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     /* Definitions related to expression printing, prefixifying, and
 168        dumping.  */
 169
 170     const struct exp_descriptor *la_exp_desc;
 171
 172     /* Parser function.  */
 173
 174     int (*la_parser) (struct parser_state *);
 175
 176     /* Parser error function.  */
 177
 178     void (*la_error) (char *);
 179
 180     /* Given an expression *EXPP created by prefixifying the result of
 181        la_parser, perform any remaining processing necessary to complete
 182        its translation.  *EXPP may change; la_post_parser is responsible
 183        for releasing its previous contents, if necessary.  If
 184        VOID_CONTEXT_P, then no value is expected from the expression.  */
 185
 186     void (*la_post_parser) (struct expression ** expp, int void_context_p);
 187
 188     void (*la_printchar) (int ch, struct type *chtype,
 189                           struct ui_file * stream);
 190
 191     void (*la_printstr) (struct ui_file * stream, struct type *elttype,
 192                          const gdb_byte *string, unsigned int length,
 193                          const char *encoding, int force_ellipses,
 194                          const struct value_print_options *);
 195
 196     void (*la_emitchar) (int ch, struct type *chtype,
 197                          struct ui_file * stream, int quoter);
 198
 199     /* Print a type using syntax appropriate for this language.  */
 200
 201     void (*la_print_type) (struct type *, const char *, struct ui_file *, int,
 202                            int, const struct type_print_options *);
 203
 204     /* Print a typedef using syntax appropriate for this language.
 205        TYPE is the underlying type.  NEW_SYMBOL is the symbol naming
 206        the type.  STREAM is the output stream on which to print.  */
 207
 208     void (*la_print_typedef) (struct type *type, struct symbol *new_symbol,
 209                               struct ui_file *stream);
 210
 211     /* Print a value using syntax appropriate for this language.
 212
 213        TYPE is the type of the sub-object to be printed.
 214
 215        CONTENTS holds the bits of the value.  This holds the entire
 216        enclosing object.
 217
 218        EMBEDDED_OFFSET is the offset into the outermost object of the
 219        sub-object represented by TYPE.  This is the object which this
 220        call should print.  Note that the enclosing type is not
 221        available.
 222
 223        ADDRESS is the address in the inferior of the enclosing object.
 224
 225        STREAM is the stream on which the value is to be printed.
 226
 227        RECURSE is the recursion depth.  It is zero-based.
 228
 229        OPTIONS are the formatting options to be used when
 230        printing.  */
 231
 232     void (*la_val_print) (struct type *type,
 233                           const gdb_byte *contents,
 234                           int embedded_offset, CORE_ADDR address,
 235                           struct ui_file *stream, int recurse,
 236                           const struct value *val,
 237                           const struct value_print_options *options);
 238
 239     /* Print a top-level value using syntax appropriate for this language.  */
 240
 241     void (*la_value_print) (struct value *, struct ui_file *,
 242                             const struct value_print_options *);
 243
 244     /* Given a symbol VAR, the corresponding block VAR_BLOCK (if any) and a
 245        stack frame id FRAME, read the value of the variable and return (pointer
 246        to a) struct value containing the value.
 247
 248        VAR_BLOCK is needed if there's a possibility for VAR to be outside
 249        FRAME.  This is what happens if FRAME correspond to a nested function
 250        and VAR is defined in the outer function.  If callers know that VAR is
 251        located in FRAME or is global/static, NULL can be passed as VAR_BLOCK.
 252
 253        Throw an error if the variable cannot be found.  */
 254
 255     struct value *(*la_read_var_value) (struct symbol *var,
 256                                         const struct block *var_block,
 257                                         struct frame_info *frame);
 258
 259     /* PC is possibly an unknown languages trampoline.
 260        If that PC falls in a trampoline belonging to this language,
 261        return the address of the first pc in the real function, or 0
 262        if it isn't a language tramp for this language.  */
 263     CORE_ADDR (*skip_trampoline) (struct frame_info *, CORE_ADDR);
 264
 265     /* Now come some hooks for lookup_symbol.  */
 266
 267     /* If this is non-NULL, specifies the name that of the implicit
 268        local variable that refers to the current object instance.  */
 269
 270     char *la_name_of_this;
 271
 272     /* This is a function that lookup_symbol will call when it gets to
 273        the part of symbol lookup where C looks up static and global
 274        variables.  */
 275
 276     struct block_symbol (*la_lookup_symbol_nonlocal)
 277       (const struct language_defn *,
 278        const char *,
 279        const struct block *,
 280        const domain_enum);
 281
 282     /* Find the definition of the type with the given name.  */
 283     struct type *(*la_lookup_transparent_type) (const char *);
 284
 285     /* Return demangled language symbol, or NULL.  */
 286     char *(*la_demangle) (const char *mangled, int options);
 287
 288     /* Return class name of a mangled method name or NULL.  */
 289     char *(*la_class_name_from_physname) (const char *physname);
 290
 291     /* Table for printing expressions.  */
 292
 293     const struct op_print *la_op_print_tab;
 294
 295     /* Zero if the language has first-class arrays.  True if there are no
 296        array values, and array objects decay to pointers, as in C.  */
 297
 298     char c_style_arrays;
 299
 300     /* Index to use for extracting the first element of a string.  */
 301     char string_lower_bound;
 302
 303     /* The list of characters forming word boundaries.  */
 304     char *(*la_word_break_characters) (void);
 305
 306     /* Should return a vector of all symbols which are possible
 307        completions for TEXT.  WORD is the entire command on which the
 308        completion is being made.  If CODE is TYPE_CODE_UNDEF, then all
 309        symbols should be examined; otherwise, only STRUCT_DOMAIN
 310        symbols whose type has a code of CODE should be matched.  */
 311     VEC (char_ptr) *(*la_make_symbol_completion_list) (const char *text,
 312                                                        const char *word,
 313                                                        enum type_code code);
 314
 315     /* The per-architecture (OS/ABI) language information.  */
 316     void (*la_language_arch_info) (struct gdbarch *,
 317                                    struct language_arch_info *);
 318
 319     /* Print the index of an element of an array.  */
 320     void (*la_print_array_index) (struct value *index_value,
 321                                   struct ui_file *stream,
 322                                   const struct value_print_options *options);
 323
 324     /* Return non-zero if TYPE should be passed (and returned) by
 325        reference at the language level.  */
 326     int (*la_pass_by_reference) (struct type *type);
 327
 328     /* Obtain a string from the inferior, storing it in a newly allocated
 329        buffer in BUFFER, which should be freed by the caller.  If the
 330        in- and out-parameter *LENGTH is specified at -1, the string is
 331        read until a null character of the appropriate width is found -
 332        otherwise the string is read to the length of characters specified.
 333        On completion, *LENGTH will hold the size of the string in characters.
 334        If a *LENGTH of -1 was specified it will count only actual
 335        characters, excluding any eventual terminating null character.
 336        Otherwise *LENGTH will include all characters - including any nulls.
 337        CHARSET will hold the encoding used in the string.  */
 338     void (*la_get_string) (struct value *value, gdb_byte **buffer, int *length,
 339                            struct type **chartype, const char **charset);
 340
 341     /* Return a pointer to the function that should be used to match
 342        a symbol name against LOOKUP_NAME. This is mostly for languages
 343        such as Ada where the matching algorithm depends on LOOKUP_NAME.
 344
 345        This field may be NULL, in which case strcmp_iw will be used
 346        to perform the matching.  */
 347     symbol_name_cmp_ftype (*la_get_symbol_name_cmp) (const char *lookup_name);
 348
 349     /* Find all symbols in the current program space matching NAME in
 350        DOMAIN, according to this language's rules.
 351
 352        The search is done in BLOCK only.
 353        The caller is responsible for iterating up through superblocks
 354        if desired.
 355
 356        For each one, call CALLBACK with the symbol and the DATA
 357        argument.  If CALLBACK returns zero, the iteration ends at that
 358        point.
 359
 360        This field may not be NULL.  If the language does not need any
 361        special processing here, 'iterate_over_symbols' should be
 362        used as the definition.  */
 363     void (*la_iterate_over_symbols) (const struct block *block,
 364                                      const char *name,
 365                                      domain_enum domain,
 366                                      symbol_found_callback_ftype *callback,
 367                                      void *data);
 368
 369     /* Various operations on varobj.  */
 370     const struct lang_varobj_ops *la_varobj_ops;
 371
 372     /* If this language allows compilation from the gdb command line,
 373        this method should be non-NULL.  When called it should return
 374        an instance of struct gcc_context appropriate to the language.
 375        When defined this method must never return NULL; instead it
 376        should throw an exception on failure.  The returned compiler
 377        instance is owned by its caller and must be deallocated by
 378        calling its 'destroy' method.  */
 379
 380     struct compile_instance *(*la_get_compile_instance) (void);
 381
 382     /* This method must be defined if 'la_get_gcc_context' is defined.
 383        If 'la_get_gcc_context' is not defined, then this method is
 384        ignored.
 385
 386        This takes the user-supplied text and returns a newly malloc'd
 387        bit of code to compile.  The caller owns the result.
 388
 389        INST is the compiler instance being used.
 390        INPUT is the user's input text.
 391        GDBARCH is the architecture to use.
 392        EXPR_BLOCK is the block in which the expression is being
 393        parsed.
 394        EXPR_PC is the PC at which the expression is being parsed.  */
 395
 396     char *(*la_compute_program) (struct compile_instance *inst,
 397                                  const char *input,
 398                                  struct gdbarch *gdbarch,
 399                                  const struct block *expr_block,
 400                                  CORE_ADDR expr_pc);
 401
 402     /* Add fields above this point, so the magic number is always last.  */
 403     /* Magic number for compat checking.  */
 404
 405     long la_magic;
 406
 407   };
 408
 409 #define LANG_MAGIC      910823L
 410
 411 /* Pointer to the language_defn for our current language.  This pointer
 412    always points to *some* valid struct; it can be used without checking
 413    it for validity.
 414
 415    The current language affects expression parsing and evaluation
 416    (FIXME: it might be cleaner to make the evaluation-related stuff
 417    separate exp_opcodes for each different set of semantics.  We
 418    should at least think this through more clearly with respect to
 419    what happens if the language is changed between parsing and
 420    evaluation) and printing of things like types and arrays.  It does
 421    *not* affect symbol-reading-- each source file in a symbol-file has
 422    its own language and we should keep track of that regardless of the
 423    language when symbols are read.  If we want some manual setting for
 424    the language of symbol files (e.g. detecting when ".c" files are
 425    C++), it should be a separate setting from the current_language.  */
 426
 427 extern const struct language_defn *current_language;
 428
 429 /* Pointer to the language_defn expected by the user, e.g. the language
 430    of main(), or the language we last mentioned in a message, or C.  */
 431
 432 extern const struct language_defn *expected_language;
 433
 434 /* language_mode ==
 435    language_mode_auto:   current_language automatically set upon selection
 436    of scope (e.g. stack frame)
 437    language_mode_manual: current_language set only by user.  */
 438
 439 extern enum language_mode
 440   {
 441     language_mode_auto, language_mode_manual
 442   }
 443 language_mode;
 444
 445 struct type *language_bool_type (const struct language_defn *l,
 446                                  struct gdbarch *gdbarch);
 447
 448 struct type *language_string_char_type (const struct language_defn *l,
 449                                         struct gdbarch *gdbarch);
 450
 451 /* Look up type NAME in language L, and return its definition for architecture
 452    GDBARCH.  Returns NULL if not found.  */
 453
 454 struct type *language_lookup_primitive_type (const struct language_defn *l,
 455                                              struct gdbarch *gdbarch,
 456                                              const char *name);
 457
 458 /* Wrapper around language_lookup_primitive_type to return the
 459    corresponding symbol.  */
 460
 461 struct symbol *
 462   language_lookup_primitive_type_as_symbol (const struct language_defn *l,
 463                                             struct gdbarch *gdbarch,
 464                                             const char *name);
 465
 466 \f
 467 /* These macros define the behaviour of the expression
 468    evaluator.  */
 469
 470 /* Should we range check values against the domain of their type?  */
 471 #define RANGE_CHECK (range_check != range_check_off)
 472
 473 /* "cast" really means conversion.  */
 474 /* FIXME -- should be a setting in language_defn.  */
 475 #define CAST_IS_CONVERSION(LANG) ((LANG)->la_language == language_c  || \
 476                                   (LANG)->la_language == language_cplus || \
 477                                   (LANG)->la_language == language_objc)
 478
 479 extern void language_info (int);
 480
 481 extern enum language set_language (enum language);
 482 \f
 483
 484 /* This page contains functions that return things that are
 485    specific to languages.  Each of these functions is based on
 486    the current setting of working_lang, which the user sets
 487    with the "set language" command.  */
 488
 489 #define LA_PRINT_TYPE(type,varstring,stream,show,level,flags)           \
 490   (current_language->la_print_type(type,varstring,stream,show,level,flags))
 491
 492 #define LA_PRINT_TYPEDEF(type,new_symbol,stream) \
 493   (current_language->la_print_typedef(type,new_symbol,stream))
 494
 495 #define LA_VAL_PRINT(type,valaddr,offset,addr,stream,val,recurse,options) \
 496   (current_language->la_val_print(type,valaddr,offset,addr,stream, \
 497                                   val,recurse,options))
 498 #define LA_VALUE_PRINT(val,stream,options) \
 499   (current_language->la_value_print(val,stream,options))
 500
 501 #define LA_PRINT_CHAR(ch, type, stream) \
 502   (current_language->la_printchar(ch, type, stream))
 503 #define LA_PRINT_STRING(stream, elttype, string, length, encoding, force_ellipses, options) \
 504   (current_language->la_printstr(stream, elttype, string, length, \
 505                                  encoding, force_ellipses,options))
 506 #define LA_EMIT_CHAR(ch, type, stream, quoter) \
 507   (current_language->la_emitchar(ch, type, stream, quoter))
 508 #define LA_GET_STRING(value, buffer, length, chartype, encoding) \
 509   (current_language->la_get_string(value, buffer, length, chartype, encoding))
 510
 511 #define LA_PRINT_ARRAY_INDEX(index_value, stream, options) \
 512   (current_language->la_print_array_index(index_value, stream, options))
 513
 514 #define LA_ITERATE_OVER_SYMBOLS(BLOCK, NAME, DOMAIN, CALLBACK, DATA) \
 515   (current_language->la_iterate_over_symbols (BLOCK, NAME, DOMAIN, CALLBACK, \
 516                                               DATA))
 517
 518 /* Test a character to decide whether it can be printed in literal form
 519    or needs to be printed in another representation.  For example,
 520    in C the literal form of the character with octal value 141 is 'a'
 521    and the "other representation" is '\141'.  The "other representation"
 522    is program language dependent.  */
 523
 524 #define PRINT_LITERAL_FORM(c)           \
 525   ((c) >= 0x20                          \
 526    && ((c) < 0x7F || (c) >= 0xA0)       \
 527    && (!sevenbit_strings || (c) < 0x80))
 528
 529 /* Type predicates */
 530
 531 extern int pointer_type (struct type *);
 532
 533 /* Error messages */
 534
 535 extern void range_error (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
 536
 537 /* Data:  Does this value represent "truth" to the current language?  */
 538
 539 extern int value_true (struct value *);
 540
 541 /* Misc:  The string representing a particular enum language.  */
 542
 543 extern enum language language_enum (char *str);
 544
 545 extern const struct language_defn *language_def (enum language);
 546
 547 extern const char *language_str (enum language);
 548
 549 /* Add a language to the set known by GDB (at initialization time).  */
 550
 551 extern void add_language (const struct language_defn *);
 552
 553 /* Check for a language-specific trampoline.  */
 554
 555 extern CORE_ADDR skip_language_trampoline (struct frame_info *, CORE_ADDR pc);
 556
 557 /* Return demangled language symbol, or NULL.  */
 558 extern char *language_demangle (const struct language_defn *current_language,
 559                                 const char *mangled, int options);
 560
 561 /* Return class name from physname, or NULL.  */
 562 extern char *language_class_name_from_physname (const struct language_defn *,
 563                                                 const char *physname);
 564
 565 /* Splitting strings into words.  */
 566 extern char *default_word_break_characters (void);
 567
 568 /* Print the index of an array element using the C99 syntax.  */
 569 extern void default_print_array_index (struct value *index_value,
 570                                        struct ui_file *stream,
 571                                        const struct value_print_options *options);
 572
 573 /* Return non-zero if TYPE should be passed (and returned) by
 574    reference at the language level.  */
 575 int language_pass_by_reference (struct type *type);
 576
 577 /* Return zero; by default, types are passed by value at the language
 578    level.  The target ABI may pass or return some structs by reference
 579    independent of this.  */
 580 int default_pass_by_reference (struct type *type);
 581
 582 /* The default implementation of la_print_typedef.  */
 583 void default_print_typedef (struct type *type, struct symbol *new_symbol,
 584                             struct ui_file *stream);
 585
 586 void default_get_string (struct value *value, gdb_byte **buffer, int *length,
 587                          struct type **char_type, const char **charset);
 588
 589 void c_get_string (struct value *value, gdb_byte **buffer, int *length,
 590                    struct type **char_type, const char **charset);
 591
 592 #endif /* defined (LANGUAGE_H) */