h"
  27
  28 /* Forward decls for prototypes.  */
  29 struct value;
  30 struct objfile;
  31 struct frame_info;
  32 struct expression;
  33 struct ui_file;
  34 struct value_print_options;
  35 struct type_print_options;
  36 struct lang_varobj_ops;
  37 struct parser_state;
  38 struct compile_instance;
  39
  40 #define MAX_FORTRAN_DIMS  7     /* Maximum number of F77 array dims.  */
  41
  42 /* range_mode ==
  43    range_mode_auto:   range_check set automatically to default of language.
  44    range_mode_manual: range_check set manually by user.  */
  45
  46 extern enum range_mode
  47   {
  48     range_mode_auto, range_mode_manual
  49   }
  50 range_mode;
  51
  52 /* range_check ==
  53    range_check_on:    Ranges are checked in GDB expressions, producing errors.
  54    range_check_warn:  Ranges are checked, producing warnings.
  55    range_check_off:   Ranges are not checked in GDB expressions.  */
  56
  57 extern enum range_check
  58   {
  59     range_check_off, range_check_warn, range_check_on
  60   }
  61 range_check;
  62
  63 /* case_mode ==
  64    case_mode_auto:   case_sensitivity set upon selection of scope.
  65    case_mode_manual: case_sensitivity set only by user.  */
  66
  67 extern enum case_mode
  68   {
  69     case_mode_auto, case_mode_manual
  70   }
  71 case_mode;
  72
  73 /* array_ordering ==
  74    array_row_major:     Arrays are in row major order.
  75    array_column_major:  Arrays are in column major order.  */
  76
  77 extern enum array_ordering
  78   {
  79     array_row_major, array_column_major
  80   }
  81 array_ordering;
  82
  83
  84 /* case_sensitivity ==
  85    case_sensitive_on:   Case sensitivity in name matching is used.
  86    case_sensitive_off:  Case sensitivity in name matching is not used.  */
  87
  88 extern enum case_sensitivity
  89   {
  90     case_sensitive_on, case_sensitive_off
  91   }
  92 case_sensitivity;
  93
  94
  95 /* macro_expansion ==
  96    macro_expansion_no:  No macro expansion is available.
  97    macro_expansion_c:   C-like macro expansion is available.  */
  98
  99 enum macro_expansion
 100   {
 101     macro_expansion_no, macro_expansion_c
 102   };
 103
 104 \f
 105 /* Per architecture (OS/ABI) language information.  */
 106
 107 struct language_arch_info
 108 {
 109   /* Its primitive types.  This is a vector ended by a NULL pointer.
 110      These types can be specified by name in parsing types in
 111      expressions, regardless of whether the program being debugged
 112      actually defines such a type.  */
 113   struct type **primitive_type_vector;
 114
 115   /* Symbol wrappers around primitive_type_vector, so that the symbol lookup
 116      machinery can return them.  */
 117   struct symbol **primitive_type_symbols;
 118
 119   /* Type of elements of strings.  */
 120   struct type *string_char_type;
 121
 122   /* Symbol name of type to use as boolean type, if defined.  */
 123   const char *bool_type_symbol;
 124   /* Otherwise, this is the default boolean builtin type.  */
 125   struct type *bool_type_default;
 126 };
 127
 128 /* A pointer to a function expected to return nonzero if
 129    SYMBOL_SEARCH_NAME matches the given LOOKUP_NAME.
 130
 131    SYMBOL_SEARCH_NAME should be a symbol's "search" name.
 132    LOOKUP_NAME should be the name of an entity after it has been
 133    transformed for lookup.  */
 134
 135 typedef int (*symbol_name_cmp_ftype) (const char *symbol_search_name,
 136                                           const char *lookup_name);
 137
 138 /* Structure tying together assorted information about a language.  */
 139
 140 struct language_defn
 141   {
 142     /* Name of the language.  */
 143
 144     const char *la_name;
 145
 146     /* Natural or official name of the language.  */
 147
 148     const char *la_natural_name;
 149
 150     /* its symtab language-enum (defs.h).  */
 151
 152     enum language la_language;
 153
 154     /* Default range checking.  */
 155
 156     enum range_check la_range_check;
 157
 158     /* Default case sensitivity.  */
 159     enum case_sensitivity la_case_sensitivity;
 160
 161     /* Multi-dimensional array ordering.  */
 162     enum array_ordering la_array_ordering;
 163
 164     /* Style of macro expansion, if any, supported by this language.  */
 165     enum macro_expansion la_macro_expansion;
 166
 167     /* A NULL-terminated array of file extensions for this language.
 168        The extension must include the ".", like ".c".  If this
 169        language doesn't need to provide any filename extensions, this
 170        may be NULL.  */
 171
 172     const char *const *la_filename_extensions;
 173
 174     /* Definitions related to expression printing, prefixifying, and
 175        dumping.  */
 176
 177     const struct exp_descriptor *la_exp_desc;
 178
 179     /* Parser function.  */
 180
 181     int (*la_parser) (struct parser_state *);
 182
 183     /* Parser error function.  */
 184
 185     void (*la_error) (char *);
 186
 187     /* Given an expression *EXPP created by prefixifying the result of
 188        la_parser, perform any remaining processing necessary to complete
 189        its translation.  *EXPP may change; la_post_parser is responsible
 190        for releasing its previous contents, if necessary.  If
 191        VOID_CONTEXT_P, then no value is expected from the expression.  */
 192
 193     void (*la_post_parser) (struct expression ** expp, int void_context_p);
 194
 195     void (*la_printchar) (int ch, struct type *chtype,
 196                           struct ui_file * stream);
 197
 198     void (*la_printstr) (struct ui_file * stream, struct type *elttype,
 199                          const gdb_byte *string, unsigned int length,
 200                          const char *encoding, int force_ellipses,
 201                          const struct value_print_options *);
 202
 203     void (*la_emitchar) (int ch, struct type *chtype,
 204                          struct ui_file * stream, int quoter);
 205
 206     /* Print a type using syntax appropriate for this language.  */
 207
 208     void (*la_print_type) (struct type *, const char *, struct ui_file *, int,
 209                            int, const struct type_print_options *);
 210
 211     /* Print a typedef using syntax appropriate for this language.
 212        TYPE is the underlying type.  NEW_SYMBOL is the symbol naming
 213        the type.  STREAM is the output stream on which to print.  */
 214
 215     void (*la_print_typedef) (struct type *type, struct symbol *new_symbol,
 216                               struct ui_file *stream);
 217
 218     /* Print a value using syntax appropriate for this language.
 219
 220        TYPE is the type of the sub-object to be printed.
 221
 222        EMBEDDED_OFFSET is the offset into the outermost object of the
 223        sub-object represented by TYPE.  This is the object which this
 224        call should print.  Note that the enclosing type is not
 225        available.
 226
 227        ADDRESS is the address in the inferior of the enclosing object.
 228
 229        STREAM is the stream on which the value is to be printed.
 230
 231        RECURSE is the recursion depth.  It is zero-based.
 232
 233        OPTIONS are the formatting options to be used when
 234        printing.  */
 235
 236     void (*la_val_print) (struct type *type,
 237                           int embedded_offset, CORE_ADDR address,
 238                           struct ui_file *stream, int recurse,
 239                           struct value *val,
 240                           const struct value_print_options *options);
 241
 242     /* Print a top-level value using syntax appropriate for this language.  */
 243
 244     void (*la_value_print) (struct value *, struct ui_file *,
 245                             const struct value_print_options *);
 246
 247     /* Given a symbol VAR, the corresponding block VAR_BLOCK (if any) and a
 248        stack frame id FRAME, read the value of the variable and return (pointer
 249        to a) struct value containing the value.
 250
 251        VAR_BLOCK is needed if there's a possibility for VAR to be outside
 252        FRAME.  This is what happens if FRAME correspond to a nested function
 253        and VAR is defined in the outer function.  If callers know that VAR is
 254        located in FRAME or is global/static, NULL can be passed as VAR_BLOCK.
 255
 256        Throw an error if the variable cannot be found.  */
 257
 258     struct value *(*la_read_var_value) (struct symbol *var,
 259                                         const struct block *var_block,
 260                                         struct frame_info *frame);
 261
 262     /* PC is possibly an unknown languages trampoline.
 263        If that PC falls in a trampoline belonging to this language,
 264        return the address of the first pc in the real function, or 0
 265        if it isn't a language tramp for this language.  */
 266     CORE_ADDR (*skip_trampoline) (struct frame_info *, CORE_ADDR);
 267
 268     /* Now come some hooks for lookup_symbol.  */
 269
 270     /* If this is non-NULL, specifies the name that of the implicit
 271        local variable that refers to the current object instance.  */
 272
 273     char *la_name_of_this;
 274
 275     /* This is a function that lookup_symbol will call when it gets to
 276        the part of symbol lookup where C looks up static and global
 277        variables.  */
 278
 279     struct block_symbol (*la_lookup_symbol_nonlocal)
 280       (const struct language_defn *,
 281        const char *,
 282        const struct block *,
 283        const domain_enum);
 284
 285     /* Find the definition of the type with the given name.  */
 286     struct type *(*la_lookup_transparent_type) (const char *);
 287
 288     /* Return demangled language symbol, or NULL.  */
 289     char *(*la_demangle) (const char *mangled, int options);
 290
 291     /* Demangle a symbol according to this language's rules.  Unlike
 292        la_demangle, this does not take any options.
 293
 294        *DEMANGLED will be set by this function.
 295
 296        If this function returns 0, then *DEMANGLED must always be set
 297        to NULL.
 298
 299        If this function returns 1, the implementation may set this to
 300        a xmalloc'd string holding the demangled form.  However, it is
 301        not required to.  The string, if any, is owned by the caller.
 302
 303        The resulting string should be of the form that will be
 304        installed into a symbol.  */
 305     int (*la_sniff_from_mangled_name) (const char *mangled, char **demangled);
 306
 307     /* Return class name of a mangled method name or NULL.  */
 308     char *(*la_class_name_from_physname) (const char *physname);
 309
 310     /* Table for printing expressions.  */
 311
 312     const struct op_print *la_op_print_tab;
 313
 314     /* Zero if the language has first-class arrays.  True if there are no
 315        array values, and array objects decay to pointers, as in C.  */
 316
 317     char c_style_arrays;
 318
 319     /* Index to use for extracting the first element of a string.  */
 320     char string_lower_bound;
 321
 322     /* The list of characters forming word boundaries.  */
 323     char *(*la_word_break_characters) (void);
 324
 325     /* Should return a vector of all symbols which are possible
 326        completions for TEXT.  WORD is the entire command on which the
 327        completion is being made.  If CODE is TYPE_CODE_UNDEF, then all
 328        symbols should be examined; otherwise, only STRUCT_DOMAIN
 329        symbols whose type has a code of CODE should be matched.  */
 330     VEC (char_ptr) *(*la_make_symbol_completion_list) (const char *text,
 331                                                        const char *word,
 332                                                        enum type_code code);
 333
 334     /* The per-architecture (OS/ABI) language information.  */
 335     void (*la_language_arch_info) (struct gdbarch *,
 336                                    struct language_arch_info *);
 337
 338     /* Print the index of an element of an array.  */
 339     void (*la_print_array_index) (struct value *index_value,
 340                                   struct ui_file *stream,
 341                                   const struct value_print_options *options);
 342
 343     /* Return non-zero if TYPE should be passed (and returned) by
 344        reference at the language level.  */
 345     int (*la_pass_by_reference) (struct type *type);
 346
 347     /* Obtain a string from the inferior, storing it in a newly allocated
 348        buffer in BUFFER, which should be freed by the caller.  If the
 349        in- and out-parameter *LENGTH is specified at -1, the string is
 350        read until a null character of the appropriate width is found -
 351        otherwise the string is read to the length of characters specified.
 352        On completion, *LENGTH will hold the size of the string in characters.
 353        If a *LENGTH of -1 was specified it will count only actual
 354        characters, excluding any eventual terminating null character.
 355        Otherwise *LENGTH will include all characters - including any nulls.
 356        CHARSET will hold the encoding used in the string.  */
 357     void (*la_get_string) (struct value *value, gdb_byte **buffer, int *length,
 358                            struct type **chartype, const char **charset);
 359
 360     /* Return a pointer to the function that should be used to match
 361        a symbol name against LOOKUP_NAME. This is mostly for languages
 362        such as Ada where the matching algorithm depends on LOOKUP_NAME.
 363
 364        This field may be NULL, in which case strcmp_iw will be used
 365        to perform the matching.  */
 366     symbol_name_cmp_ftype (*la_get_symbol_name_cmp) (const char *lookup_name);
 367
 368     /* Find all symbols in the current program space matching NAME in
 369        DOMAIN, according to this language's rules.
 370
 371        The search is done in BLOCK only.
 372        The caller is responsible for iterating up through superblocks
 373        if desired.
 374
 375        For each one, call CALLBACK with the symbol and the DATA
 376        argument.  If CALLBACK returns zero, the iteration ends at that
 377        point.
 378
 379        This field may not be NULL.  If the language does not need any
 380        special processing here, 'iterate_over_symbols' should be
 381        used as the definition.  */
 382     void (*la_iterate_over_symbols) (const struct block *block,
 383                                      const char *name,
 384                                      domain_enum domain,
 385                                      symbol_found_callback_ftype *callback,
 386                                      void *data);
 387
 388     /* Various operations on varobj.  */
 389     const struct lang_varobj_ops *la_varobj_ops;
 390
 391     /* If this language allows compilation from the gdb command line,
 392        this method should be non-NULL.  When called it should return
 393        an instance of struct gcc_context appropriate to the language.
 394        When defined this method must never return NULL; instead it
 395        should throw an exception on failure.  The returned compiler
 396        instance is owned by its caller and must be deallocated by
 397        calling its 'destroy' method.  */
 398
 399     struct compile_instance *(*la_get_compile_instance) (void);
 400
 401     /* This method must be defined if 'la_get_gcc_context' is defined.
 402        If 'la_get_gcc_context' is not defined, then this method is
 403        ignored.
 404
 405        This takes the user-supplied text and returns a new bit of code
 406        to compile.
 407
 408        INST is the compiler instance being used.
 409        INPUT is the user's input text.
 410        GDBARCH is the architecture to use.
 411        EXPR_BLOCK is the block in which the expression is being
 412        parsed.
 413        EXPR_PC is the PC at which the expression is being parsed.  */
 414
 415     std::string (*la_compute_program) (struct compile_instance *inst,
 416                                        const char *input,
 417                                        struct gdbarch *gdbarch,
 418                                        const struct block *expr_block,
 419                                        CORE_ADDR expr_pc);
 420
 421     /* Add fields above this point, so the magic number is always last.  */
 422     /* Magic number for compat checking.  */
 423
 424     long la_magic;
 425
 426   };
 427
 428 #define LANG_MAGIC      910823L
 429
 430 /* Pointer to the language_defn for our current language.  This pointer
 431    always points to *some* valid struct; it can be used without checking
 432    it for validity.
 433
 434    The current language affects expression parsing and evaluation
 435    (FIXME: it might be cleaner to make the evaluation-related stuff
 436    separate exp_opcodes for each different set of semantics.  We
 437    should at least think this through more clearly with respect to
 438    what happens if the language is changed between parsing and
 439    evaluation) and printing of things like types and arrays.  It does
 440    *not* affect symbol-reading-- each source file in a symbol-file has
 441    its own language and we should keep track of that regardless of the
 442    language when symbols are read.  If we want some manual setting for
 443    the language of symbol files (e.g. detecting when ".c" files are
 444    C++), it should be a separate setting from the current_language.  */
 445
 446 extern const struct language_defn *current_language;
 447
 448 /* Pointer to the language_defn expected by the user, e.g. the language
 449    of main(), or the language we last mentioned in a message, or C.  */
 450
 451 extern const struct language_defn *expected_language;
 452
 453 /* language_mode ==
 454    language_mode_auto:   current_language automatically set upon selection
 455    of scope (e.g. stack frame)
 456    language_mode_manual: current_language set only by user.  */
 457
 458 extern enum language_mode
 459   {
 460     language_mode_auto, language_mode_manual
 461   }
 462 language_mode;
 463
 464 struct type *language_bool_type (const struct language_defn *l,
 465                                  struct gdbarch *gdbarch);
 466
 467 struct type *language_string_char_type (const struct language_defn *l,
 468                                         struct gdbarch *gdbarch);
 469
 470 /* Look up type NAME in language L, and return its definition for architecture
 471    GDBARCH.  Returns NULL if not found.  */
 472
 473 struct type *language_lookup_primitive_type (const struct language_defn *l,
 474                                              struct gdbarch *gdbarch,
 475                                              const char *name);
 476
 477 /* Wrapper around language_lookup_primitive_type to return the
 478    corresponding symbol.  */
 479
 480 struct symbol *
 481   language_lookup_primitive_type_as_symbol (const struct language_defn *l,
 482                                             struct gdbarch *gdbarch,
 483                                             const char *name);
 484
 485 \f
 486 /* These macros define the behaviour of the expression
 487    evaluator.  */
 488
 489 /* Should we range check values against the domain of their type?  */
 490 #define RANGE_CHECK (range_check != range_check_off)
 491
 492 /* "cast" really means conversion.  */
 493 /* FIXME -- should be a setting in language_defn.  */
 494 #define CAST_IS_CONVERSION(LANG) ((LANG)->la_language == language_c  || \
 495                                   (LANG)->la_language == language_cplus || \
 496                                   (LANG)->la_language == language_objc)
 497
 498 extern void language_info (int);
 499
 500 extern enum language set_language (enum language);
 501 \f
 502
 503 /* This page contains functions that return things that are
 504    specific to languages.  Each of these functions is based on
 505    the current setting of working_lang, which the user sets
 506    with the "set language" command.  */
 507
 508 #define LA_PRINT_TYPE(type,varstring,stream,show,level,flags)           \
 509   (current_language->la_print_type(type,varstring,stream,show,level,flags))
 510
 511 #define LA_PRINT_TYPEDEF(type,new_symbol,stream) \
 512   (current_language->la_print_typedef(type,new_symbol,stream))
 513
 514 #define LA_VALUE_PRINT(val,stream,options) \
 515   (current_language->la_value_print(val,stream,options))
 516
 517 #define LA_PRINT_CHAR(ch, type, stream) \
 518   (current_language->la_printchar(ch, type, stream))
 519 #define LA_PRINT_STRING(stream, elttype, string, length, encoding, force_ellipses, options) \
 520   (current_language->la_printstr(stream, elttype, string, length, \
 521                                  encoding, force_ellipses,options))
 522 #define LA_EMIT_CHAR(ch, type, stream, quoter) \
 523   (current_language->la_emitchar(ch, type, stream, quoter))
 524 #define LA_GET_STRING(value, buffer, length, chartype, encoding) \
 525   (current_language->la_get_string(value, buffer, length, chartype, encoding))
 526
 527 #define LA_PRINT_ARRAY_INDEX(index_value, stream, options) \
 528   (current_language->la_print_array_index(index_value, stream, options))
 529
 530 #define LA_ITERATE_OVER_SYMBOLS(BLOCK, NAME, DOMAIN, CALLBACK, DATA) \
 531   (current_language->la_iterate_over_symbols (BLOCK, NAME, DOMAIN, CALLBACK, \
 532                                               DATA))
 533
 534 /* Test a character to decide whether it can be printed in literal form
 535    or needs to be printed in another representation.  For example,
 536    in C the literal form of the character with octal value 141 is 'a'
 537    and the "other representation" is '\141'.  The "other representation"
 538    is program language dependent.  */
 539
 540 #define PRINT_LITERAL_FORM(c)           \
 541   ((c) >= 0x20                          \
 542    && ((c) < 0x7F || (c) >= 0xA0)       \
 543    && (!sevenbit_strings || (c) < 0x80))
 544
 545 /* Type predicates */
 546
 547 extern int pointer_type (struct type *);
 548
 549 /* Error messages */
 550
 551 extern void range_error (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
 552
 553 /* Data:  Does this value represent "truth" to the current language?  */
 554
 555 extern int value_true (struct value *);
 556
 557 /* Misc:  The string representing a particular enum language.  */
 558
 559 extern enum language language_enum (char *str);
 560
 561 extern const struct language_defn *language_def (enum language);
 562
 563 extern const char *language_str (enum language);
 564
 565 /* Add a language to the set known by GDB (at initialization time).  */
 566
 567 extern void add_language (const struct language_defn *);
 568
 569 /* Check for a language-specific trampoline.  */
 570
 571 extern CORE_ADDR skip_language_trampoline (struct frame_info *, CORE_ADDR pc);
 572
 573 /* Return demangled language symbol, or NULL.  */
 574 extern char *language_demangle (const struct language_defn *current_language,
 575                                 const char *mangled, int options);
 576
 577 /* A wrapper for la_sniff_from_mangled_name.  The arguments and result
 578    are as for the method.  */
 579
 580 extern int language_sniff_from_mangled_name (const struct language_defn *lang,
 581                                              const char *mangled,
 582                                              char **demangled);
 583
 584 /* Return class name from physname, or NULL.  */
 585 extern char *language_class_name_from_physname (const struct language_defn *,
 586                                                 const char *physname);
 587
 588 /* Splitting strings into words.  */
 589 extern char *default_word_break_characters (void);
 590
 591 /* Print the index of an array element using the C99 syntax.  */
 592 extern void default_print_array_index (struct value *index_value,
 593                                        struct ui_file *stream,
 594                                        const struct value_print_options *options);
 595
 596 /* Return non-zero if TYPE should be passed (and returned) by
 597    reference at the language level.  */
 598 int language_pass_by_reference (struct type *type);
 599
 600 /* Return zero; by default, types are passed by value at the language
 601    level.  The target ABI may pass or return some structs by reference
 602    independent of this.  */
 603 int default_pass_by_reference (struct type *type);
 604
 605 /* The default implementation of la_print_typedef.  */
 606 void default_print_typedef (struct type *type, struct symbol *new_symbol,
 607                             struct ui_file *stream);
 608
 609 void default_get_string (struct value *value, gdb_byte **buffer, int *length,
 610                          struct type **char_type, const char **charset);
 611
 612 void c_get_string (struct value *value, gdb_byte **buffer, int *length,
 613                    struct type **char_type, const char **charset);
 614
 615 #endif /* defined (LANGUAGE_H) */