gdb/language.h

   1 /* Source-language-related definitions for GDB.
   2
   3    Copyright (C) 1991-2016 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        CONTENTS holds the bits of the value.  This holds the entire
 223        enclosing object.
 224
 225        EMBEDDED_OFFSET is the offset into the outermost object of the
 226        sub-object represented by TYPE.  This is the object which this
 227        call should print.  Note that the enclosing type is not
 228        available.
 229
 230        ADDRESS is the address in the inferior of the enclosing object.
 231
 232        STREAM is the stream on which the value is to be printed.
 233
 234        RECURSE is the recursion depth.  It is zero-based.
 235
 236        OPTIONS are the formatting options to be used when
 237        printing.  */
 238
 239     void (*la_val_print) (struct type *type,
 240                           const gdb_byte *contents,
 241                           int embedded_offset, CORE_ADDR address,
 242                           struct ui_file *stream, int recurse,
 243                           const struct value *val,
 244                           const struct value_print_options *options);
 245
 246     /* Print a top-level value using syntax appropriate for this language.  */
 247
 248     void (*la_value_print) (struct value *, struct ui_file *,
 249                             const struct value_print_options *);
 250
 251     /* Given a symbol VAR, the corresponding block VAR_BLOCK (if any) and a
 252        stack frame id FRAME, read the value of the variable and return (pointer
 253        to a) struct value containing the value.
 254
 255        VAR_BLOCK is needed if there's a possibility for VAR to be outside
 256        FRAME.  This is what happens if FRAME correspond to a nested function
 257        and VAR is defined in the outer function.  If callers know that VAR is
 258        located in FRAME or is global/static, NULL can be passed as VAR_BLOCK.
 259
 260        Throw an error if the variable cannot be found.  */
 261
 262     struct value *(*la_read_var_value) (struct symbol *var,
 263                                         const struct block *var_block,
 264                                         struct frame_info *frame);
 265
 266     /* PC is possibly an unknown languages trampoline.
 267        If that PC falls in a trampoline belonging to this language,
 268        return the address of the first pc in the real function, or 0
 269        if it isn't a language tramp for this language.  */
 270     CORE_ADDR (*skip_trampoline) (struct frame_info *, CORE_ADDR);
 271
 272     /* Now come some hooks for lookup_symbol.  */
 273
 274     /* If this is non-NULL, specifies the name that of the implicit
 275        local variable that refers to the current object instance.  */
 276
 277     char *la_name_of_this;
 278
 279     /* This is a function that lookup_symbol will call when it gets to
 280        the part of symbol lookup where C looks up static and global
 281        variables.  */
 282
 283     struct block_symbol (*la_lookup_symbol_nonlocal)
 284       (const struct language_defn *,
 285        const char *,
 286        const struct block *,
 287        const domain_enum);
 288
 289     /* Find the definition of the type with the given name.  */
 290     struct type *(*la_lookup_transparent_type) (const char *);
 291
 292     /* Return demangled language symbol, or NULL.  */
 293     char *(*la_demangle) (const char *mangled, int options);
 294
 295     /* Return class name of a mangled method name or NULL.  */
 296     char *(*la_class_name_from_physname) (const char *physname);
 297
 298     /* Table for printing expressions.  */
 299
 300     const struct op_print *la_op_print_tab;
 301
 302     /* Zero if the language has first-class arrays.  True if there are no
 303        array values, and array objects decay to pointers, as in C.  */
 304
 305     char c_style_arrays;
 306
 307     /* Index to use for extracting the first element of a string.  */
 308     char string_lower_bound;
 309
 310     /* The list of characters forming word boundaries.  */
 311     char *(*la_word_break_characters) (void);
 312
 313     /* Should return a vector of all symbols which are possible
 314        completions for TEXT.  WORD is the entire command on which the
 315        completion is being made.  If CODE is TYPE_CODE_UNDEF, then all
 316        symbols should be examined; otherwise, only STRUCT_DOMAIN
 317        symbols whose type has a code of CODE should be matched.  */
 318     VEC (char_ptr) *(*la_make_symbol_completion_list) (const char *text,
 319                                                        const char *word,
 320                                                        enum type_code code);
 321
 322     /* The per-architecture (OS/ABI) language information.  */
 323     void (*la_language_arch_info) (struct gdbarch *,
 324                                    struct language_arch_info *);
 325
 326     /* Print the index of an element of an array.  */
 327     void (*la_print_array_index) (struct value *index_value,
 328                                   struct ui_file *stream,
 329                                   const struct value_print_options *options);
 330
 331     /* Return non-zero if TYPE should be passed (and returned) by
 332        reference at the language level.  */
 333     int (*la_pass_by_reference) (struct type *type);
 334
 335     /* Obtain a string from the inferior, storing it in a newly allocated
 336        buffer in BUFFER, which should be freed by the caller.  If the
 337        in- and out-parameter *LENGTH is specified at -1, the string is
 338        read until a null character of the appropriate width is found -
 339        otherwise the string is read to the length of characters specified.
 340        On completion, *LENGTH will hold the size of the string in characters.
 341        If a *LENGTH of -1 was specified it will count only actual
 342        characters, excluding any eventual terminating null character.
 343        Otherwise *LENGTH will include all characters - including any nulls.
 344        CHARSET will hold the encoding used in the string.  */
 345     void (*la_get_string) (struct value *value, gdb_byte **buffer, int *length,
 346                            struct type **chartype, const char **charset);
 347
 348     /* Return a pointer to the function that should be used to match
 349        a symbol name against LOOKUP_NAME. This is mostly for languages
 350        such as Ada where the matching algorithm depends on LOOKUP_NAME.
 351
 352        This field may be NULL, in which case strcmp_iw will be used
 353        to perform the matching.  */
 354     symbol_name_cmp_ftype (*la_get_symbol_name_cmp) (const char *lookup_name);
 355
 356     /* Find all symbols in the current program space matching NAME in
 357        DOMAIN, according to this language's rules.
 358
 359        The search is done in BLOCK only.
 360        The caller is responsible for iterating up through superblocks
 361        if desired.
 362
 363        For each one, call CALLBACK with the symbol and the DATA
 364        argument.  If CALLBACK returns zero, the iteration ends at that
 365        point.
 366
 367        This field may not be NULL.  If the language does not need any
 368        special processing here, 'iterate_over_symbols' should be
 369        used as the definition.  */
 370     void (*la_iterate_over_symbols) (const struct block *block,
 371                                      const char *name,
 372                                      domain_enum domain,
 373                                      symbol_found_callback_ftype *callback,
 374                                      void *data);
 375
 376     /* Various operations on varobj.  */
 377     const struct lang_varobj_ops *la_varobj_ops;
 378
 379     /* If this language allows compilation from the gdb command line,
 380        this method should be non-NULL.  When called it should return
 381        an instance of struct gcc_context appropriate to the language.
 382        When defined this method must never return NULL; instead it
 383        should throw an exception on failure.  The returned compiler
 384        instance is owned by its caller and must be deallocated by
 385        calling its 'destroy' method.  */
 386
 387     struct compile_instance *(*la_get_compile_instance) (void);
 388
 389     /* This method must be defined if 'la_get_gcc_context' is defined.
 390        If 'la_get_gcc_context' is not defined, then this method is
 391        ignored.
 392
 393        This takes the user-supplied text and returns a newly malloc'd
 394        bit of code to compile.  The caller owns the result.
 395
 396        INST is the compiler instance being used.
 397        INPUT is the user's input text.
 398        GDBARCH is the architecture to use.
 399        EXPR_BLOCK is the block in which the expression is being
 400        parsed.
 401        EXPR_PC is the PC at which the expression is being parsed.  */
 402
 403     char *(*la_compute_program) (struct compile_instance *inst,
 404                                  const char *input,
 405                                  struct gdbarch *gdbarch,
 406                                  const struct block *expr_block,
 407                                  CORE_ADDR expr_pc);
 408
 409     /* Add fields above this point, so the magic number is always last.  */
 410     /* Magic number for compat checking.  */
 411
 412     long la_magic;
 413
 414   };
 415
 416 #define LANG_MAGIC      910823L
 417
 418 /* Pointer to the language_defn for our current language.  This pointer
 419    always points to *some* valid struct; it can be used without checking
 420    it for validity.
 421
 422    The current language affects expression parsing and evaluation
 423    (FIXME: it might be cleaner to make the evaluation-related stuff
 424    separate exp_opcodes for each different set of semantics.  We
 425    should at least think this through more clearly with respect to
 426    what happens if the language is changed between parsing and
 427    evaluation) and printing of things like types and arrays.  It does
 428    *not* affect symbol-reading-- each source file in a symbol-file has
 429    its own language and we should keep track of that regardless of the
 430    language when symbols are read.  If we want some manual setting for
 431    the language of symbol files (e.g. detecting when ".c" files are
 432    C++), it should be a separate setting from the current_language.  */
 433
 434 extern const struct language_defn *current_language;
 435
 436 /* Pointer to the language_defn expected by the user, e.g. the language
 437    of main(), or the language we last mentioned in a message, or C.  */
 438
 439 extern const struct language_defn *expected_language;
 440
 441 /* language_mode ==
 442    language_mode_auto:   current_language automatically set upon selection
 443    of scope (e.g. stack frame)
 444    language_mode_manual: current_language set only by user.  */
 445
 446 extern enum language_mode
 447   {
 448     language_mode_auto, language_mode_manual
 449   }
 450 language_mode;
 451
 452 struct type *language_bool_type (const struct language_defn *l,
 453                                  struct gdbarch *gdbarch);
 454
 455 struct type *language_string_char_type (const struct language_defn *l,
 456                                         struct gdbarch *gdbarch);
 457
 458 /* Look up type NAME in language L, and return its definition for architecture
 459    GDBARCH.  Returns NULL if not found.  */
 460
 461 struct type *language_lookup_primitive_type (const struct language_defn *l,
 462                                              struct gdbarch *gdbarch,
 463                                              const char *name);
 464
 465 /* Wrapper around language_lookup_primitive_type to return the
 466    corresponding symbol.  */
 467
 468 struct symbol *
 469   language_lookup_primitive_type_as_symbol (const struct language_defn *l,
 470                                             struct gdbarch *gdbarch,
 471                                             const char *name);
 472
 473 \f
 474 /* These macros define the behaviour of the expression
 475    evaluator.  */
 476
 477 /* Should we range check values against the domain of their type?  */
 478 #define RANGE_CHECK (range_check != range_check_off)
 479
 480 /* "cast" really means conversion.  */
 481 /* FIXME -- should be a setting in language_defn.  */
 482 #define CAST_IS_CONVERSION(LANG) ((LANG)->la_language == language_c  || \
 483                                   (LANG)->la_language == language_cplus || \
 484                                   (LANG)->la_language == language_objc)
 485
 486 extern void language_info (int);
 487
 488 extern enum language set_language (enum language);
 489 \f
 490
 491 /* This page contains functions that return things that are
 492    specific to languages.  Each of these functions is based on
 493    the current setting of working_lang, which the user sets
 494    with the "set language" command.  */
 495
 496 #define LA_PRINT_TYPE(type,varstring,stream,show,level,flags)           \
 497   (current_language->la_print_type(type,varstring,stream,show,level,flags))
 498
 499 #define LA_PRINT_TYPEDEF(type,new_symbol,stream) \
 500   (current_language->la_print_typedef(type,new_symbol,stream))
 501
 502 #define LA_VAL_PRINT(type,valaddr,offset,addr,stream,val,recurse,options) \
 503   (current_language->la_val_print(type,valaddr,offset,addr,stream, \
 504                                   val,recurse,options))
 505 #define LA_VALUE_PRINT(val,stream,options) \
 506   (current_language->la_value_print(val,stream,options))
 507
 508 #define LA_PRINT_CHAR(ch, type, stream) \
 509   (current_language->la_printchar(ch, type, stream))
 510 #define LA_PRINT_STRING(stream, elttype, string, length, encoding, force_ellipses, options) \
 511   (current_language->la_printstr(stream, elttype, string, length, \
 512                                  encoding, force_ellipses,options))
 513 #define LA_EMIT_CHAR(ch, type, stream, quoter) \
 514   (current_language->la_emitchar(ch, type, stream, quoter))
 515 #define LA_GET_STRING(value, buffer, length, chartype, encoding) \
 516   (current_language->la_get_string(value, buffer, length, chartype, encoding))
 517
 518 #define LA_PRINT_ARRAY_INDEX(index_value, stream, options) \
 519   (current_language->la_print_array_index(index_value, stream, options))
 520
 521 #define LA_ITERATE_OVER_SYMBOLS(BLOCK, NAME, DOMAIN, CALLBACK, DATA) \
 522   (current_language->la_iterate_over_symbols (BLOCK, NAME, DOMAIN, CALLBACK, \
 523                                               DATA))
 524
 525 /* Test a character to decide whether it can be printed in literal form
 526    or needs to be printed in another representation.  For example,
 527    in C the literal form of the character with octal value 141 is 'a'
 528    and the "other representation" is '\141'.  The "other representation"
 529    is program language dependent.  */
 530
 531 #define PRINT_LITERAL_FORM(c)           \
 532   ((c) >= 0x20                          \
 533    && ((c) < 0x7F || (c) >= 0xA0)       \
 534    && (!sevenbit_strings || (c) < 0x80))
 535
 536 /* Type predicates */
 537
 538 extern int pointer_type (struct type *);
 539
 540 /* Error messages */
 541
 542 extern void range_error (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
 543
 544 /* Data:  Does this value represent "truth" to the current language?  */
 545
 546 extern int value_true (struct value *);
 547
 548 /* Misc:  The string representing a particular enum language.  */
 549
 550 extern enum language language_enum (char *str);
 551
 552 extern const struct language_defn *language_def (enum language);
 553
 554 extern const char *language_str (enum language);
 555
 556 /* Add a language to the set known by GDB (at initialization time).  */
 557
 558 extern void add_language (const struct language_defn *);
 559
 560 /* Check for a language-specific trampoline.  */
 561
 562 extern CORE_ADDR skip_language_trampoline (struct frame_info *, CORE_ADDR pc);
 563
 564 /* Return demangled language symbol, or NULL.  */
 565 extern char *language_demangle (const struct language_defn *current_language,
 566                                 const char *mangled, int options);
 567
 568 /* Return class name from physname, or NULL.  */
 569 extern char *language_class_name_from_physname (const struct language_defn *,
 570                                                 const char *physname);
 571
 572 /* Splitting strings into words.  */
 573 extern char *default_word_break_characters (void);
 574
 575 /* Print the index of an array element using the C99 syntax.  */
 576 extern void default_print_array_index (struct value *index_value,
 577                                        struct ui_file *stream,
 578                                        const struct value_print_options *options);
 579
 580 /* Return non-zero if TYPE should be passed (and returned) by
 581    reference at the language level.  */
 582 int language_pass_by_reference (struct type *type);
 583
 584 /* Return zero; by default, types are passed by value at the language
 585    level.  The target ABI may pass or return some structs by reference
 586    independent of this.  */
 587 int default_pass_by_reference (struct type *type);
 588
 589 /* The default implementation of la_print_typedef.  */
 590 void default_print_typedef (struct type *type, struct symbol *new_symbol,
 591                             struct ui_file *stream);
 592
 593 void default_get_string (struct value *value, gdb_byte **buffer, int *length,
 594                          struct type **char_type, const char **charset);
 595
 596 void c_get_string (struct value *value, gdb_byte **buffer, int *length,
 597                    struct type **char_type, const char **charset);
 598
 599 #endif /* defined (LANGUAGE_H) */