846cf8ce7d7c9d797c5d0f188b9311efcec23864
[binutils-gdb.git] / gdb / ada-lang.c
1 /* Ada language support routines for GDB, the GNU debugger.
2
3 Copyright (C) 1992-2018 Free Software Foundation, Inc.
4
5 This file is part of GDB.
6
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 3 of the License, or
10 (at your option) any later version.
11
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
16
17 You should have received a copy of the GNU General Public License
18 along with this program. If not, see <http://www.gnu.org/licenses/>. */
19
20
21 #include "defs.h"
22 #include <ctype.h>
23 #include "demangle.h"
24 #include "gdb_regex.h"
25 #include "frame.h"
26 #include "symtab.h"
27 #include "gdbtypes.h"
28 #include "gdbcmd.h"
29 #include "expression.h"
30 #include "parser-defs.h"
31 #include "language.h"
32 #include "varobj.h"
33 #include "c-lang.h"
34 #include "inferior.h"
35 #include "symfile.h"
36 #include "objfiles.h"
37 #include "breakpoint.h"
38 #include "gdbcore.h"
39 #include "hashtab.h"
40 #include "gdb_obstack.h"
41 #include "ada-lang.h"
42 #include "completer.h"
43 #include <sys/stat.h>
44 #include "ui-out.h"
45 #include "block.h"
46 #include "infcall.h"
47 #include "dictionary.h"
48 #include "annotate.h"
49 #include "valprint.h"
50 #include "source.h"
51 #include "observer.h"
52 #include "vec.h"
53 #include "stack.h"
54 #include "gdb_vecs.h"
55 #include "typeprint.h"
56 #include "namespace.h"
57
58 #include "psymtab.h"
59 #include "value.h"
60 #include "mi/mi-common.h"
61 #include "arch-utils.h"
62 #include "cli/cli-utils.h"
63 #include "common/function-view.h"
64 #include "common/byte-vector.h"
65 #include <algorithm>
66
67 /* Define whether or not the C operator '/' truncates towards zero for
68 differently signed operands (truncation direction is undefined in C).
69 Copied from valarith.c. */
70
71 #ifndef TRUNCATION_TOWARDS_ZERO
72 #define TRUNCATION_TOWARDS_ZERO ((-5 / 2) == -2)
73 #endif
74
75 static struct type *desc_base_type (struct type *);
76
77 static struct type *desc_bounds_type (struct type *);
78
79 static struct value *desc_bounds (struct value *);
80
81 static int fat_pntr_bounds_bitpos (struct type *);
82
83 static int fat_pntr_bounds_bitsize (struct type *);
84
85 static struct type *desc_data_target_type (struct type *);
86
87 static struct value *desc_data (struct value *);
88
89 static int fat_pntr_data_bitpos (struct type *);
90
91 static int fat_pntr_data_bitsize (struct type *);
92
93 static struct value *desc_one_bound (struct value *, int, int);
94
95 static int desc_bound_bitpos (struct type *, int, int);
96
97 static int desc_bound_bitsize (struct type *, int, int);
98
99 static struct type *desc_index_type (struct type *, int);
100
101 static int desc_arity (struct type *);
102
103 static int ada_type_match (struct type *, struct type *, int);
104
105 static int ada_args_match (struct symbol *, struct value **, int);
106
107 static struct value *make_array_descriptor (struct type *, struct value *);
108
109 static void ada_add_block_symbols (struct obstack *,
110 const struct block *,
111 const lookup_name_info &lookup_name,
112 domain_enum, struct objfile *);
113
114 static void ada_add_all_symbols (struct obstack *, const struct block *,
115 const lookup_name_info &lookup_name,
116 domain_enum, int, int *);
117
118 static int is_nonfunction (struct block_symbol *, int);
119
120 static void add_defn_to_vec (struct obstack *, struct symbol *,
121 const struct block *);
122
123 static int num_defns_collected (struct obstack *);
124
125 static struct block_symbol *defns_collected (struct obstack *, int);
126
127 static struct value *resolve_subexp (expression_up *, int *, int,
128 struct type *);
129
130 static void replace_operator_with_call (expression_up *, int, int, int,
131 struct symbol *, const struct block *);
132
133 static int possible_user_operator_p (enum exp_opcode, struct value **);
134
135 static const char *ada_op_name (enum exp_opcode);
136
137 static const char *ada_decoded_op_name (enum exp_opcode);
138
139 static int numeric_type_p (struct type *);
140
141 static int integer_type_p (struct type *);
142
143 static int scalar_type_p (struct type *);
144
145 static int discrete_type_p (struct type *);
146
147 static enum ada_renaming_category parse_old_style_renaming (struct type *,
148 const char **,
149 int *,
150 const char **);
151
152 static struct symbol *find_old_style_renaming_symbol (const char *,
153 const struct block *);
154
155 static struct type *ada_lookup_struct_elt_type (struct type *, const char *,
156 int, int);
157
158 static struct value *evaluate_subexp_type (struct expression *, int *);
159
160 static struct type *ada_find_parallel_type_with_name (struct type *,
161 const char *);
162
163 static int is_dynamic_field (struct type *, int);
164
165 static struct type *to_fixed_variant_branch_type (struct type *,
166 const gdb_byte *,
167 CORE_ADDR, struct value *);
168
169 static struct type *to_fixed_array_type (struct type *, struct value *, int);
170
171 static struct type *to_fixed_range_type (struct type *, struct value *);
172
173 static struct type *to_static_fixed_type (struct type *);
174 static struct type *static_unwrap_type (struct type *type);
175
176 static struct value *unwrap_value (struct value *);
177
178 static struct type *constrained_packed_array_type (struct type *, long *);
179
180 static struct type *decode_constrained_packed_array_type (struct type *);
181
182 static long decode_packed_array_bitsize (struct type *);
183
184 static struct value *decode_constrained_packed_array (struct value *);
185
186 static int ada_is_packed_array_type (struct type *);
187
188 static int ada_is_unconstrained_packed_array_type (struct type *);
189
190 static struct value *value_subscript_packed (struct value *, int,
191 struct value **);
192
193 static void move_bits (gdb_byte *, int, const gdb_byte *, int, int, int);
194
195 static struct value *coerce_unspec_val_to_type (struct value *,
196 struct type *);
197
198 static int lesseq_defined_than (struct symbol *, struct symbol *);
199
200 static int equiv_types (struct type *, struct type *);
201
202 static int is_name_suffix (const char *);
203
204 static int advance_wild_match (const char **, const char *, int);
205
206 static bool wild_match (const char *name, const char *patn);
207
208 static struct value *ada_coerce_ref (struct value *);
209
210 static LONGEST pos_atr (struct value *);
211
212 static struct value *value_pos_atr (struct type *, struct value *);
213
214 static struct value *value_val_atr (struct type *, struct value *);
215
216 static struct symbol *standard_lookup (const char *, const struct block *,
217 domain_enum);
218
219 static struct value *ada_search_struct_field (const char *, struct value *, int,
220 struct type *);
221
222 static struct value *ada_value_primitive_field (struct value *, int, int,
223 struct type *);
224
225 static int find_struct_field (const char *, struct type *, int,
226 struct type **, int *, int *, int *, int *);
227
228 static struct value *ada_to_fixed_value_create (struct type *, CORE_ADDR,
229 struct value *);
230
231 static int ada_resolve_function (struct block_symbol *, int,
232 struct value **, int, const char *,
233 struct type *);
234
235 static int ada_is_direct_array_type (struct type *);
236
237 static void ada_language_arch_info (struct gdbarch *,
238 struct language_arch_info *);
239
240 static struct value *ada_index_struct_field (int, struct value *, int,
241 struct type *);
242
243 static struct value *assign_aggregate (struct value *, struct value *,
244 struct expression *,
245 int *, enum noside);
246
247 static void aggregate_assign_from_choices (struct value *, struct value *,
248 struct expression *,
249 int *, LONGEST *, int *,
250 int, LONGEST, LONGEST);
251
252 static void aggregate_assign_positional (struct value *, struct value *,
253 struct expression *,
254 int *, LONGEST *, int *, int,
255 LONGEST, LONGEST);
256
257
258 static void aggregate_assign_others (struct value *, struct value *,
259 struct expression *,
260 int *, LONGEST *, int, LONGEST, LONGEST);
261
262
263 static void add_component_interval (LONGEST, LONGEST, LONGEST *, int *, int);
264
265
266 static struct value *ada_evaluate_subexp (struct type *, struct expression *,
267 int *, enum noside);
268
269 static void ada_forward_operator_length (struct expression *, int, int *,
270 int *);
271
272 static struct type *ada_find_any_type (const char *name);
273
274 static symbol_name_matcher_ftype *ada_get_symbol_name_matcher
275 (const lookup_name_info &lookup_name);
276
277 \f
278
279 /* The result of a symbol lookup to be stored in our symbol cache. */
280
281 struct cache_entry
282 {
283 /* The name used to perform the lookup. */
284 const char *name;
285 /* The namespace used during the lookup. */
286 domain_enum domain;
287 /* The symbol returned by the lookup, or NULL if no matching symbol
288 was found. */
289 struct symbol *sym;
290 /* The block where the symbol was found, or NULL if no matching
291 symbol was found. */
292 const struct block *block;
293 /* A pointer to the next entry with the same hash. */
294 struct cache_entry *next;
295 };
296
297 /* The Ada symbol cache, used to store the result of Ada-mode symbol
298 lookups in the course of executing the user's commands.
299
300 The cache is implemented using a simple, fixed-sized hash.
301 The size is fixed on the grounds that there are not likely to be
302 all that many symbols looked up during any given session, regardless
303 of the size of the symbol table. If we decide to go to a resizable
304 table, let's just use the stuff from libiberty instead. */
305
306 #define HASH_SIZE 1009
307
308 struct ada_symbol_cache
309 {
310 /* An obstack used to store the entries in our cache. */
311 struct obstack cache_space;
312
313 /* The root of the hash table used to implement our symbol cache. */
314 struct cache_entry *root[HASH_SIZE];
315 };
316
317 static void ada_free_symbol_cache (struct ada_symbol_cache *sym_cache);
318
319 /* Maximum-sized dynamic type. */
320 static unsigned int varsize_limit;
321
322 static const char ada_completer_word_break_characters[] =
323 #ifdef VMS
324 " \t\n!@#%^&*()+=|~`}{[]\";:?/,-";
325 #else
326 " \t\n!@#$%^&*()+=|~`}{[]\";:?/,-";
327 #endif
328
329 /* The name of the symbol to use to get the name of the main subprogram. */
330 static const char ADA_MAIN_PROGRAM_SYMBOL_NAME[]
331 = "__gnat_ada_main_program_name";
332
333 /* Limit on the number of warnings to raise per expression evaluation. */
334 static int warning_limit = 2;
335
336 /* Number of warning messages issued; reset to 0 by cleanups after
337 expression evaluation. */
338 static int warnings_issued = 0;
339
340 static const char *known_runtime_file_name_patterns[] = {
341 ADA_KNOWN_RUNTIME_FILE_NAME_PATTERNS NULL
342 };
343
344 static const char *known_auxiliary_function_name_patterns[] = {
345 ADA_KNOWN_AUXILIARY_FUNCTION_NAME_PATTERNS NULL
346 };
347
348 /* Maintenance-related settings for this module. */
349
350 static struct cmd_list_element *maint_set_ada_cmdlist;
351 static struct cmd_list_element *maint_show_ada_cmdlist;
352
353 /* Implement the "maintenance set ada" (prefix) command. */
354
355 static void
356 maint_set_ada_cmd (const char *args, int from_tty)
357 {
358 help_list (maint_set_ada_cmdlist, "maintenance set ada ", all_commands,
359 gdb_stdout);
360 }
361
362 /* Implement the "maintenance show ada" (prefix) command. */
363
364 static void
365 maint_show_ada_cmd (const char *args, int from_tty)
366 {
367 cmd_show_list (maint_show_ada_cmdlist, from_tty, "");
368 }
369
370 /* The "maintenance ada set/show ignore-descriptive-type" value. */
371
372 static int ada_ignore_descriptive_types_p = 0;
373
374 /* Inferior-specific data. */
375
376 /* Per-inferior data for this module. */
377
378 struct ada_inferior_data
379 {
380 /* The ada__tags__type_specific_data type, which is used when decoding
381 tagged types. With older versions of GNAT, this type was directly
382 accessible through a component ("tsd") in the object tag. But this
383 is no longer the case, so we cache it for each inferior. */
384 struct type *tsd_type;
385
386 /* The exception_support_info data. This data is used to determine
387 how to implement support for Ada exception catchpoints in a given
388 inferior. */
389 const struct exception_support_info *exception_info;
390 };
391
392 /* Our key to this module's inferior data. */
393 static const struct inferior_data *ada_inferior_data;
394
395 /* A cleanup routine for our inferior data. */
396 static void
397 ada_inferior_data_cleanup (struct inferior *inf, void *arg)
398 {
399 struct ada_inferior_data *data;
400
401 data = (struct ada_inferior_data *) inferior_data (inf, ada_inferior_data);
402 if (data != NULL)
403 xfree (data);
404 }
405
406 /* Return our inferior data for the given inferior (INF).
407
408 This function always returns a valid pointer to an allocated
409 ada_inferior_data structure. If INF's inferior data has not
410 been previously set, this functions creates a new one with all
411 fields set to zero, sets INF's inferior to it, and then returns
412 a pointer to that newly allocated ada_inferior_data. */
413
414 static struct ada_inferior_data *
415 get_ada_inferior_data (struct inferior *inf)
416 {
417 struct ada_inferior_data *data;
418
419 data = (struct ada_inferior_data *) inferior_data (inf, ada_inferior_data);
420 if (data == NULL)
421 {
422 data = XCNEW (struct ada_inferior_data);
423 set_inferior_data (inf, ada_inferior_data, data);
424 }
425
426 return data;
427 }
428
429 /* Perform all necessary cleanups regarding our module's inferior data
430 that is required after the inferior INF just exited. */
431
432 static void
433 ada_inferior_exit (struct inferior *inf)
434 {
435 ada_inferior_data_cleanup (inf, NULL);
436 set_inferior_data (inf, ada_inferior_data, NULL);
437 }
438
439
440 /* program-space-specific data. */
441
442 /* This module's per-program-space data. */
443 struct ada_pspace_data
444 {
445 /* The Ada symbol cache. */
446 struct ada_symbol_cache *sym_cache;
447 };
448
449 /* Key to our per-program-space data. */
450 static const struct program_space_data *ada_pspace_data_handle;
451
452 /* Return this module's data for the given program space (PSPACE).
453 If not is found, add a zero'ed one now.
454
455 This function always returns a valid object. */
456
457 static struct ada_pspace_data *
458 get_ada_pspace_data (struct program_space *pspace)
459 {
460 struct ada_pspace_data *data;
461
462 data = ((struct ada_pspace_data *)
463 program_space_data (pspace, ada_pspace_data_handle));
464 if (data == NULL)
465 {
466 data = XCNEW (struct ada_pspace_data);
467 set_program_space_data (pspace, ada_pspace_data_handle, data);
468 }
469
470 return data;
471 }
472
473 /* The cleanup callback for this module's per-program-space data. */
474
475 static void
476 ada_pspace_data_cleanup (struct program_space *pspace, void *data)
477 {
478 struct ada_pspace_data *pspace_data = (struct ada_pspace_data *) data;
479
480 if (pspace_data->sym_cache != NULL)
481 ada_free_symbol_cache (pspace_data->sym_cache);
482 xfree (pspace_data);
483 }
484
485 /* Utilities */
486
487 /* If TYPE is a TYPE_CODE_TYPEDEF type, return the target type after
488 all typedef layers have been peeled. Otherwise, return TYPE.
489
490 Normally, we really expect a typedef type to only have 1 typedef layer.
491 In other words, we really expect the target type of a typedef type to be
492 a non-typedef type. This is particularly true for Ada units, because
493 the language does not have a typedef vs not-typedef distinction.
494 In that respect, the Ada compiler has been trying to eliminate as many
495 typedef definitions in the debugging information, since they generally
496 do not bring any extra information (we still use typedef under certain
497 circumstances related mostly to the GNAT encoding).
498
499 Unfortunately, we have seen situations where the debugging information
500 generated by the compiler leads to such multiple typedef layers. For
501 instance, consider the following example with stabs:
502
503 .stabs "pck__float_array___XUP:Tt(0,46)=s16P_ARRAY:(0,47)=[...]"[...]
504 .stabs "pck__float_array___XUP:t(0,36)=(0,46)",128,0,6,0
505
506 This is an error in the debugging information which causes type
507 pck__float_array___XUP to be defined twice, and the second time,
508 it is defined as a typedef of a typedef.
509
510 This is on the fringe of legality as far as debugging information is
511 concerned, and certainly unexpected. But it is easy to handle these
512 situations correctly, so we can afford to be lenient in this case. */
513
514 static struct type *
515 ada_typedef_target_type (struct type *type)
516 {
517 while (TYPE_CODE (type) == TYPE_CODE_TYPEDEF)
518 type = TYPE_TARGET_TYPE (type);
519 return type;
520 }
521
522 /* Given DECODED_NAME a string holding a symbol name in its
523 decoded form (ie using the Ada dotted notation), returns
524 its unqualified name. */
525
526 static const char *
527 ada_unqualified_name (const char *decoded_name)
528 {
529 const char *result;
530
531 /* If the decoded name starts with '<', it means that the encoded
532 name does not follow standard naming conventions, and thus that
533 it is not your typical Ada symbol name. Trying to unqualify it
534 is therefore pointless and possibly erroneous. */
535 if (decoded_name[0] == '<')
536 return decoded_name;
537
538 result = strrchr (decoded_name, '.');
539 if (result != NULL)
540 result++; /* Skip the dot... */
541 else
542 result = decoded_name;
543
544 return result;
545 }
546
547 /* Return a string starting with '<', followed by STR, and '>'.
548 The result is good until the next call. */
549
550 static char *
551 add_angle_brackets (const char *str)
552 {
553 static char *result = NULL;
554
555 xfree (result);
556 result = xstrprintf ("<%s>", str);
557 return result;
558 }
559
560 static const char *
561 ada_get_gdb_completer_word_break_characters (void)
562 {
563 return ada_completer_word_break_characters;
564 }
565
566 /* Print an array element index using the Ada syntax. */
567
568 static void
569 ada_print_array_index (struct value *index_value, struct ui_file *stream,
570 const struct value_print_options *options)
571 {
572 LA_VALUE_PRINT (index_value, stream, options);
573 fprintf_filtered (stream, " => ");
574 }
575
576 /* Assuming VECT points to an array of *SIZE objects of size
577 ELEMENT_SIZE, grow it to contain at least MIN_SIZE objects,
578 updating *SIZE as necessary and returning the (new) array. */
579
580 void *
581 grow_vect (void *vect, size_t *size, size_t min_size, int element_size)
582 {
583 if (*size < min_size)
584 {
585 *size *= 2;
586 if (*size < min_size)
587 *size = min_size;
588 vect = xrealloc (vect, *size * element_size);
589 }
590 return vect;
591 }
592
593 /* True (non-zero) iff TARGET matches FIELD_NAME up to any trailing
594 suffix of FIELD_NAME beginning "___". */
595
596 static int
597 field_name_match (const char *field_name, const char *target)
598 {
599 int len = strlen (target);
600
601 return
602 (strncmp (field_name, target, len) == 0
603 && (field_name[len] == '\0'
604 || (startswith (field_name + len, "___")
605 && strcmp (field_name + strlen (field_name) - 6,
606 "___XVN") != 0)));
607 }
608
609
610 /* Assuming TYPE is a TYPE_CODE_STRUCT or a TYPE_CODE_TYPDEF to
611 a TYPE_CODE_STRUCT, find the field whose name matches FIELD_NAME,
612 and return its index. This function also handles fields whose name
613 have ___ suffixes because the compiler sometimes alters their name
614 by adding such a suffix to represent fields with certain constraints.
615 If the field could not be found, return a negative number if
616 MAYBE_MISSING is set. Otherwise raise an error. */
617
618 int
619 ada_get_field_index (const struct type *type, const char *field_name,
620 int maybe_missing)
621 {
622 int fieldno;
623 struct type *struct_type = check_typedef ((struct type *) type);
624
625 for (fieldno = 0; fieldno < TYPE_NFIELDS (struct_type); fieldno++)
626 if (field_name_match (TYPE_FIELD_NAME (struct_type, fieldno), field_name))
627 return fieldno;
628
629 if (!maybe_missing)
630 error (_("Unable to find field %s in struct %s. Aborting"),
631 field_name, TYPE_NAME (struct_type));
632
633 return -1;
634 }
635
636 /* The length of the prefix of NAME prior to any "___" suffix. */
637
638 int
639 ada_name_prefix_len (const char *name)
640 {
641 if (name == NULL)
642 return 0;
643 else
644 {
645 const char *p = strstr (name, "___");
646
647 if (p == NULL)
648 return strlen (name);
649 else
650 return p - name;
651 }
652 }
653
654 /* Return non-zero if SUFFIX is a suffix of STR.
655 Return zero if STR is null. */
656
657 static int
658 is_suffix (const char *str, const char *suffix)
659 {
660 int len1, len2;
661
662 if (str == NULL)
663 return 0;
664 len1 = strlen (str);
665 len2 = strlen (suffix);
666 return (len1 >= len2 && strcmp (str + len1 - len2, suffix) == 0);
667 }
668
669 /* The contents of value VAL, treated as a value of type TYPE. The
670 result is an lval in memory if VAL is. */
671
672 static struct value *
673 coerce_unspec_val_to_type (struct value *val, struct type *type)
674 {
675 type = ada_check_typedef (type);
676 if (value_type (val) == type)
677 return val;
678 else
679 {
680 struct value *result;
681
682 /* Make sure that the object size is not unreasonable before
683 trying to allocate some memory for it. */
684 ada_ensure_varsize_limit (type);
685
686 if (value_lazy (val)
687 || TYPE_LENGTH (type) > TYPE_LENGTH (value_type (val)))
688 result = allocate_value_lazy (type);
689 else
690 {
691 result = allocate_value (type);
692 value_contents_copy_raw (result, 0, val, 0, TYPE_LENGTH (type));
693 }
694 set_value_component_location (result, val);
695 set_value_bitsize (result, value_bitsize (val));
696 set_value_bitpos (result, value_bitpos (val));
697 set_value_address (result, value_address (val));
698 return result;
699 }
700 }
701
702 static const gdb_byte *
703 cond_offset_host (const gdb_byte *valaddr, long offset)
704 {
705 if (valaddr == NULL)
706 return NULL;
707 else
708 return valaddr + offset;
709 }
710
711 static CORE_ADDR
712 cond_offset_target (CORE_ADDR address, long offset)
713 {
714 if (address == 0)
715 return 0;
716 else
717 return address + offset;
718 }
719
720 /* Issue a warning (as for the definition of warning in utils.c, but
721 with exactly one argument rather than ...), unless the limit on the
722 number of warnings has passed during the evaluation of the current
723 expression. */
724
725 /* FIXME: cagney/2004-10-10: This function is mimicking the behavior
726 provided by "complaint". */
727 static void lim_warning (const char *format, ...) ATTRIBUTE_PRINTF (1, 2);
728
729 static void
730 lim_warning (const char *format, ...)
731 {
732 va_list args;
733
734 va_start (args, format);
735 warnings_issued += 1;
736 if (warnings_issued <= warning_limit)
737 vwarning (format, args);
738
739 va_end (args);
740 }
741
742 /* Issue an error if the size of an object of type T is unreasonable,
743 i.e. if it would be a bad idea to allocate a value of this type in
744 GDB. */
745
746 void
747 ada_ensure_varsize_limit (const struct type *type)
748 {
749 if (TYPE_LENGTH (type) > varsize_limit)
750 error (_("object size is larger than varsize-limit"));
751 }
752
753 /* Maximum value of a SIZE-byte signed integer type. */
754 static LONGEST
755 max_of_size (int size)
756 {
757 LONGEST top_bit = (LONGEST) 1 << (size * 8 - 2);
758
759 return top_bit | (top_bit - 1);
760 }
761
762 /* Minimum value of a SIZE-byte signed integer type. */
763 static LONGEST
764 min_of_size (int size)
765 {
766 return -max_of_size (size) - 1;
767 }
768
769 /* Maximum value of a SIZE-byte unsigned integer type. */
770 static ULONGEST
771 umax_of_size (int size)
772 {
773 ULONGEST top_bit = (ULONGEST) 1 << (size * 8 - 1);
774
775 return top_bit | (top_bit - 1);
776 }
777
778 /* Maximum value of integral type T, as a signed quantity. */
779 static LONGEST
780 max_of_type (struct type *t)
781 {
782 if (TYPE_UNSIGNED (t))
783 return (LONGEST) umax_of_size (TYPE_LENGTH (t));
784 else
785 return max_of_size (TYPE_LENGTH (t));
786 }
787
788 /* Minimum value of integral type T, as a signed quantity. */
789 static LONGEST
790 min_of_type (struct type *t)
791 {
792 if (TYPE_UNSIGNED (t))
793 return 0;
794 else
795 return min_of_size (TYPE_LENGTH (t));
796 }
797
798 /* The largest value in the domain of TYPE, a discrete type, as an integer. */
799 LONGEST
800 ada_discrete_type_high_bound (struct type *type)
801 {
802 type = resolve_dynamic_type (type, NULL, 0);
803 switch (TYPE_CODE (type))
804 {
805 case TYPE_CODE_RANGE:
806 return TYPE_HIGH_BOUND (type);
807 case TYPE_CODE_ENUM:
808 return TYPE_FIELD_ENUMVAL (type, TYPE_NFIELDS (type) - 1);
809 case TYPE_CODE_BOOL:
810 return 1;
811 case TYPE_CODE_CHAR:
812 case TYPE_CODE_INT:
813 return max_of_type (type);
814 default:
815 error (_("Unexpected type in ada_discrete_type_high_bound."));
816 }
817 }
818
819 /* The smallest value in the domain of TYPE, a discrete type, as an integer. */
820 LONGEST
821 ada_discrete_type_low_bound (struct type *type)
822 {
823 type = resolve_dynamic_type (type, NULL, 0);
824 switch (TYPE_CODE (type))
825 {
826 case TYPE_CODE_RANGE:
827 return TYPE_LOW_BOUND (type);
828 case TYPE_CODE_ENUM:
829 return TYPE_FIELD_ENUMVAL (type, 0);
830 case TYPE_CODE_BOOL:
831 return 0;
832 case TYPE_CODE_CHAR:
833 case TYPE_CODE_INT:
834 return min_of_type (type);
835 default:
836 error (_("Unexpected type in ada_discrete_type_low_bound."));
837 }
838 }
839
840 /* The identity on non-range types. For range types, the underlying
841 non-range scalar type. */
842
843 static struct type *
844 get_base_type (struct type *type)
845 {
846 while (type != NULL && TYPE_CODE (type) == TYPE_CODE_RANGE)
847 {
848 if (type == TYPE_TARGET_TYPE (type) || TYPE_TARGET_TYPE (type) == NULL)
849 return type;
850 type = TYPE_TARGET_TYPE (type);
851 }
852 return type;
853 }
854
855 /* Return a decoded version of the given VALUE. This means returning
856 a value whose type is obtained by applying all the GNAT-specific
857 encondings, making the resulting type a static but standard description
858 of the initial type. */
859
860 struct value *
861 ada_get_decoded_value (struct value *value)
862 {
863 struct type *type = ada_check_typedef (value_type (value));
864
865 if (ada_is_array_descriptor_type (type)
866 || (ada_is_constrained_packed_array_type (type)
867 && TYPE_CODE (type) != TYPE_CODE_PTR))
868 {
869 if (TYPE_CODE (type) == TYPE_CODE_TYPEDEF) /* array access type. */
870 value = ada_coerce_to_simple_array_ptr (value);
871 else
872 value = ada_coerce_to_simple_array (value);
873 }
874 else
875 value = ada_to_fixed_value (value);
876
877 return value;
878 }
879
880 /* Same as ada_get_decoded_value, but with the given TYPE.
881 Because there is no associated actual value for this type,
882 the resulting type might be a best-effort approximation in
883 the case of dynamic types. */
884
885 struct type *
886 ada_get_decoded_type (struct type *type)
887 {
888 type = to_static_fixed_type (type);
889 if (ada_is_constrained_packed_array_type (type))
890 type = ada_coerce_to_simple_array_type (type);
891 return type;
892 }
893
894 \f
895
896 /* Language Selection */
897
898 /* If the main program is in Ada, return language_ada, otherwise return LANG
899 (the main program is in Ada iif the adainit symbol is found). */
900
901 enum language
902 ada_update_initial_language (enum language lang)
903 {
904 if (lookup_minimal_symbol ("adainit", (const char *) NULL,
905 (struct objfile *) NULL).minsym != NULL)
906 return language_ada;
907
908 return lang;
909 }
910
911 /* If the main procedure is written in Ada, then return its name.
912 The result is good until the next call. Return NULL if the main
913 procedure doesn't appear to be in Ada. */
914
915 char *
916 ada_main_name (void)
917 {
918 struct bound_minimal_symbol msym;
919 static char *main_program_name = NULL;
920
921 /* For Ada, the name of the main procedure is stored in a specific
922 string constant, generated by the binder. Look for that symbol,
923 extract its address, and then read that string. If we didn't find
924 that string, then most probably the main procedure is not written
925 in Ada. */
926 msym = lookup_minimal_symbol (ADA_MAIN_PROGRAM_SYMBOL_NAME, NULL, NULL);
927
928 if (msym.minsym != NULL)
929 {
930 CORE_ADDR main_program_name_addr;
931 int err_code;
932
933 main_program_name_addr = BMSYMBOL_VALUE_ADDRESS (msym);
934 if (main_program_name_addr == 0)
935 error (_("Invalid address for Ada main program name."));
936
937 xfree (main_program_name);
938 target_read_string (main_program_name_addr, &main_program_name,
939 1024, &err_code);
940
941 if (err_code != 0)
942 return NULL;
943 return main_program_name;
944 }
945
946 /* The main procedure doesn't seem to be in Ada. */
947 return NULL;
948 }
949 \f
950 /* Symbols */
951
952 /* Table of Ada operators and their GNAT-encoded names. Last entry is pair
953 of NULLs. */
954
955 const struct ada_opname_map ada_opname_table[] = {
956 {"Oadd", "\"+\"", BINOP_ADD},
957 {"Osubtract", "\"-\"", BINOP_SUB},
958 {"Omultiply", "\"*\"", BINOP_MUL},
959 {"Odivide", "\"/\"", BINOP_DIV},
960 {"Omod", "\"mod\"", BINOP_MOD},
961 {"Orem", "\"rem\"", BINOP_REM},
962 {"Oexpon", "\"**\"", BINOP_EXP},
963 {"Olt", "\"<\"", BINOP_LESS},
964 {"Ole", "\"<=\"", BINOP_LEQ},
965 {"Ogt", "\">\"", BINOP_GTR},
966 {"Oge", "\">=\"", BINOP_GEQ},
967 {"Oeq", "\"=\"", BINOP_EQUAL},
968 {"One", "\"/=\"", BINOP_NOTEQUAL},
969 {"Oand", "\"and\"", BINOP_BITWISE_AND},
970 {"Oor", "\"or\"", BINOP_BITWISE_IOR},
971 {"Oxor", "\"xor\"", BINOP_BITWISE_XOR},
972 {"Oconcat", "\"&\"", BINOP_CONCAT},
973 {"Oabs", "\"abs\"", UNOP_ABS},
974 {"Onot", "\"not\"", UNOP_LOGICAL_NOT},
975 {"Oadd", "\"+\"", UNOP_PLUS},
976 {"Osubtract", "\"-\"", UNOP_NEG},
977 {NULL, NULL}
978 };
979
980 /* The "encoded" form of DECODED, according to GNAT conventions. The
981 result is valid until the next call to ada_encode. If
982 THROW_ERRORS, throw an error if invalid operator name is found.
983 Otherwise, return NULL in that case. */
984
985 static char *
986 ada_encode_1 (const char *decoded, bool throw_errors)
987 {
988 static char *encoding_buffer = NULL;
989 static size_t encoding_buffer_size = 0;
990 const char *p;
991 int k;
992
993 if (decoded == NULL)
994 return NULL;
995
996 GROW_VECT (encoding_buffer, encoding_buffer_size,
997 2 * strlen (decoded) + 10);
998
999 k = 0;
1000 for (p = decoded; *p != '\0'; p += 1)
1001 {
1002 if (*p == '.')
1003 {
1004 encoding_buffer[k] = encoding_buffer[k + 1] = '_';
1005 k += 2;
1006 }
1007 else if (*p == '"')
1008 {
1009 const struct ada_opname_map *mapping;
1010
1011 for (mapping = ada_opname_table;
1012 mapping->encoded != NULL
1013 && !startswith (p, mapping->decoded); mapping += 1)
1014 ;
1015 if (mapping->encoded == NULL)
1016 {
1017 if (throw_errors)
1018 error (_("invalid Ada operator name: %s"), p);
1019 else
1020 return NULL;
1021 }
1022 strcpy (encoding_buffer + k, mapping->encoded);
1023 k += strlen (mapping->encoded);
1024 break;
1025 }
1026 else
1027 {
1028 encoding_buffer[k] = *p;
1029 k += 1;
1030 }
1031 }
1032
1033 encoding_buffer[k] = '\0';
1034 return encoding_buffer;
1035 }
1036
1037 /* The "encoded" form of DECODED, according to GNAT conventions.
1038 The result is valid until the next call to ada_encode. */
1039
1040 char *
1041 ada_encode (const char *decoded)
1042 {
1043 return ada_encode_1 (decoded, true);
1044 }
1045
1046 /* Return NAME folded to lower case, or, if surrounded by single
1047 quotes, unfolded, but with the quotes stripped away. Result good
1048 to next call. */
1049
1050 char *
1051 ada_fold_name (const char *name)
1052 {
1053 static char *fold_buffer = NULL;
1054 static size_t fold_buffer_size = 0;
1055
1056 int len = strlen (name);
1057 GROW_VECT (fold_buffer, fold_buffer_size, len + 1);
1058
1059 if (name[0] == '\'')
1060 {
1061 strncpy (fold_buffer, name + 1, len - 2);
1062 fold_buffer[len - 2] = '\000';
1063 }
1064 else
1065 {
1066 int i;
1067
1068 for (i = 0; i <= len; i += 1)
1069 fold_buffer[i] = tolower (name[i]);
1070 }
1071
1072 return fold_buffer;
1073 }
1074
1075 /* Return nonzero if C is either a digit or a lowercase alphabet character. */
1076
1077 static int
1078 is_lower_alphanum (const char c)
1079 {
1080 return (isdigit (c) || (isalpha (c) && islower (c)));
1081 }
1082
1083 /* ENCODED is the linkage name of a symbol and LEN contains its length.
1084 This function saves in LEN the length of that same symbol name but
1085 without either of these suffixes:
1086 . .{DIGIT}+
1087 . ${DIGIT}+
1088 . ___{DIGIT}+
1089 . __{DIGIT}+.
1090
1091 These are suffixes introduced by the compiler for entities such as
1092 nested subprogram for instance, in order to avoid name clashes.
1093 They do not serve any purpose for the debugger. */
1094
1095 static void
1096 ada_remove_trailing_digits (const char *encoded, int *len)
1097 {
1098 if (*len > 1 && isdigit (encoded[*len - 1]))
1099 {
1100 int i = *len - 2;
1101
1102 while (i > 0 && isdigit (encoded[i]))
1103 i--;
1104 if (i >= 0 && encoded[i] == '.')
1105 *len = i;
1106 else if (i >= 0 && encoded[i] == '$')
1107 *len = i;
1108 else if (i >= 2 && startswith (encoded + i - 2, "___"))
1109 *len = i - 2;
1110 else if (i >= 1 && startswith (encoded + i - 1, "__"))
1111 *len = i - 1;
1112 }
1113 }
1114
1115 /* Remove the suffix introduced by the compiler for protected object
1116 subprograms. */
1117
1118 static void
1119 ada_remove_po_subprogram_suffix (const char *encoded, int *len)
1120 {
1121 /* Remove trailing N. */
1122
1123 /* Protected entry subprograms are broken into two
1124 separate subprograms: The first one is unprotected, and has
1125 a 'N' suffix; the second is the protected version, and has
1126 the 'P' suffix. The second calls the first one after handling
1127 the protection. Since the P subprograms are internally generated,
1128 we leave these names undecoded, giving the user a clue that this
1129 entity is internal. */
1130
1131 if (*len > 1
1132 && encoded[*len - 1] == 'N'
1133 && (isdigit (encoded[*len - 2]) || islower (encoded[*len - 2])))
1134 *len = *len - 1;
1135 }
1136
1137 /* Remove trailing X[bn]* suffixes (indicating names in package bodies). */
1138
1139 static void
1140 ada_remove_Xbn_suffix (const char *encoded, int *len)
1141 {
1142 int i = *len - 1;
1143
1144 while (i > 0 && (encoded[i] == 'b' || encoded[i] == 'n'))
1145 i--;
1146
1147 if (encoded[i] != 'X')
1148 return;
1149
1150 if (i == 0)
1151 return;
1152
1153 if (isalnum (encoded[i-1]))
1154 *len = i;
1155 }
1156
1157 /* If ENCODED follows the GNAT entity encoding conventions, then return
1158 the decoded form of ENCODED. Otherwise, return "<%s>" where "%s" is
1159 replaced by ENCODED.
1160
1161 The resulting string is valid until the next call of ada_decode.
1162 If the string is unchanged by decoding, the original string pointer
1163 is returned. */
1164
1165 const char *
1166 ada_decode (const char *encoded)
1167 {
1168 int i, j;
1169 int len0;
1170 const char *p;
1171 char *decoded;
1172 int at_start_name;
1173 static char *decoding_buffer = NULL;
1174 static size_t decoding_buffer_size = 0;
1175
1176 /* The name of the Ada main procedure starts with "_ada_".
1177 This prefix is not part of the decoded name, so skip this part
1178 if we see this prefix. */
1179 if (startswith (encoded, "_ada_"))
1180 encoded += 5;
1181
1182 /* If the name starts with '_', then it is not a properly encoded
1183 name, so do not attempt to decode it. Similarly, if the name
1184 starts with '<', the name should not be decoded. */
1185 if (encoded[0] == '_' || encoded[0] == '<')
1186 goto Suppress;
1187
1188 len0 = strlen (encoded);
1189
1190 ada_remove_trailing_digits (encoded, &len0);
1191 ada_remove_po_subprogram_suffix (encoded, &len0);
1192
1193 /* Remove the ___X.* suffix if present. Do not forget to verify that
1194 the suffix is located before the current "end" of ENCODED. We want
1195 to avoid re-matching parts of ENCODED that have previously been
1196 marked as discarded (by decrementing LEN0). */
1197 p = strstr (encoded, "___");
1198 if (p != NULL && p - encoded < len0 - 3)
1199 {
1200 if (p[3] == 'X')
1201 len0 = p - encoded;
1202 else
1203 goto Suppress;
1204 }
1205
1206 /* Remove any trailing TKB suffix. It tells us that this symbol
1207 is for the body of a task, but that information does not actually
1208 appear in the decoded name. */
1209
1210 if (len0 > 3 && startswith (encoded + len0 - 3, "TKB"))
1211 len0 -= 3;
1212
1213 /* Remove any trailing TB suffix. The TB suffix is slightly different
1214 from the TKB suffix because it is used for non-anonymous task
1215 bodies. */
1216
1217 if (len0 > 2 && startswith (encoded + len0 - 2, "TB"))
1218 len0 -= 2;
1219
1220 /* Remove trailing "B" suffixes. */
1221 /* FIXME: brobecker/2006-04-19: Not sure what this are used for... */
1222
1223 if (len0 > 1 && startswith (encoded + len0 - 1, "B"))
1224 len0 -= 1;
1225
1226 /* Make decoded big enough for possible expansion by operator name. */
1227
1228 GROW_VECT (decoding_buffer, decoding_buffer_size, 2 * len0 + 1);
1229 decoded = decoding_buffer;
1230
1231 /* Remove trailing __{digit}+ or trailing ${digit}+. */
1232
1233 if (len0 > 1 && isdigit (encoded[len0 - 1]))
1234 {
1235 i = len0 - 2;
1236 while ((i >= 0 && isdigit (encoded[i]))
1237 || (i >= 1 && encoded[i] == '_' && isdigit (encoded[i - 1])))
1238 i -= 1;
1239 if (i > 1 && encoded[i] == '_' && encoded[i - 1] == '_')
1240 len0 = i - 1;
1241 else if (encoded[i] == '$')
1242 len0 = i;
1243 }
1244
1245 /* The first few characters that are not alphabetic are not part
1246 of any encoding we use, so we can copy them over verbatim. */
1247
1248 for (i = 0, j = 0; i < len0 && !isalpha (encoded[i]); i += 1, j += 1)
1249 decoded[j] = encoded[i];
1250
1251 at_start_name = 1;
1252 while (i < len0)
1253 {
1254 /* Is this a symbol function? */
1255 if (at_start_name && encoded[i] == 'O')
1256 {
1257 int k;
1258
1259 for (k = 0; ada_opname_table[k].encoded != NULL; k += 1)
1260 {
1261 int op_len = strlen (ada_opname_table[k].encoded);
1262 if ((strncmp (ada_opname_table[k].encoded + 1, encoded + i + 1,
1263 op_len - 1) == 0)
1264 && !isalnum (encoded[i + op_len]))
1265 {
1266 strcpy (decoded + j, ada_opname_table[k].decoded);
1267 at_start_name = 0;
1268 i += op_len;
1269 j += strlen (ada_opname_table[k].decoded);
1270 break;
1271 }
1272 }
1273 if (ada_opname_table[k].encoded != NULL)
1274 continue;
1275 }
1276 at_start_name = 0;
1277
1278 /* Replace "TK__" with "__", which will eventually be translated
1279 into "." (just below). */
1280
1281 if (i < len0 - 4 && startswith (encoded + i, "TK__"))
1282 i += 2;
1283
1284 /* Replace "__B_{DIGITS}+__" sequences by "__", which will eventually
1285 be translated into "." (just below). These are internal names
1286 generated for anonymous blocks inside which our symbol is nested. */
1287
1288 if (len0 - i > 5 && encoded [i] == '_' && encoded [i+1] == '_'
1289 && encoded [i+2] == 'B' && encoded [i+3] == '_'
1290 && isdigit (encoded [i+4]))
1291 {
1292 int k = i + 5;
1293
1294 while (k < len0 && isdigit (encoded[k]))
1295 k++; /* Skip any extra digit. */
1296
1297 /* Double-check that the "__B_{DIGITS}+" sequence we found
1298 is indeed followed by "__". */
1299 if (len0 - k > 2 && encoded [k] == '_' && encoded [k+1] == '_')
1300 i = k;
1301 }
1302
1303 /* Remove _E{DIGITS}+[sb] */
1304
1305 /* Just as for protected object subprograms, there are 2 categories
1306 of subprograms created by the compiler for each entry. The first
1307 one implements the actual entry code, and has a suffix following
1308 the convention above; the second one implements the barrier and
1309 uses the same convention as above, except that the 'E' is replaced
1310 by a 'B'.
1311
1312 Just as above, we do not decode the name of barrier functions
1313 to give the user a clue that the code he is debugging has been
1314 internally generated. */
1315
1316 if (len0 - i > 3 && encoded [i] == '_' && encoded[i+1] == 'E'
1317 && isdigit (encoded[i+2]))
1318 {
1319 int k = i + 3;
1320
1321 while (k < len0 && isdigit (encoded[k]))
1322 k++;
1323
1324 if (k < len0
1325 && (encoded[k] == 'b' || encoded[k] == 's'))
1326 {
1327 k++;
1328 /* Just as an extra precaution, make sure that if this
1329 suffix is followed by anything else, it is a '_'.
1330 Otherwise, we matched this sequence by accident. */
1331 if (k == len0
1332 || (k < len0 && encoded[k] == '_'))
1333 i = k;
1334 }
1335 }
1336
1337 /* Remove trailing "N" in [a-z0-9]+N__. The N is added by
1338 the GNAT front-end in protected object subprograms. */
1339
1340 if (i < len0 + 3
1341 && encoded[i] == 'N' && encoded[i+1] == '_' && encoded[i+2] == '_')
1342 {
1343 /* Backtrack a bit up until we reach either the begining of
1344 the encoded name, or "__". Make sure that we only find
1345 digits or lowercase characters. */
1346 const char *ptr = encoded + i - 1;
1347
1348 while (ptr >= encoded && is_lower_alphanum (ptr[0]))
1349 ptr--;
1350 if (ptr < encoded
1351 || (ptr > encoded && ptr[0] == '_' && ptr[-1] == '_'))
1352 i++;
1353 }
1354
1355 if (encoded[i] == 'X' && i != 0 && isalnum (encoded[i - 1]))
1356 {
1357 /* This is a X[bn]* sequence not separated from the previous
1358 part of the name with a non-alpha-numeric character (in other
1359 words, immediately following an alpha-numeric character), then
1360 verify that it is placed at the end of the encoded name. If
1361 not, then the encoding is not valid and we should abort the
1362 decoding. Otherwise, just skip it, it is used in body-nested
1363 package names. */
1364 do
1365 i += 1;
1366 while (i < len0 && (encoded[i] == 'b' || encoded[i] == 'n'));
1367 if (i < len0)
1368 goto Suppress;
1369 }
1370 else if (i < len0 - 2 && encoded[i] == '_' && encoded[i + 1] == '_')
1371 {
1372 /* Replace '__' by '.'. */
1373 decoded[j] = '.';
1374 at_start_name = 1;
1375 i += 2;
1376 j += 1;
1377 }
1378 else
1379 {
1380 /* It's a character part of the decoded name, so just copy it
1381 over. */
1382 decoded[j] = encoded[i];
1383 i += 1;
1384 j += 1;
1385 }
1386 }
1387 decoded[j] = '\000';
1388
1389 /* Decoded names should never contain any uppercase character.
1390 Double-check this, and abort the decoding if we find one. */
1391
1392 for (i = 0; decoded[i] != '\0'; i += 1)
1393 if (isupper (decoded[i]) || decoded[i] == ' ')
1394 goto Suppress;
1395
1396 if (strcmp (decoded, encoded) == 0)
1397 return encoded;
1398 else
1399 return decoded;
1400
1401 Suppress:
1402 GROW_VECT (decoding_buffer, decoding_buffer_size, strlen (encoded) + 3);
1403 decoded = decoding_buffer;
1404 if (encoded[0] == '<')
1405 strcpy (decoded, encoded);
1406 else
1407 xsnprintf (decoded, decoding_buffer_size, "<%s>", encoded);
1408 return decoded;
1409
1410 }
1411
1412 /* Table for keeping permanent unique copies of decoded names. Once
1413 allocated, names in this table are never released. While this is a
1414 storage leak, it should not be significant unless there are massive
1415 changes in the set of decoded names in successive versions of a
1416 symbol table loaded during a single session. */
1417 static struct htab *decoded_names_store;
1418
1419 /* Returns the decoded name of GSYMBOL, as for ada_decode, caching it
1420 in the language-specific part of GSYMBOL, if it has not been
1421 previously computed. Tries to save the decoded name in the same
1422 obstack as GSYMBOL, if possible, and otherwise on the heap (so that,
1423 in any case, the decoded symbol has a lifetime at least that of
1424 GSYMBOL).
1425 The GSYMBOL parameter is "mutable" in the C++ sense: logically
1426 const, but nevertheless modified to a semantically equivalent form
1427 when a decoded name is cached in it. */
1428
1429 const char *
1430 ada_decode_symbol (const struct general_symbol_info *arg)
1431 {
1432 struct general_symbol_info *gsymbol = (struct general_symbol_info *) arg;
1433 const char **resultp =
1434 &gsymbol->language_specific.demangled_name;
1435
1436 if (!gsymbol->ada_mangled)
1437 {
1438 const char *decoded = ada_decode (gsymbol->name);
1439 struct obstack *obstack = gsymbol->language_specific.obstack;
1440
1441 gsymbol->ada_mangled = 1;
1442
1443 if (obstack != NULL)
1444 *resultp
1445 = (const char *) obstack_copy0 (obstack, decoded, strlen (decoded));
1446 else
1447 {
1448 /* Sometimes, we can't find a corresponding objfile, in
1449 which case, we put the result on the heap. Since we only
1450 decode when needed, we hope this usually does not cause a
1451 significant memory leak (FIXME). */
1452
1453 char **slot = (char **) htab_find_slot (decoded_names_store,
1454 decoded, INSERT);
1455
1456 if (*slot == NULL)
1457 *slot = xstrdup (decoded);
1458 *resultp = *slot;
1459 }
1460 }
1461
1462 return *resultp;
1463 }
1464
1465 static char *
1466 ada_la_decode (const char *encoded, int options)
1467 {
1468 return xstrdup (ada_decode (encoded));
1469 }
1470
1471 /* Implement la_sniff_from_mangled_name for Ada. */
1472
1473 static int
1474 ada_sniff_from_mangled_name (const char *mangled, char **out)
1475 {
1476 const char *demangled = ada_decode (mangled);
1477
1478 *out = NULL;
1479
1480 if (demangled != mangled && demangled != NULL && demangled[0] != '<')
1481 {
1482 /* Set the gsymbol language to Ada, but still return 0.
1483 Two reasons for that:
1484
1485 1. For Ada, we prefer computing the symbol's decoded name
1486 on the fly rather than pre-compute it, in order to save
1487 memory (Ada projects are typically very large).
1488
1489 2. There are some areas in the definition of the GNAT
1490 encoding where, with a bit of bad luck, we might be able
1491 to decode a non-Ada symbol, generating an incorrect
1492 demangled name (Eg: names ending with "TB" for instance
1493 are identified as task bodies and so stripped from
1494 the decoded name returned).
1495
1496 Returning 1, here, but not setting *DEMANGLED, helps us get a
1497 little bit of the best of both worlds. Because we're last,
1498 we should not affect any of the other languages that were
1499 able to demangle the symbol before us; we get to correctly
1500 tag Ada symbols as such; and even if we incorrectly tagged a
1501 non-Ada symbol, which should be rare, any routing through the
1502 Ada language should be transparent (Ada tries to behave much
1503 like C/C++ with non-Ada symbols). */
1504 return 1;
1505 }
1506
1507 return 0;
1508 }
1509
1510 \f
1511
1512 /* Arrays */
1513
1514 /* Assuming that INDEX_DESC_TYPE is an ___XA structure, a structure
1515 generated by the GNAT compiler to describe the index type used
1516 for each dimension of an array, check whether it follows the latest
1517 known encoding. If not, fix it up to conform to the latest encoding.
1518 Otherwise, do nothing. This function also does nothing if
1519 INDEX_DESC_TYPE is NULL.
1520
1521 The GNAT encoding used to describle the array index type evolved a bit.
1522 Initially, the information would be provided through the name of each
1523 field of the structure type only, while the type of these fields was
1524 described as unspecified and irrelevant. The debugger was then expected
1525 to perform a global type lookup using the name of that field in order
1526 to get access to the full index type description. Because these global
1527 lookups can be very expensive, the encoding was later enhanced to make
1528 the global lookup unnecessary by defining the field type as being
1529 the full index type description.
1530
1531 The purpose of this routine is to allow us to support older versions
1532 of the compiler by detecting the use of the older encoding, and by
1533 fixing up the INDEX_DESC_TYPE to follow the new one (at this point,
1534 we essentially replace each field's meaningless type by the associated
1535 index subtype). */
1536
1537 void
1538 ada_fixup_array_indexes_type (struct type *index_desc_type)
1539 {
1540 int i;
1541
1542 if (index_desc_type == NULL)
1543 return;
1544 gdb_assert (TYPE_NFIELDS (index_desc_type) > 0);
1545
1546 /* Check if INDEX_DESC_TYPE follows the older encoding (it is sufficient
1547 to check one field only, no need to check them all). If not, return
1548 now.
1549
1550 If our INDEX_DESC_TYPE was generated using the older encoding,
1551 the field type should be a meaningless integer type whose name
1552 is not equal to the field name. */
1553 if (TYPE_NAME (TYPE_FIELD_TYPE (index_desc_type, 0)) != NULL
1554 && strcmp (TYPE_NAME (TYPE_FIELD_TYPE (index_desc_type, 0)),
1555 TYPE_FIELD_NAME (index_desc_type, 0)) == 0)
1556 return;
1557
1558 /* Fixup each field of INDEX_DESC_TYPE. */
1559 for (i = 0; i < TYPE_NFIELDS (index_desc_type); i++)
1560 {
1561 const char *name = TYPE_FIELD_NAME (index_desc_type, i);
1562 struct type *raw_type = ada_check_typedef (ada_find_any_type (name));
1563
1564 if (raw_type)
1565 TYPE_FIELD_TYPE (index_desc_type, i) = raw_type;
1566 }
1567 }
1568
1569 /* Names of MAX_ADA_DIMENS bounds in P_BOUNDS fields of array descriptors. */
1570
1571 static const char *bound_name[] = {
1572 "LB0", "UB0", "LB1", "UB1", "LB2", "UB2", "LB3", "UB3",
1573 "LB4", "UB4", "LB5", "UB5", "LB6", "UB6", "LB7", "UB7"
1574 };
1575
1576 /* Maximum number of array dimensions we are prepared to handle. */
1577
1578 #define MAX_ADA_DIMENS (sizeof(bound_name) / (2*sizeof(char *)))
1579
1580
1581 /* The desc_* routines return primitive portions of array descriptors
1582 (fat pointers). */
1583
1584 /* The descriptor or array type, if any, indicated by TYPE; removes
1585 level of indirection, if needed. */
1586
1587 static struct type *
1588 desc_base_type (struct type *type)
1589 {
1590 if (type == NULL)
1591 return NULL;
1592 type = ada_check_typedef (type);
1593 if (TYPE_CODE (type) == TYPE_CODE_TYPEDEF)
1594 type = ada_typedef_target_type (type);
1595
1596 if (type != NULL
1597 && (TYPE_CODE (type) == TYPE_CODE_PTR
1598 || TYPE_CODE (type) == TYPE_CODE_REF))
1599 return ada_check_typedef (TYPE_TARGET_TYPE (type));
1600 else
1601 return type;
1602 }
1603
1604 /* True iff TYPE indicates a "thin" array pointer type. */
1605
1606 static int
1607 is_thin_pntr (struct type *type)
1608 {
1609 return
1610 is_suffix (ada_type_name (desc_base_type (type)), "___XUT")
1611 || is_suffix (ada_type_name (desc_base_type (type)), "___XUT___XVE");
1612 }
1613
1614 /* The descriptor type for thin pointer type TYPE. */
1615
1616 static struct type *
1617 thin_descriptor_type (struct type *type)
1618 {
1619 struct type *base_type = desc_base_type (type);
1620
1621 if (base_type == NULL)
1622 return NULL;
1623 if (is_suffix (ada_type_name (base_type), "___XVE"))
1624 return base_type;
1625 else
1626 {
1627 struct type *alt_type = ada_find_parallel_type (base_type, "___XVE");
1628
1629 if (alt_type == NULL)
1630 return base_type;
1631 else
1632 return alt_type;
1633 }
1634 }
1635
1636 /* A pointer to the array data for thin-pointer value VAL. */
1637
1638 static struct value *
1639 thin_data_pntr (struct value *val)
1640 {
1641 struct type *type = ada_check_typedef (value_type (val));
1642 struct type *data_type = desc_data_target_type (thin_descriptor_type (type));
1643
1644 data_type = lookup_pointer_type (data_type);
1645
1646 if (TYPE_CODE (type) == TYPE_CODE_PTR)
1647 return value_cast (data_type, value_copy (val));
1648 else
1649 return value_from_longest (data_type, value_address (val));
1650 }
1651
1652 /* True iff TYPE indicates a "thick" array pointer type. */
1653
1654 static int
1655 is_thick_pntr (struct type *type)
1656 {
1657 type = desc_base_type (type);
1658 return (type != NULL && TYPE_CODE (type) == TYPE_CODE_STRUCT
1659 && lookup_struct_elt_type (type, "P_BOUNDS", 1) != NULL);
1660 }
1661
1662 /* If TYPE is the type of an array descriptor (fat or thin pointer) or a
1663 pointer to one, the type of its bounds data; otherwise, NULL. */
1664
1665 static struct type *
1666 desc_bounds_type (struct type *type)
1667 {
1668 struct type *r;
1669
1670 type = desc_base_type (type);
1671
1672 if (type == NULL)
1673 return NULL;
1674 else if (is_thin_pntr (type))
1675 {
1676 type = thin_descriptor_type (type);
1677 if (type == NULL)
1678 return NULL;
1679 r = lookup_struct_elt_type (type, "BOUNDS", 1);
1680 if (r != NULL)
1681 return ada_check_typedef (r);
1682 }
1683 else if (TYPE_CODE (type) == TYPE_CODE_STRUCT)
1684 {
1685 r = lookup_struct_elt_type (type, "P_BOUNDS", 1);
1686 if (r != NULL)
1687 return ada_check_typedef (TYPE_TARGET_TYPE (ada_check_typedef (r)));
1688 }
1689 return NULL;
1690 }
1691
1692 /* If ARR is an array descriptor (fat or thin pointer), or pointer to
1693 one, a pointer to its bounds data. Otherwise NULL. */
1694
1695 static struct value *
1696 desc_bounds (struct value *arr)
1697 {
1698 struct type *type = ada_check_typedef (value_type (arr));
1699
1700 if (is_thin_pntr (type))
1701 {
1702 struct type *bounds_type =
1703 desc_bounds_type (thin_descriptor_type (type));
1704 LONGEST addr;
1705
1706 if (bounds_type == NULL)
1707 error (_("Bad GNAT array descriptor"));
1708
1709 /* NOTE: The following calculation is not really kosher, but
1710 since desc_type is an XVE-encoded type (and shouldn't be),
1711 the correct calculation is a real pain. FIXME (and fix GCC). */
1712 if (TYPE_CODE (type) == TYPE_CODE_PTR)
1713 addr = value_as_long (arr);
1714 else
1715 addr = value_address (arr);
1716
1717 return
1718 value_from_longest (lookup_pointer_type (bounds_type),
1719 addr - TYPE_LENGTH (bounds_type));
1720 }
1721
1722 else if (is_thick_pntr (type))
1723 {
1724 struct value *p_bounds = value_struct_elt (&arr, NULL, "P_BOUNDS", NULL,
1725 _("Bad GNAT array descriptor"));
1726 struct type *p_bounds_type = value_type (p_bounds);
1727
1728 if (p_bounds_type
1729 && TYPE_CODE (p_bounds_type) == TYPE_CODE_PTR)
1730 {
1731 struct type *target_type = TYPE_TARGET_TYPE (p_bounds_type);
1732
1733 if (TYPE_STUB (target_type))
1734 p_bounds = value_cast (lookup_pointer_type
1735 (ada_check_typedef (target_type)),
1736 p_bounds);
1737 }
1738 else
1739 error (_("Bad GNAT array descriptor"));
1740
1741 return p_bounds;
1742 }
1743 else
1744 return NULL;
1745 }
1746
1747 /* If TYPE is the type of an array-descriptor (fat pointer), the bit
1748 position of the field containing the address of the bounds data. */
1749
1750 static int
1751 fat_pntr_bounds_bitpos (struct type *type)
1752 {
1753 return TYPE_FIELD_BITPOS (desc_base_type (type), 1);
1754 }
1755
1756 /* If TYPE is the type of an array-descriptor (fat pointer), the bit
1757 size of the field containing the address of the bounds data. */
1758
1759 static int
1760 fat_pntr_bounds_bitsize (struct type *type)
1761 {
1762 type = desc_base_type (type);
1763
1764 if (TYPE_FIELD_BITSIZE (type, 1) > 0)
1765 return TYPE_FIELD_BITSIZE (type, 1);
1766 else
1767 return 8 * TYPE_LENGTH (ada_check_typedef (TYPE_FIELD_TYPE (type, 1)));
1768 }
1769
1770 /* If TYPE is the type of an array descriptor (fat or thin pointer) or a
1771 pointer to one, the type of its array data (a array-with-no-bounds type);
1772 otherwise, NULL. Use ada_type_of_array to get an array type with bounds
1773 data. */
1774
1775 static struct type *
1776 desc_data_target_type (struct type *type)
1777 {
1778 type = desc_base_type (type);
1779
1780 /* NOTE: The following is bogus; see comment in desc_bounds. */
1781 if (is_thin_pntr (type))
1782 return desc_base_type (TYPE_FIELD_TYPE (thin_descriptor_type (type), 1));
1783 else if (is_thick_pntr (type))
1784 {
1785 struct type *data_type = lookup_struct_elt_type (type, "P_ARRAY", 1);
1786
1787 if (data_type
1788 && TYPE_CODE (ada_check_typedef (data_type)) == TYPE_CODE_PTR)
1789 return ada_check_typedef (TYPE_TARGET_TYPE (data_type));
1790 }
1791
1792 return NULL;
1793 }
1794
1795 /* If ARR is an array descriptor (fat or thin pointer), a pointer to
1796 its array data. */
1797
1798 static struct value *
1799 desc_data (struct value *arr)
1800 {
1801 struct type *type = value_type (arr);
1802
1803 if (is_thin_pntr (type))
1804 return thin_data_pntr (arr);
1805 else if (is_thick_pntr (type))
1806 return value_struct_elt (&arr, NULL, "P_ARRAY", NULL,
1807 _("Bad GNAT array descriptor"));
1808 else
1809 return NULL;
1810 }
1811
1812
1813 /* If TYPE is the type of an array-descriptor (fat pointer), the bit
1814 position of the field containing the address of the data. */
1815
1816 static int
1817 fat_pntr_data_bitpos (struct type *type)
1818 {
1819 return TYPE_FIELD_BITPOS (desc_base_type (type), 0);
1820 }
1821
1822 /* If TYPE is the type of an array-descriptor (fat pointer), the bit
1823 size of the field containing the address of the data. */
1824
1825 static int
1826 fat_pntr_data_bitsize (struct type *type)
1827 {
1828 type = desc_base_type (type);
1829
1830 if (TYPE_FIELD_BITSIZE (type, 0) > 0)
1831 return TYPE_FIELD_BITSIZE (type, 0);
1832 else
1833 return TARGET_CHAR_BIT * TYPE_LENGTH (TYPE_FIELD_TYPE (type, 0));
1834 }
1835
1836 /* If BOUNDS is an array-bounds structure (or pointer to one), return
1837 the Ith lower bound stored in it, if WHICH is 0, and the Ith upper
1838 bound, if WHICH is 1. The first bound is I=1. */
1839
1840 static struct value *
1841 desc_one_bound (struct value *bounds, int i, int which)
1842 {
1843 return value_struct_elt (&bounds, NULL, bound_name[2 * i + which - 2], NULL,
1844 _("Bad GNAT array descriptor bounds"));
1845 }
1846
1847 /* If BOUNDS is an array-bounds structure type, return the bit position
1848 of the Ith lower bound stored in it, if WHICH is 0, and the Ith upper
1849 bound, if WHICH is 1. The first bound is I=1. */
1850
1851 static int
1852 desc_bound_bitpos (struct type *type, int i, int which)
1853 {
1854 return TYPE_FIELD_BITPOS (desc_base_type (type), 2 * i + which - 2);
1855 }
1856
1857 /* If BOUNDS is an array-bounds structure type, return the bit field size
1858 of the Ith lower bound stored in it, if WHICH is 0, and the Ith upper
1859 bound, if WHICH is 1. The first bound is I=1. */
1860
1861 static int
1862 desc_bound_bitsize (struct type *type, int i, int which)
1863 {
1864 type = desc_base_type (type);
1865
1866 if (TYPE_FIELD_BITSIZE (type, 2 * i + which - 2) > 0)
1867 return TYPE_FIELD_BITSIZE (type, 2 * i + which - 2);
1868 else
1869 return 8 * TYPE_LENGTH (TYPE_FIELD_TYPE (type, 2 * i + which - 2));
1870 }
1871
1872 /* If TYPE is the type of an array-bounds structure, the type of its
1873 Ith bound (numbering from 1). Otherwise, NULL. */
1874
1875 static struct type *
1876 desc_index_type (struct type *type, int i)
1877 {
1878 type = desc_base_type (type);
1879
1880 if (TYPE_CODE (type) == TYPE_CODE_STRUCT)
1881 return lookup_struct_elt_type (type, bound_name[2 * i - 2], 1);
1882 else
1883 return NULL;
1884 }
1885
1886 /* The number of index positions in the array-bounds type TYPE.
1887 Return 0 if TYPE is NULL. */
1888
1889 static int
1890 desc_arity (struct type *type)
1891 {
1892 type = desc_base_type (type);
1893
1894 if (type != NULL)
1895 return TYPE_NFIELDS (type) / 2;
1896 return 0;
1897 }
1898
1899 /* Non-zero iff TYPE is a simple array type (not a pointer to one) or
1900 an array descriptor type (representing an unconstrained array
1901 type). */
1902
1903 static int
1904 ada_is_direct_array_type (struct type *type)
1905 {
1906 if (type == NULL)
1907 return 0;
1908 type = ada_check_typedef (type);
1909 return (TYPE_CODE (type) == TYPE_CODE_ARRAY
1910 || ada_is_array_descriptor_type (type));
1911 }
1912
1913 /* Non-zero iff TYPE represents any kind of array in Ada, or a pointer
1914 * to one. */
1915
1916 static int
1917 ada_is_array_type (struct type *type)
1918 {
1919 while (type != NULL
1920 && (TYPE_CODE (type) == TYPE_CODE_PTR
1921 || TYPE_CODE (type) == TYPE_CODE_REF))
1922 type = TYPE_TARGET_TYPE (type);
1923 return ada_is_direct_array_type (type);
1924 }
1925
1926 /* Non-zero iff TYPE is a simple array type or pointer to one. */
1927
1928 int
1929 ada_is_simple_array_type (struct type *type)
1930 {
1931 if (type == NULL)
1932 return 0;
1933 type = ada_check_typedef (type);
1934 return (TYPE_CODE (type) == TYPE_CODE_ARRAY
1935 || (TYPE_CODE (type) == TYPE_CODE_PTR
1936 && TYPE_CODE (ada_check_typedef (TYPE_TARGET_TYPE (type)))
1937 == TYPE_CODE_ARRAY));
1938 }
1939
1940 /* Non-zero iff TYPE belongs to a GNAT array descriptor. */
1941
1942 int
1943 ada_is_array_descriptor_type (struct type *type)
1944 {
1945 struct type *data_type = desc_data_target_type (type);
1946
1947 if (type == NULL)
1948 return 0;
1949 type = ada_check_typedef (type);
1950 return (data_type != NULL
1951 && TYPE_CODE (data_type) == TYPE_CODE_ARRAY
1952 && desc_arity (desc_bounds_type (type)) > 0);
1953 }
1954
1955 /* Non-zero iff type is a partially mal-formed GNAT array
1956 descriptor. FIXME: This is to compensate for some problems with
1957 debugging output from GNAT. Re-examine periodically to see if it
1958 is still needed. */
1959
1960 int
1961 ada_is_bogus_array_descriptor (struct type *type)
1962 {
1963 return
1964 type != NULL
1965 && TYPE_CODE (type) == TYPE_CODE_STRUCT
1966 && (lookup_struct_elt_type (type, "P_BOUNDS", 1) != NULL
1967 || lookup_struct_elt_type (type, "P_ARRAY", 1) != NULL)
1968 && !ada_is_array_descriptor_type (type);
1969 }
1970
1971
1972 /* If ARR has a record type in the form of a standard GNAT array descriptor,
1973 (fat pointer) returns the type of the array data described---specifically,
1974 a pointer-to-array type. If BOUNDS is non-zero, the bounds data are filled
1975 in from the descriptor; otherwise, they are left unspecified. If
1976 the ARR denotes a null array descriptor and BOUNDS is non-zero,
1977 returns NULL. The result is simply the type of ARR if ARR is not
1978 a descriptor. */
1979 struct type *
1980 ada_type_of_array (struct value *arr, int bounds)
1981 {
1982 if (ada_is_constrained_packed_array_type (value_type (arr)))
1983 return decode_constrained_packed_array_type (value_type (arr));
1984
1985 if (!ada_is_array_descriptor_type (value_type (arr)))
1986 return value_type (arr);
1987
1988 if (!bounds)
1989 {
1990 struct type *array_type =
1991 ada_check_typedef (desc_data_target_type (value_type (arr)));
1992
1993 if (ada_is_unconstrained_packed_array_type (value_type (arr)))
1994 TYPE_FIELD_BITSIZE (array_type, 0) =
1995 decode_packed_array_bitsize (value_type (arr));
1996
1997 return array_type;
1998 }
1999 else
2000 {
2001 struct type *elt_type;
2002 int arity;
2003 struct value *descriptor;
2004
2005 elt_type = ada_array_element_type (value_type (arr), -1);
2006 arity = ada_array_arity (value_type (arr));
2007
2008 if (elt_type == NULL || arity == 0)
2009 return ada_check_typedef (value_type (arr));
2010
2011 descriptor = desc_bounds (arr);
2012 if (value_as_long (descriptor) == 0)
2013 return NULL;
2014 while (arity > 0)
2015 {
2016 struct type *range_type = alloc_type_copy (value_type (arr));
2017 struct type *array_type = alloc_type_copy (value_type (arr));
2018 struct value *low = desc_one_bound (descriptor, arity, 0);
2019 struct value *high = desc_one_bound (descriptor, arity, 1);
2020
2021 arity -= 1;
2022 create_static_range_type (range_type, value_type (low),
2023 longest_to_int (value_as_long (low)),
2024 longest_to_int (value_as_long (high)));
2025 elt_type = create_array_type (array_type, elt_type, range_type);
2026
2027 if (ada_is_unconstrained_packed_array_type (value_type (arr)))
2028 {
2029 /* We need to store the element packed bitsize, as well as
2030 recompute the array size, because it was previously
2031 computed based on the unpacked element size. */
2032 LONGEST lo = value_as_long (low);
2033 LONGEST hi = value_as_long (high);
2034
2035 TYPE_FIELD_BITSIZE (elt_type, 0) =
2036 decode_packed_array_bitsize (value_type (arr));
2037 /* If the array has no element, then the size is already
2038 zero, and does not need to be recomputed. */
2039 if (lo < hi)
2040 {
2041 int array_bitsize =
2042 (hi - lo + 1) * TYPE_FIELD_BITSIZE (elt_type, 0);
2043
2044 TYPE_LENGTH (array_type) = (array_bitsize + 7) / 8;
2045 }
2046 }
2047 }
2048
2049 return lookup_pointer_type (elt_type);
2050 }
2051 }
2052
2053 /* If ARR does not represent an array, returns ARR unchanged.
2054 Otherwise, returns either a standard GDB array with bounds set
2055 appropriately or, if ARR is a non-null fat pointer, a pointer to a standard
2056 GDB array. Returns NULL if ARR is a null fat pointer. */
2057
2058 struct value *
2059 ada_coerce_to_simple_array_ptr (struct value *arr)
2060 {
2061 if (ada_is_array_descriptor_type (value_type (arr)))
2062 {
2063 struct type *arrType = ada_type_of_array (arr, 1);
2064
2065 if (arrType == NULL)
2066 return NULL;
2067 return value_cast (arrType, value_copy (desc_data (arr)));
2068 }
2069 else if (ada_is_constrained_packed_array_type (value_type (arr)))
2070 return decode_constrained_packed_array (arr);
2071 else
2072 return arr;
2073 }
2074
2075 /* If ARR does not represent an array, returns ARR unchanged.
2076 Otherwise, returns a standard GDB array describing ARR (which may
2077 be ARR itself if it already is in the proper form). */
2078
2079 struct value *
2080 ada_coerce_to_simple_array (struct value *arr)
2081 {
2082 if (ada_is_array_descriptor_type (value_type (arr)))
2083 {
2084 struct value *arrVal = ada_coerce_to_simple_array_ptr (arr);
2085
2086 if (arrVal == NULL)
2087 error (_("Bounds unavailable for null array pointer."));
2088 ada_ensure_varsize_limit (TYPE_TARGET_TYPE (value_type (arrVal)));
2089 return value_ind (arrVal);
2090 }
2091 else if (ada_is_constrained_packed_array_type (value_type (arr)))
2092 return decode_constrained_packed_array (arr);
2093 else
2094 return arr;
2095 }
2096
2097 /* If TYPE represents a GNAT array type, return it translated to an
2098 ordinary GDB array type (possibly with BITSIZE fields indicating
2099 packing). For other types, is the identity. */
2100
2101 struct type *
2102 ada_coerce_to_simple_array_type (struct type *type)
2103 {
2104 if (ada_is_constrained_packed_array_type (type))
2105 return decode_constrained_packed_array_type (type);
2106
2107 if (ada_is_array_descriptor_type (type))
2108 return ada_check_typedef (desc_data_target_type (type));
2109
2110 return type;
2111 }
2112
2113 /* Non-zero iff TYPE represents a standard GNAT packed-array type. */
2114
2115 static int
2116 ada_is_packed_array_type (struct type *type)
2117 {
2118 if (type == NULL)
2119 return 0;
2120 type = desc_base_type (type);
2121 type = ada_check_typedef (type);
2122 return
2123 ada_type_name (type) != NULL
2124 && strstr (ada_type_name (type), "___XP") != NULL;
2125 }
2126
2127 /* Non-zero iff TYPE represents a standard GNAT constrained
2128 packed-array type. */
2129
2130 int
2131 ada_is_constrained_packed_array_type (struct type *type)
2132 {
2133 return ada_is_packed_array_type (type)
2134 && !ada_is_array_descriptor_type (type);
2135 }
2136
2137 /* Non-zero iff TYPE represents an array descriptor for a
2138 unconstrained packed-array type. */
2139
2140 static int
2141 ada_is_unconstrained_packed_array_type (struct type *type)
2142 {
2143 return ada_is_packed_array_type (type)
2144 && ada_is_array_descriptor_type (type);
2145 }
2146
2147 /* Given that TYPE encodes a packed array type (constrained or unconstrained),
2148 return the size of its elements in bits. */
2149
2150 static long
2151 decode_packed_array_bitsize (struct type *type)
2152 {
2153 const char *raw_name;
2154 const char *tail;
2155 long bits;
2156
2157 /* Access to arrays implemented as fat pointers are encoded as a typedef
2158 of the fat pointer type. We need the name of the fat pointer type
2159 to do the decoding, so strip the typedef layer. */
2160 if (TYPE_CODE (type) == TYPE_CODE_TYPEDEF)
2161 type = ada_typedef_target_type (type);
2162
2163 raw_name = ada_type_name (ada_check_typedef (type));
2164 if (!raw_name)
2165 raw_name = ada_type_name (desc_base_type (type));
2166
2167 if (!raw_name)
2168 return 0;
2169
2170 tail = strstr (raw_name, "___XP");
2171 gdb_assert (tail != NULL);
2172
2173 if (sscanf (tail + sizeof ("___XP") - 1, "%ld", &bits) != 1)
2174 {
2175 lim_warning
2176 (_("could not understand bit size information on packed array"));
2177 return 0;
2178 }
2179
2180 return bits;
2181 }
2182
2183 /* Given that TYPE is a standard GDB array type with all bounds filled
2184 in, and that the element size of its ultimate scalar constituents
2185 (that is, either its elements, or, if it is an array of arrays, its
2186 elements' elements, etc.) is *ELT_BITS, return an identical type,
2187 but with the bit sizes of its elements (and those of any
2188 constituent arrays) recorded in the BITSIZE components of its
2189 TYPE_FIELD_BITSIZE values, and with *ELT_BITS set to its total size
2190 in bits.
2191
2192 Note that, for arrays whose index type has an XA encoding where
2193 a bound references a record discriminant, getting that discriminant,
2194 and therefore the actual value of that bound, is not possible
2195 because none of the given parameters gives us access to the record.
2196 This function assumes that it is OK in the context where it is being
2197 used to return an array whose bounds are still dynamic and where
2198 the length is arbitrary. */
2199
2200 static struct type *
2201 constrained_packed_array_type (struct type *type, long *elt_bits)
2202 {
2203 struct type *new_elt_type;
2204 struct type *new_type;
2205 struct type *index_type_desc;
2206 struct type *index_type;
2207 LONGEST low_bound, high_bound;
2208
2209 type = ada_check_typedef (type);
2210 if (TYPE_CODE (type) != TYPE_CODE_ARRAY)
2211 return type;
2212
2213 index_type_desc = ada_find_parallel_type (type, "___XA");
2214 if (index_type_desc)
2215 index_type = to_fixed_range_type (TYPE_FIELD_TYPE (index_type_desc, 0),
2216 NULL);
2217 else
2218 index_type = TYPE_INDEX_TYPE (type);
2219
2220 new_type = alloc_type_copy (type);
2221 new_elt_type =
2222 constrained_packed_array_type (ada_check_typedef (TYPE_TARGET_TYPE (type)),
2223 elt_bits);
2224 create_array_type (new_type, new_elt_type, index_type);
2225 TYPE_FIELD_BITSIZE (new_type, 0) = *elt_bits;
2226 TYPE_NAME (new_type) = ada_type_name (type);
2227
2228 if ((TYPE_CODE (check_typedef (index_type)) == TYPE_CODE_RANGE
2229 && is_dynamic_type (check_typedef (index_type)))
2230 || get_discrete_bounds (index_type, &low_bound, &high_bound) < 0)
2231 low_bound = high_bound = 0;
2232 if (high_bound < low_bound)
2233 *elt_bits = TYPE_LENGTH (new_type) = 0;
2234 else
2235 {
2236 *elt_bits *= (high_bound - low_bound + 1);
2237 TYPE_LENGTH (new_type) =
2238 (*elt_bits + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT;
2239 }
2240
2241 TYPE_FIXED_INSTANCE (new_type) = 1;
2242 return new_type;
2243 }
2244
2245 /* The array type encoded by TYPE, where
2246 ada_is_constrained_packed_array_type (TYPE). */
2247
2248 static struct type *
2249 decode_constrained_packed_array_type (struct type *type)
2250 {
2251 const char *raw_name = ada_type_name (ada_check_typedef (type));
2252 char *name;
2253 const char *tail;
2254 struct type *shadow_type;
2255 long bits;
2256
2257 if (!raw_name)
2258 raw_name = ada_type_name (desc_base_type (type));
2259
2260 if (!raw_name)
2261 return NULL;
2262
2263 name = (char *) alloca (strlen (raw_name) + 1);
2264 tail = strstr (raw_name, "___XP");
2265 type = desc_base_type (type);
2266
2267 memcpy (name, raw_name, tail - raw_name);
2268 name[tail - raw_name] = '\000';
2269
2270 shadow_type = ada_find_parallel_type_with_name (type, name);
2271
2272 if (shadow_type == NULL)
2273 {
2274 lim_warning (_("could not find bounds information on packed array"));
2275 return NULL;
2276 }
2277 shadow_type = check_typedef (shadow_type);
2278
2279 if (TYPE_CODE (shadow_type) != TYPE_CODE_ARRAY)
2280 {
2281 lim_warning (_("could not understand bounds "
2282 "information on packed array"));
2283 return NULL;
2284 }
2285
2286 bits = decode_packed_array_bitsize (type);
2287 return constrained_packed_array_type (shadow_type, &bits);
2288 }
2289
2290 /* Given that ARR is a struct value *indicating a GNAT constrained packed
2291 array, returns a simple array that denotes that array. Its type is a
2292 standard GDB array type except that the BITSIZEs of the array
2293 target types are set to the number of bits in each element, and the
2294 type length is set appropriately. */
2295
2296 static struct value *
2297 decode_constrained_packed_array (struct value *arr)
2298 {
2299 struct type *type;
2300
2301 /* If our value is a pointer, then dereference it. Likewise if
2302 the value is a reference. Make sure that this operation does not
2303 cause the target type to be fixed, as this would indirectly cause
2304 this array to be decoded. The rest of the routine assumes that
2305 the array hasn't been decoded yet, so we use the basic "coerce_ref"
2306 and "value_ind" routines to perform the dereferencing, as opposed
2307 to using "ada_coerce_ref" or "ada_value_ind". */
2308 arr = coerce_ref (arr);
2309 if (TYPE_CODE (ada_check_typedef (value_type (arr))) == TYPE_CODE_PTR)
2310 arr = value_ind (arr);
2311
2312 type = decode_constrained_packed_array_type (value_type (arr));
2313 if (type == NULL)
2314 {
2315 error (_("can't unpack array"));
2316 return NULL;
2317 }
2318
2319 if (gdbarch_bits_big_endian (get_type_arch (value_type (arr)))
2320 && ada_is_modular_type (value_type (arr)))
2321 {
2322 /* This is a (right-justified) modular type representing a packed
2323 array with no wrapper. In order to interpret the value through
2324 the (left-justified) packed array type we just built, we must
2325 first left-justify it. */
2326 int bit_size, bit_pos;
2327 ULONGEST mod;
2328
2329 mod = ada_modulus (value_type (arr)) - 1;
2330 bit_size = 0;
2331 while (mod > 0)
2332 {
2333 bit_size += 1;
2334 mod >>= 1;
2335 }
2336 bit_pos = HOST_CHAR_BIT * TYPE_LENGTH (value_type (arr)) - bit_size;
2337 arr = ada_value_primitive_packed_val (arr, NULL,
2338 bit_pos / HOST_CHAR_BIT,
2339 bit_pos % HOST_CHAR_BIT,
2340 bit_size,
2341 type);
2342 }
2343
2344 return coerce_unspec_val_to_type (arr, type);
2345 }
2346
2347
2348 /* The value of the element of packed array ARR at the ARITY indices
2349 given in IND. ARR must be a simple array. */
2350
2351 static struct value *
2352 value_subscript_packed (struct value *arr, int arity, struct value **ind)
2353 {
2354 int i;
2355 int bits, elt_off, bit_off;
2356 long elt_total_bit_offset;
2357 struct type *elt_type;
2358 struct value *v;
2359
2360 bits = 0;
2361 elt_total_bit_offset = 0;
2362 elt_type = ada_check_typedef (value_type (arr));
2363 for (i = 0; i < arity; i += 1)
2364 {
2365 if (TYPE_CODE (elt_type) != TYPE_CODE_ARRAY
2366 || TYPE_FIELD_BITSIZE (elt_type, 0) == 0)
2367 error
2368 (_("attempt to do packed indexing of "
2369 "something other than a packed array"));
2370 else
2371 {
2372 struct type *range_type = TYPE_INDEX_TYPE (elt_type);
2373 LONGEST lowerbound, upperbound;
2374 LONGEST idx;
2375
2376 if (get_discrete_bounds (range_type, &lowerbound, &upperbound) < 0)
2377 {
2378 lim_warning (_("don't know bounds of array"));
2379 lowerbound = upperbound = 0;
2380 }
2381
2382 idx = pos_atr (ind[i]);
2383 if (idx < lowerbound || idx > upperbound)
2384 lim_warning (_("packed array index %ld out of bounds"),
2385 (long) idx);
2386 bits = TYPE_FIELD_BITSIZE (elt_type, 0);
2387 elt_total_bit_offset += (idx - lowerbound) * bits;
2388 elt_type = ada_check_typedef (TYPE_TARGET_TYPE (elt_type));
2389 }
2390 }
2391 elt_off = elt_total_bit_offset / HOST_CHAR_BIT;
2392 bit_off = elt_total_bit_offset % HOST_CHAR_BIT;
2393
2394 v = ada_value_primitive_packed_val (arr, NULL, elt_off, bit_off,
2395 bits, elt_type);
2396 return v;
2397 }
2398
2399 /* Non-zero iff TYPE includes negative integer values. */
2400
2401 static int
2402 has_negatives (struct type *type)
2403 {
2404 switch (TYPE_CODE (type))
2405 {
2406 default:
2407 return 0;
2408 case TYPE_CODE_INT:
2409 return !TYPE_UNSIGNED (type);
2410 case TYPE_CODE_RANGE:
2411 return TYPE_LOW_BOUND (type) < 0;
2412 }
2413 }
2414
2415 /* With SRC being a buffer containing BIT_SIZE bits of data at BIT_OFFSET,
2416 unpack that data into UNPACKED. UNPACKED_LEN is the size in bytes of
2417 the unpacked buffer.
2418
2419 The size of the unpacked buffer (UNPACKED_LEN) is expected to be large
2420 enough to contain at least BIT_OFFSET bits. If not, an error is raised.
2421
2422 IS_BIG_ENDIAN is nonzero if the data is stored in big endian mode,
2423 zero otherwise.
2424
2425 IS_SIGNED_TYPE is nonzero if the data corresponds to a signed type.
2426
2427 IS_SCALAR is nonzero if the data corresponds to a signed type. */
2428
2429 static void
2430 ada_unpack_from_contents (const gdb_byte *src, int bit_offset, int bit_size,
2431 gdb_byte *unpacked, int unpacked_len,
2432 int is_big_endian, int is_signed_type,
2433 int is_scalar)
2434 {
2435 int src_len = (bit_size + bit_offset + HOST_CHAR_BIT - 1) / 8;
2436 int src_idx; /* Index into the source area */
2437 int src_bytes_left; /* Number of source bytes left to process. */
2438 int srcBitsLeft; /* Number of source bits left to move */
2439 int unusedLS; /* Number of bits in next significant
2440 byte of source that are unused */
2441
2442 int unpacked_idx; /* Index into the unpacked buffer */
2443 int unpacked_bytes_left; /* Number of bytes left to set in unpacked. */
2444
2445 unsigned long accum; /* Staging area for bits being transferred */
2446 int accumSize; /* Number of meaningful bits in accum */
2447 unsigned char sign;
2448
2449 /* Transmit bytes from least to most significant; delta is the direction
2450 the indices move. */
2451 int delta = is_big_endian ? -1 : 1;
2452
2453 /* Make sure that unpacked is large enough to receive the BIT_SIZE
2454 bits from SRC. .*/
2455 if ((bit_size + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT > unpacked_len)
2456 error (_("Cannot unpack %d bits into buffer of %d bytes"),
2457 bit_size, unpacked_len);
2458
2459 srcBitsLeft = bit_size;
2460 src_bytes_left = src_len;
2461 unpacked_bytes_left = unpacked_len;
2462 sign = 0;
2463
2464 if (is_big_endian)
2465 {
2466 src_idx = src_len - 1;
2467 if (is_signed_type
2468 && ((src[0] << bit_offset) & (1 << (HOST_CHAR_BIT - 1))))
2469 sign = ~0;
2470
2471 unusedLS =
2472 (HOST_CHAR_BIT - (bit_size + bit_offset) % HOST_CHAR_BIT)
2473 % HOST_CHAR_BIT;
2474
2475 if (is_scalar)
2476 {
2477 accumSize = 0;
2478 unpacked_idx = unpacked_len - 1;
2479 }
2480 else
2481 {
2482 /* Non-scalar values must be aligned at a byte boundary... */
2483 accumSize =
2484 (HOST_CHAR_BIT - bit_size % HOST_CHAR_BIT) % HOST_CHAR_BIT;
2485 /* ... And are placed at the beginning (most-significant) bytes
2486 of the target. */
2487 unpacked_idx = (bit_size + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT - 1;
2488 unpacked_bytes_left = unpacked_idx + 1;
2489 }
2490 }
2491 else
2492 {
2493 int sign_bit_offset = (bit_size + bit_offset - 1) % 8;
2494
2495 src_idx = unpacked_idx = 0;
2496 unusedLS = bit_offset;
2497 accumSize = 0;
2498
2499 if (is_signed_type && (src[src_len - 1] & (1 << sign_bit_offset)))
2500 sign = ~0;
2501 }
2502
2503 accum = 0;
2504 while (src_bytes_left > 0)
2505 {
2506 /* Mask for removing bits of the next source byte that are not
2507 part of the value. */
2508 unsigned int unusedMSMask =
2509 (1 << (srcBitsLeft >= HOST_CHAR_BIT ? HOST_CHAR_BIT : srcBitsLeft)) -
2510 1;
2511 /* Sign-extend bits for this byte. */
2512 unsigned int signMask = sign & ~unusedMSMask;
2513
2514 accum |=
2515 (((src[src_idx] >> unusedLS) & unusedMSMask) | signMask) << accumSize;
2516 accumSize += HOST_CHAR_BIT - unusedLS;
2517 if (accumSize >= HOST_CHAR_BIT)
2518 {
2519 unpacked[unpacked_idx] = accum & ~(~0UL << HOST_CHAR_BIT);
2520 accumSize -= HOST_CHAR_BIT;
2521 accum >>= HOST_CHAR_BIT;
2522 unpacked_bytes_left -= 1;
2523 unpacked_idx += delta;
2524 }
2525 srcBitsLeft -= HOST_CHAR_BIT - unusedLS;
2526 unusedLS = 0;
2527 src_bytes_left -= 1;
2528 src_idx += delta;
2529 }
2530 while (unpacked_bytes_left > 0)
2531 {
2532 accum |= sign << accumSize;
2533 unpacked[unpacked_idx] = accum & ~(~0UL << HOST_CHAR_BIT);
2534 accumSize -= HOST_CHAR_BIT;
2535 if (accumSize < 0)
2536 accumSize = 0;
2537 accum >>= HOST_CHAR_BIT;
2538 unpacked_bytes_left -= 1;
2539 unpacked_idx += delta;
2540 }
2541 }
2542
2543 /* Create a new value of type TYPE from the contents of OBJ starting
2544 at byte OFFSET, and bit offset BIT_OFFSET within that byte,
2545 proceeding for BIT_SIZE bits. If OBJ is an lval in memory, then
2546 assigning through the result will set the field fetched from.
2547 VALADDR is ignored unless OBJ is NULL, in which case,
2548 VALADDR+OFFSET must address the start of storage containing the
2549 packed value. The value returned in this case is never an lval.
2550 Assumes 0 <= BIT_OFFSET < HOST_CHAR_BIT. */
2551
2552 struct value *
2553 ada_value_primitive_packed_val (struct value *obj, const gdb_byte *valaddr,
2554 long offset, int bit_offset, int bit_size,
2555 struct type *type)
2556 {
2557 struct value *v;
2558 const gdb_byte *src; /* First byte containing data to unpack */
2559 gdb_byte *unpacked;
2560 const int is_scalar = is_scalar_type (type);
2561 const int is_big_endian = gdbarch_bits_big_endian (get_type_arch (type));
2562 gdb::byte_vector staging;
2563
2564 type = ada_check_typedef (type);
2565
2566 if (obj == NULL)
2567 src = valaddr + offset;
2568 else
2569 src = value_contents (obj) + offset;
2570
2571 if (is_dynamic_type (type))
2572 {
2573 /* The length of TYPE might by dynamic, so we need to resolve
2574 TYPE in order to know its actual size, which we then use
2575 to create the contents buffer of the value we return.
2576 The difficulty is that the data containing our object is
2577 packed, and therefore maybe not at a byte boundary. So, what
2578 we do, is unpack the data into a byte-aligned buffer, and then
2579 use that buffer as our object's value for resolving the type. */
2580 int staging_len = (bit_size + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT;
2581 staging.resize (staging_len);
2582
2583 ada_unpack_from_contents (src, bit_offset, bit_size,
2584 staging.data (), staging.size (),
2585 is_big_endian, has_negatives (type),
2586 is_scalar);
2587 type = resolve_dynamic_type (type, staging.data (), 0);
2588 if (TYPE_LENGTH (type) < (bit_size + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT)
2589 {
2590 /* This happens when the length of the object is dynamic,
2591 and is actually smaller than the space reserved for it.
2592 For instance, in an array of variant records, the bit_size
2593 we're given is the array stride, which is constant and
2594 normally equal to the maximum size of its element.
2595 But, in reality, each element only actually spans a portion
2596 of that stride. */
2597 bit_size = TYPE_LENGTH (type) * HOST_CHAR_BIT;
2598 }
2599 }
2600
2601 if (obj == NULL)
2602 {
2603 v = allocate_value (type);
2604 src = valaddr + offset;
2605 }
2606 else if (VALUE_LVAL (obj) == lval_memory && value_lazy (obj))
2607 {
2608 int src_len = (bit_size + bit_offset + HOST_CHAR_BIT - 1) / 8;
2609 gdb_byte *buf;
2610
2611 v = value_at (type, value_address (obj) + offset);
2612 buf = (gdb_byte *) alloca (src_len);
2613 read_memory (value_address (v), buf, src_len);
2614 src = buf;
2615 }
2616 else
2617 {
2618 v = allocate_value (type);
2619 src = value_contents (obj) + offset;
2620 }
2621
2622 if (obj != NULL)
2623 {
2624 long new_offset = offset;
2625
2626 set_value_component_location (v, obj);
2627 set_value_bitpos (v, bit_offset + value_bitpos (obj));
2628 set_value_bitsize (v, bit_size);
2629 if (value_bitpos (v) >= HOST_CHAR_BIT)
2630 {
2631 ++new_offset;
2632 set_value_bitpos (v, value_bitpos (v) - HOST_CHAR_BIT);
2633 }
2634 set_value_offset (v, new_offset);
2635
2636 /* Also set the parent value. This is needed when trying to
2637 assign a new value (in inferior memory). */
2638 set_value_parent (v, obj);
2639 }
2640 else
2641 set_value_bitsize (v, bit_size);
2642 unpacked = value_contents_writeable (v);
2643
2644 if (bit_size == 0)
2645 {
2646 memset (unpacked, 0, TYPE_LENGTH (type));
2647 return v;
2648 }
2649
2650 if (staging.size () == TYPE_LENGTH (type))
2651 {
2652 /* Small short-cut: If we've unpacked the data into a buffer
2653 of the same size as TYPE's length, then we can reuse that,
2654 instead of doing the unpacking again. */
2655 memcpy (unpacked, staging.data (), staging.size ());
2656 }
2657 else
2658 ada_unpack_from_contents (src, bit_offset, bit_size,
2659 unpacked, TYPE_LENGTH (type),
2660 is_big_endian, has_negatives (type), is_scalar);
2661
2662 return v;
2663 }
2664
2665 /* Move N bits from SOURCE, starting at bit offset SRC_OFFSET to
2666 TARGET, starting at bit offset TARG_OFFSET. SOURCE and TARGET must
2667 not overlap. */
2668 static void
2669 move_bits (gdb_byte *target, int targ_offset, const gdb_byte *source,
2670 int src_offset, int n, int bits_big_endian_p)
2671 {
2672 unsigned int accum, mask;
2673 int accum_bits, chunk_size;
2674
2675 target += targ_offset / HOST_CHAR_BIT;
2676 targ_offset %= HOST_CHAR_BIT;
2677 source += src_offset / HOST_CHAR_BIT;
2678 src_offset %= HOST_CHAR_BIT;
2679 if (bits_big_endian_p)
2680 {
2681 accum = (unsigned char) *source;
2682 source += 1;
2683 accum_bits = HOST_CHAR_BIT - src_offset;
2684
2685 while (n > 0)
2686 {
2687 int unused_right;
2688
2689 accum = (accum << HOST_CHAR_BIT) + (unsigned char) *source;
2690 accum_bits += HOST_CHAR_BIT;
2691 source += 1;
2692 chunk_size = HOST_CHAR_BIT - targ_offset;
2693 if (chunk_size > n)
2694 chunk_size = n;
2695 unused_right = HOST_CHAR_BIT - (chunk_size + targ_offset);
2696 mask = ((1 << chunk_size) - 1) << unused_right;
2697 *target =
2698 (*target & ~mask)
2699 | ((accum >> (accum_bits - chunk_size - unused_right)) & mask);
2700 n -= chunk_size;
2701 accum_bits -= chunk_size;
2702 target += 1;
2703 targ_offset = 0;
2704 }
2705 }
2706 else
2707 {
2708 accum = (unsigned char) *source >> src_offset;
2709 source += 1;
2710 accum_bits = HOST_CHAR_BIT - src_offset;
2711
2712 while (n > 0)
2713 {
2714 accum = accum + ((unsigned char) *source << accum_bits);
2715 accum_bits += HOST_CHAR_BIT;
2716 source += 1;
2717 chunk_size = HOST_CHAR_BIT - targ_offset;
2718 if (chunk_size > n)
2719 chunk_size = n;
2720 mask = ((1 << chunk_size) - 1) << targ_offset;
2721 *target = (*target & ~mask) | ((accum << targ_offset) & mask);
2722 n -= chunk_size;
2723 accum_bits -= chunk_size;
2724 accum >>= chunk_size;
2725 target += 1;
2726 targ_offset = 0;
2727 }
2728 }
2729 }
2730
2731 /* Store the contents of FROMVAL into the location of TOVAL.
2732 Return a new value with the location of TOVAL and contents of
2733 FROMVAL. Handles assignment into packed fields that have
2734 floating-point or non-scalar types. */
2735
2736 static struct value *
2737 ada_value_assign (struct value *toval, struct value *fromval)
2738 {
2739 struct type *type = value_type (toval);
2740 int bits = value_bitsize (toval);
2741
2742 toval = ada_coerce_ref (toval);
2743 fromval = ada_coerce_ref (fromval);
2744
2745 if (ada_is_direct_array_type (value_type (toval)))
2746 toval = ada_coerce_to_simple_array (toval);
2747 if (ada_is_direct_array_type (value_type (fromval)))
2748 fromval = ada_coerce_to_simple_array (fromval);
2749
2750 if (!deprecated_value_modifiable (toval))
2751 error (_("Left operand of assignment is not a modifiable lvalue."));
2752
2753 if (VALUE_LVAL (toval) == lval_memory
2754 && bits > 0
2755 && (TYPE_CODE (type) == TYPE_CODE_FLT
2756 || TYPE_CODE (type) == TYPE_CODE_STRUCT))
2757 {
2758 int len = (value_bitpos (toval)
2759 + bits + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT;
2760 int from_size;
2761 gdb_byte *buffer = (gdb_byte *) alloca (len);
2762 struct value *val;
2763 CORE_ADDR to_addr = value_address (toval);
2764
2765 if (TYPE_CODE (type) == TYPE_CODE_FLT)
2766 fromval = value_cast (type, fromval);
2767
2768 read_memory (to_addr, buffer, len);
2769 from_size = value_bitsize (fromval);
2770 if (from_size == 0)
2771 from_size = TYPE_LENGTH (value_type (fromval)) * TARGET_CHAR_BIT;
2772 if (gdbarch_bits_big_endian (get_type_arch (type)))
2773 move_bits (buffer, value_bitpos (toval),
2774 value_contents (fromval), from_size - bits, bits, 1);
2775 else
2776 move_bits (buffer, value_bitpos (toval),
2777 value_contents (fromval), 0, bits, 0);
2778 write_memory_with_notification (to_addr, buffer, len);
2779
2780 val = value_copy (toval);
2781 memcpy (value_contents_raw (val), value_contents (fromval),
2782 TYPE_LENGTH (type));
2783 deprecated_set_value_type (val, type);
2784
2785 return val;
2786 }
2787
2788 return value_assign (toval, fromval);
2789 }
2790
2791
2792 /* Given that COMPONENT is a memory lvalue that is part of the lvalue
2793 CONTAINER, assign the contents of VAL to COMPONENTS's place in
2794 CONTAINER. Modifies the VALUE_CONTENTS of CONTAINER only, not
2795 COMPONENT, and not the inferior's memory. The current contents
2796 of COMPONENT are ignored.
2797
2798 Although not part of the initial design, this function also works
2799 when CONTAINER and COMPONENT are not_lval's: it works as if CONTAINER
2800 had a null address, and COMPONENT had an address which is equal to
2801 its offset inside CONTAINER. */
2802
2803 static void
2804 value_assign_to_component (struct value *container, struct value *component,
2805 struct value *val)
2806 {
2807 LONGEST offset_in_container =
2808 (LONGEST) (value_address (component) - value_address (container));
2809 int bit_offset_in_container =
2810 value_bitpos (component) - value_bitpos (container);
2811 int bits;
2812
2813 val = value_cast (value_type (component), val);
2814
2815 if (value_bitsize (component) == 0)
2816 bits = TARGET_CHAR_BIT * TYPE_LENGTH (value_type (component));
2817 else
2818 bits = value_bitsize (component);
2819
2820 if (gdbarch_bits_big_endian (get_type_arch (value_type (container))))
2821 move_bits (value_contents_writeable (container) + offset_in_container,
2822 value_bitpos (container) + bit_offset_in_container,
2823 value_contents (val),
2824 TYPE_LENGTH (value_type (component)) * TARGET_CHAR_BIT - bits,
2825 bits, 1);
2826 else
2827 move_bits (value_contents_writeable (container) + offset_in_container,
2828 value_bitpos (container) + bit_offset_in_container,
2829 value_contents (val), 0, bits, 0);
2830 }
2831
2832 /* The value of the element of array ARR at the ARITY indices given in IND.
2833 ARR may be either a simple array, GNAT array descriptor, or pointer
2834 thereto. */
2835
2836 struct value *
2837 ada_value_subscript (struct value *arr, int arity, struct value **ind)
2838 {
2839 int k;
2840 struct value *elt;
2841 struct type *elt_type;
2842
2843 elt = ada_coerce_to_simple_array (arr);
2844
2845 elt_type = ada_check_typedef (value_type (elt));
2846 if (TYPE_CODE (elt_type) == TYPE_CODE_ARRAY
2847 && TYPE_FIELD_BITSIZE (elt_type, 0) > 0)
2848 return value_subscript_packed (elt, arity, ind);
2849
2850 for (k = 0; k < arity; k += 1)
2851 {
2852 if (TYPE_CODE (elt_type) != TYPE_CODE_ARRAY)
2853 error (_("too many subscripts (%d expected)"), k);
2854 elt = value_subscript (elt, pos_atr (ind[k]));
2855 }
2856 return elt;
2857 }
2858
2859 /* Assuming ARR is a pointer to a GDB array, the value of the element
2860 of *ARR at the ARITY indices given in IND.
2861 Does not read the entire array into memory.
2862
2863 Note: Unlike what one would expect, this function is used instead of
2864 ada_value_subscript for basically all non-packed array types. The reason
2865 for this is that a side effect of doing our own pointer arithmetics instead
2866 of relying on value_subscript is that there is no implicit typedef peeling.
2867 This is important for arrays of array accesses, where it allows us to
2868 preserve the fact that the array's element is an array access, where the
2869 access part os encoded in a typedef layer. */
2870
2871 static struct value *
2872 ada_value_ptr_subscript (struct value *arr, int arity, struct value **ind)
2873 {
2874 int k;
2875 struct value *array_ind = ada_value_ind (arr);
2876 struct type *type
2877 = check_typedef (value_enclosing_type (array_ind));
2878
2879 if (TYPE_CODE (type) == TYPE_CODE_ARRAY
2880 && TYPE_FIELD_BITSIZE (type, 0) > 0)
2881 return value_subscript_packed (array_ind, arity, ind);
2882
2883 for (k = 0; k < arity; k += 1)
2884 {
2885 LONGEST lwb, upb;
2886 struct value *lwb_value;
2887
2888 if (TYPE_CODE (type) != TYPE_CODE_ARRAY)
2889 error (_("too many subscripts (%d expected)"), k);
2890 arr = value_cast (lookup_pointer_type (TYPE_TARGET_TYPE (type)),
2891 value_copy (arr));
2892 get_discrete_bounds (TYPE_INDEX_TYPE (type), &lwb, &upb);
2893 lwb_value = value_from_longest (value_type(ind[k]), lwb);
2894 arr = value_ptradd (arr, pos_atr (ind[k]) - pos_atr (lwb_value));
2895 type = TYPE_TARGET_TYPE (type);
2896 }
2897
2898 return value_ind (arr);
2899 }
2900
2901 /* Given that ARRAY_PTR is a pointer or reference to an array of type TYPE (the
2902 actual type of ARRAY_PTR is ignored), returns the Ada slice of
2903 HIGH'Pos-LOW'Pos+1 elements starting at index LOW. The lower bound of
2904 this array is LOW, as per Ada rules. */
2905 static struct value *
2906 ada_value_slice_from_ptr (struct value *array_ptr, struct type *type,
2907 int low, int high)
2908 {
2909 struct type *type0 = ada_check_typedef (type);
2910 struct type *base_index_type = TYPE_TARGET_TYPE (TYPE_INDEX_TYPE (type0));
2911 struct type *index_type
2912 = create_static_range_type (NULL, base_index_type, low, high);
2913 struct type *slice_type = create_array_type_with_stride
2914 (NULL, TYPE_TARGET_TYPE (type0), index_type,
2915 get_dyn_prop (DYN_PROP_BYTE_STRIDE, type0),
2916 TYPE_FIELD_BITSIZE (type0, 0));
2917 int base_low = ada_discrete_type_low_bound (TYPE_INDEX_TYPE (type0));
2918 LONGEST base_low_pos, low_pos;
2919 CORE_ADDR base;
2920
2921 if (!discrete_position (base_index_type, low, &low_pos)
2922 || !discrete_position (base_index_type, base_low, &base_low_pos))
2923 {
2924 warning (_("unable to get positions in slice, use bounds instead"));
2925 low_pos = low;
2926 base_low_pos = base_low;
2927 }
2928
2929 base = value_as_address (array_ptr)
2930 + ((low_pos - base_low_pos)
2931 * TYPE_LENGTH (TYPE_TARGET_TYPE (type0)));
2932 return value_at_lazy (slice_type, base);
2933 }
2934
2935
2936 static struct value *
2937 ada_value_slice (struct value *array, int low, int high)
2938 {
2939 struct type *type = ada_check_typedef (value_type (array));
2940 struct type *base_index_type = TYPE_TARGET_TYPE (TYPE_INDEX_TYPE (type));
2941 struct type *index_type
2942 = create_static_range_type (NULL, TYPE_INDEX_TYPE (type), low, high);
2943 struct type *slice_type = create_array_type_with_stride
2944 (NULL, TYPE_TARGET_TYPE (type), index_type,
2945 get_dyn_prop (DYN_PROP_BYTE_STRIDE, type),
2946 TYPE_FIELD_BITSIZE (type, 0));
2947 LONGEST low_pos, high_pos;
2948
2949 if (!discrete_position (base_index_type, low, &low_pos)
2950 || !discrete_position (base_index_type, high, &high_pos))
2951 {
2952 warning (_("unable to get positions in slice, use bounds instead"));
2953 low_pos = low;
2954 high_pos = high;
2955 }
2956
2957 return value_cast (slice_type,
2958 value_slice (array, low, high_pos - low_pos + 1));
2959 }
2960
2961 /* If type is a record type in the form of a standard GNAT array
2962 descriptor, returns the number of dimensions for type. If arr is a
2963 simple array, returns the number of "array of"s that prefix its
2964 type designation. Otherwise, returns 0. */
2965
2966 int
2967 ada_array_arity (struct type *type)
2968 {
2969 int arity;
2970
2971 if (type == NULL)
2972 return 0;
2973
2974 type = desc_base_type (type);
2975
2976 arity = 0;
2977 if (TYPE_CODE (type) == TYPE_CODE_STRUCT)
2978 return desc_arity (desc_bounds_type (type));
2979 else
2980 while (TYPE_CODE (type) == TYPE_CODE_ARRAY)
2981 {
2982 arity += 1;
2983 type = ada_check_typedef (TYPE_TARGET_TYPE (type));
2984 }
2985
2986 return arity;
2987 }
2988
2989 /* If TYPE is a record type in the form of a standard GNAT array
2990 descriptor or a simple array type, returns the element type for
2991 TYPE after indexing by NINDICES indices, or by all indices if
2992 NINDICES is -1. Otherwise, returns NULL. */
2993
2994 struct type *
2995 ada_array_element_type (struct type *type, int nindices)
2996 {
2997 type = desc_base_type (type);
2998
2999 if (TYPE_CODE (type) == TYPE_CODE_STRUCT)
3000 {
3001 int k;
3002 struct type *p_array_type;
3003
3004 p_array_type = desc_data_target_type (type);
3005
3006 k = ada_array_arity (type);
3007 if (k == 0)
3008 return NULL;
3009
3010 /* Initially p_array_type = elt_type(*)[]...(k times)...[]. */
3011 if (nindices >= 0 && k > nindices)
3012 k = nindices;
3013 while (k > 0 && p_array_type != NULL)
3014 {
3015 p_array_type = ada_check_typedef (TYPE_TARGET_TYPE (p_array_type));
3016 k -= 1;
3017 }
3018 return p_array_type;
3019 }
3020 else if (TYPE_CODE (type) == TYPE_CODE_ARRAY)
3021 {
3022 while (nindices != 0 && TYPE_CODE (type) == TYPE_CODE_ARRAY)
3023 {
3024 type = TYPE_TARGET_TYPE (type);
3025 nindices -= 1;
3026 }
3027 return type;
3028 }
3029
3030 return NULL;
3031 }
3032
3033 /* The type of nth index in arrays of given type (n numbering from 1).
3034 Does not examine memory. Throws an error if N is invalid or TYPE
3035 is not an array type. NAME is the name of the Ada attribute being
3036 evaluated ('range, 'first, 'last, or 'length); it is used in building
3037 the error message. */
3038
3039 static struct type *
3040 ada_index_type (struct type *type, int n, const char *name)
3041 {
3042 struct type *result_type;
3043
3044 type = desc_base_type (type);
3045
3046 if (n < 0 || n > ada_array_arity (type))
3047 error (_("invalid dimension number to '%s"), name);
3048
3049 if (ada_is_simple_array_type (type))
3050 {
3051 int i;
3052
3053 for (i = 1; i < n; i += 1)
3054 type = TYPE_TARGET_TYPE (type);
3055 result_type = TYPE_TARGET_TYPE (TYPE_INDEX_TYPE (type));
3056 /* FIXME: The stabs type r(0,0);bound;bound in an array type
3057 has a target type of TYPE_CODE_UNDEF. We compensate here, but
3058 perhaps stabsread.c would make more sense. */
3059 if (result_type && TYPE_CODE (result_type) == TYPE_CODE_UNDEF)
3060 result_type = NULL;
3061 }
3062 else
3063 {
3064 result_type = desc_index_type (desc_bounds_type (type), n);
3065 if (result_type == NULL)
3066 error (_("attempt to take bound of something that is not an array"));
3067 }
3068
3069 return result_type;
3070 }
3071
3072 /* Given that arr is an array type, returns the lower bound of the
3073 Nth index (numbering from 1) if WHICH is 0, and the upper bound if
3074 WHICH is 1. This returns bounds 0 .. -1 if ARR_TYPE is an
3075 array-descriptor type. It works for other arrays with bounds supplied
3076 by run-time quantities other than discriminants. */
3077
3078 static LONGEST
3079 ada_array_bound_from_type (struct type *arr_type, int n, int which)
3080 {
3081 struct type *type, *index_type_desc, *index_type;
3082 int i;
3083
3084 gdb_assert (which == 0 || which == 1);
3085
3086 if (ada_is_constrained_packed_array_type (arr_type))
3087 arr_type = decode_constrained_packed_array_type (arr_type);
3088
3089 if (arr_type == NULL || !ada_is_simple_array_type (arr_type))
3090 return (LONGEST) - which;
3091
3092 if (TYPE_CODE (arr_type) == TYPE_CODE_PTR)
3093 type = TYPE_TARGET_TYPE (arr_type);
3094 else
3095 type = arr_type;
3096
3097 if (TYPE_FIXED_INSTANCE (type))
3098 {
3099 /* The array has already been fixed, so we do not need to
3100 check the parallel ___XA type again. That encoding has
3101 already been applied, so ignore it now. */
3102 index_type_desc = NULL;
3103 }
3104 else
3105 {
3106 index_type_desc = ada_find_parallel_type (type, "___XA");
3107 ada_fixup_array_indexes_type (index_type_desc);
3108 }
3109
3110 if (index_type_desc != NULL)
3111 index_type = to_fixed_range_type (TYPE_FIELD_TYPE (index_type_desc, n - 1),
3112 NULL);
3113 else
3114 {
3115 struct type *elt_type = check_typedef (type);
3116
3117 for (i = 1; i < n; i++)
3118 elt_type = check_typedef (TYPE_TARGET_TYPE (elt_type));
3119
3120 index_type = TYPE_INDEX_TYPE (elt_type);
3121 }
3122
3123 return
3124 (LONGEST) (which == 0
3125 ? ada_discrete_type_low_bound (index_type)
3126 : ada_discrete_type_high_bound (index_type));
3127 }
3128
3129 /* Given that arr is an array value, returns the lower bound of the
3130 nth index (numbering from 1) if WHICH is 0, and the upper bound if
3131 WHICH is 1. This routine will also work for arrays with bounds
3132 supplied by run-time quantities other than discriminants. */
3133
3134 static LONGEST
3135 ada_array_bound (struct value *arr, int n, int which)
3136 {
3137 struct type *arr_type;
3138
3139 if (TYPE_CODE (check_typedef (value_type (arr))) == TYPE_CODE_PTR)
3140 arr = value_ind (arr);
3141 arr_type = value_enclosing_type (arr);
3142
3143 if (ada_is_constrained_packed_array_type (arr_type))
3144 return ada_array_bound (decode_constrained_packed_array (arr), n, which);
3145 else if (ada_is_simple_array_type (arr_type))
3146 return ada_array_bound_from_type (arr_type, n, which);
3147 else
3148 return value_as_long (desc_one_bound (desc_bounds (arr), n, which));
3149 }
3150
3151 /* Given that arr is an array value, returns the length of the
3152 nth index. This routine will also work for arrays with bounds
3153 supplied by run-time quantities other than discriminants.
3154 Does not work for arrays indexed by enumeration types with representation
3155 clauses at the moment. */
3156
3157 static LONGEST
3158 ada_array_length (struct value *arr, int n)
3159 {
3160 struct type *arr_type, *index_type;
3161 int low, high;
3162
3163 if (TYPE_CODE (check_typedef (value_type (arr))) == TYPE_CODE_PTR)
3164 arr = value_ind (arr);
3165 arr_type = value_enclosing_type (arr);
3166
3167 if (ada_is_constrained_packed_array_type (arr_type))
3168 return ada_array_length (decode_constrained_packed_array (arr), n);
3169
3170 if (ada_is_simple_array_type (arr_type))
3171 {
3172 low = ada_array_bound_from_type (arr_type, n, 0);
3173 high = ada_array_bound_from_type (arr_type, n, 1);
3174 }
3175 else
3176 {
3177 low = value_as_long (desc_one_bound (desc_bounds (arr), n, 0));
3178 high = value_as_long (desc_one_bound (desc_bounds (arr), n, 1));
3179 }
3180
3181 arr_type = check_typedef (arr_type);
3182 index_type = ada_index_type (arr_type, n, "length");
3183 if (index_type != NULL)
3184 {
3185 struct type *base_type;
3186 if (TYPE_CODE (index_type) == TYPE_CODE_RANGE)
3187 base_type = TYPE_TARGET_TYPE (index_type);
3188 else
3189 base_type = index_type;
3190
3191 low = pos_atr (value_from_longest (base_type, low));
3192 high = pos_atr (value_from_longest (base_type, high));
3193 }
3194 return high - low + 1;
3195 }
3196
3197 /* An empty array whose type is that of ARR_TYPE (an array type),
3198 with bounds LOW to LOW-1. */
3199
3200 static struct value *
3201 empty_array (struct type *arr_type, int low)
3202 {
3203 struct type *arr_type0 = ada_check_typedef (arr_type);
3204 struct type *index_type
3205 = create_static_range_type
3206 (NULL, TYPE_TARGET_TYPE (TYPE_INDEX_TYPE (arr_type0)), low, low - 1);
3207 struct type *elt_type = ada_array_element_type (arr_type0, 1);
3208
3209 return allocate_value (create_array_type (NULL, elt_type, index_type));
3210 }
3211 \f
3212
3213 /* Name resolution */
3214
3215 /* The "decoded" name for the user-definable Ada operator corresponding
3216 to OP. */
3217
3218 static const char *
3219 ada_decoded_op_name (enum exp_opcode op)
3220 {
3221 int i;
3222
3223 for (i = 0; ada_opname_table[i].encoded != NULL; i += 1)
3224 {
3225 if (ada_opname_table[i].op == op)
3226 return ada_opname_table[i].decoded;
3227 }
3228 error (_("Could not find operator name for opcode"));
3229 }
3230
3231
3232 /* Same as evaluate_type (*EXP), but resolves ambiguous symbol
3233 references (marked by OP_VAR_VALUE nodes in which the symbol has an
3234 undefined namespace) and converts operators that are
3235 user-defined into appropriate function calls. If CONTEXT_TYPE is
3236 non-null, it provides a preferred result type [at the moment, only
3237 type void has any effect---causing procedures to be preferred over
3238 functions in calls]. A null CONTEXT_TYPE indicates that a non-void
3239 return type is preferred. May change (expand) *EXP. */
3240
3241 static void
3242 resolve (expression_up *expp, int void_context_p)
3243 {
3244 struct type *context_type = NULL;
3245 int pc = 0;
3246
3247 if (void_context_p)
3248 context_type = builtin_type ((*expp)->gdbarch)->builtin_void;
3249
3250 resolve_subexp (expp, &pc, 1, context_type);
3251 }
3252
3253 /* Resolve the operator of the subexpression beginning at
3254 position *POS of *EXPP. "Resolving" consists of replacing
3255 the symbols that have undefined namespaces in OP_VAR_VALUE nodes
3256 with their resolutions, replacing built-in operators with
3257 function calls to user-defined operators, where appropriate, and,
3258 when DEPROCEDURE_P is non-zero, converting function-valued variables
3259 into parameterless calls. May expand *EXPP. The CONTEXT_TYPE functions
3260 are as in ada_resolve, above. */
3261
3262 static struct value *
3263 resolve_subexp (expression_up *expp, int *pos, int deprocedure_p,
3264 struct type *context_type)
3265 {
3266 int pc = *pos;
3267 int i;
3268 struct expression *exp; /* Convenience: == *expp. */
3269 enum exp_opcode op = (*expp)->elts[pc].opcode;
3270 struct value **argvec; /* Vector of operand types (alloca'ed). */
3271 int nargs; /* Number of operands. */
3272 int oplen;
3273 struct cleanup *old_chain = make_cleanup (null_cleanup, NULL);
3274
3275 argvec = NULL;
3276 nargs = 0;
3277 exp = expp->get ();
3278
3279 /* Pass one: resolve operands, saving their types and updating *pos,
3280 if needed. */
3281 switch (op)
3282 {
3283 case OP_FUNCALL:
3284 if (exp->elts[pc + 3].opcode == OP_VAR_VALUE
3285 && SYMBOL_DOMAIN (exp->elts[pc + 5].symbol) == UNDEF_DOMAIN)
3286 *pos += 7;
3287 else
3288 {
3289 *pos += 3;
3290 resolve_subexp (expp, pos, 0, NULL);
3291 }
3292 nargs = longest_to_int (exp->elts[pc + 1].longconst);
3293 break;
3294
3295 case UNOP_ADDR:
3296 *pos += 1;
3297 resolve_subexp (expp, pos, 0, NULL);
3298 break;
3299
3300 case UNOP_QUAL:
3301 *pos += 3;
3302 resolve_subexp (expp, pos, 1, check_typedef (exp->elts[pc + 1].type));
3303 break;
3304
3305 case OP_ATR_MODULUS:
3306 case OP_ATR_SIZE:
3307 case OP_ATR_TAG:
3308 case OP_ATR_FIRST:
3309 case OP_ATR_LAST:
3310 case OP_ATR_LENGTH:
3311 case OP_ATR_POS:
3312 case OP_ATR_VAL:
3313 case OP_ATR_MIN:
3314 case OP_ATR_MAX:
3315 case TERNOP_IN_RANGE:
3316 case BINOP_IN_BOUNDS:
3317 case UNOP_IN_RANGE:
3318 case OP_AGGREGATE:
3319 case OP_OTHERS:
3320 case OP_CHOICES:
3321 case OP_POSITIONAL:
3322 case OP_DISCRETE_RANGE:
3323 case OP_NAME:
3324 ada_forward_operator_length (exp, pc, &oplen, &nargs);
3325 *pos += oplen;
3326 break;
3327
3328 case BINOP_ASSIGN:
3329 {
3330 struct value *arg1;
3331
3332 *pos += 1;
3333 arg1 = resolve_subexp (expp, pos, 0, NULL);
3334 if (arg1 == NULL)
3335 resolve_subexp (expp, pos, 1, NULL);
3336 else
3337 resolve_subexp (expp, pos, 1, value_type (arg1));
3338 break;
3339 }
3340
3341 case UNOP_CAST:
3342 *pos += 3;
3343 nargs = 1;
3344 break;
3345
3346 case BINOP_ADD:
3347 case BINOP_SUB:
3348 case BINOP_MUL:
3349 case BINOP_DIV:
3350 case BINOP_REM:
3351 case BINOP_MOD:
3352 case BINOP_EXP:
3353 case BINOP_CONCAT:
3354 case BINOP_LOGICAL_AND:
3355 case BINOP_LOGICAL_OR:
3356 case BINOP_BITWISE_AND:
3357 case BINOP_BITWISE_IOR:
3358 case BINOP_BITWISE_XOR:
3359
3360 case BINOP_EQUAL:
3361 case BINOP_NOTEQUAL:
3362 case BINOP_LESS:
3363 case BINOP_GTR:
3364 case BINOP_LEQ:
3365 case BINOP_GEQ:
3366
3367 case BINOP_REPEAT:
3368 case BINOP_SUBSCRIPT:
3369 case BINOP_COMMA:
3370 *pos += 1;
3371 nargs = 2;
3372 break;
3373
3374 case UNOP_NEG:
3375 case UNOP_PLUS:
3376 case UNOP_LOGICAL_NOT:
3377 case UNOP_ABS:
3378 case UNOP_IND:
3379 *pos += 1;
3380 nargs = 1;
3381 break;
3382
3383 case OP_LONG:
3384 case OP_FLOAT:
3385 case OP_VAR_VALUE:
3386 case OP_VAR_MSYM_VALUE:
3387 *pos += 4;
3388 break;
3389
3390 case OP_TYPE:
3391 case OP_BOOL:
3392 case OP_LAST:
3393 case OP_INTERNALVAR:
3394 *pos += 3;
3395 break;
3396
3397 case UNOP_MEMVAL:
3398 *pos += 3;
3399 nargs = 1;
3400 break;
3401
3402 case OP_REGISTER:
3403 *pos += 4 + BYTES_TO_EXP_ELEM (exp->elts[pc + 1].longconst + 1);
3404 break;
3405
3406 case STRUCTOP_STRUCT:
3407 *pos += 4 + BYTES_TO_EXP_ELEM (exp->elts[pc + 1].longconst + 1);
3408 nargs = 1;
3409 break;
3410
3411 case TERNOP_SLICE:
3412 *pos += 1;
3413 nargs = 3;
3414 break;
3415
3416 case OP_STRING:
3417 break;
3418
3419 default:
3420 error (_("Unexpected operator during name resolution"));
3421 }
3422
3423 argvec = XALLOCAVEC (struct value *, nargs + 1);
3424 for (i = 0; i < nargs; i += 1)
3425 argvec[i] = resolve_subexp (expp, pos, 1, NULL);
3426 argvec[i] = NULL;
3427 exp = expp->get ();
3428
3429 /* Pass two: perform any resolution on principal operator. */
3430 switch (op)
3431 {
3432 default:
3433 break;
3434
3435 case OP_VAR_VALUE:
3436 if (SYMBOL_DOMAIN (exp->elts[pc + 2].symbol) == UNDEF_DOMAIN)
3437 {
3438 struct block_symbol *candidates;
3439 int n_candidates;
3440
3441 n_candidates =
3442 ada_lookup_symbol_list (SYMBOL_LINKAGE_NAME
3443 (exp->elts[pc + 2].symbol),
3444 exp->elts[pc + 1].block, VAR_DOMAIN,
3445 &candidates);
3446 make_cleanup (xfree, candidates);
3447
3448 if (n_candidates > 1)
3449 {
3450 /* Types tend to get re-introduced locally, so if there
3451 are any local symbols that are not types, first filter
3452 out all types. */
3453 int j;
3454 for (j = 0; j < n_candidates; j += 1)
3455 switch (SYMBOL_CLASS (candidates[j].symbol))
3456 {
3457 case LOC_REGISTER:
3458 case LOC_ARG:
3459 case LOC_REF_ARG:
3460 case LOC_REGPARM_ADDR:
3461 case LOC_LOCAL:
3462 case LOC_COMPUTED:
3463 goto FoundNonType;
3464 default:
3465 break;
3466 }
3467 FoundNonType:
3468 if (j < n_candidates)
3469 {
3470 j = 0;
3471 while (j < n_candidates)
3472 {
3473 if (SYMBOL_CLASS (candidates[j].symbol) == LOC_TYPEDEF)
3474 {
3475 candidates[j] = candidates[n_candidates - 1];
3476 n_candidates -= 1;
3477 }
3478 else
3479 j += 1;
3480 }
3481 }
3482 }
3483
3484 if (n_candidates == 0)
3485 error (_("No definition found for %s"),
3486 SYMBOL_PRINT_NAME (exp->elts[pc + 2].symbol));
3487 else if (n_candidates == 1)
3488 i = 0;
3489 else if (deprocedure_p
3490 && !is_nonfunction (candidates, n_candidates))
3491 {
3492 i = ada_resolve_function
3493 (candidates, n_candidates, NULL, 0,
3494 SYMBOL_LINKAGE_NAME (exp->elts[pc + 2].symbol),
3495 context_type);
3496 if (i < 0)
3497 error (_("Could not find a match for %s"),
3498 SYMBOL_PRINT_NAME (exp->elts[pc + 2].symbol));
3499 }
3500 else
3501 {
3502 printf_filtered (_("Multiple matches for %s\n"),
3503 SYMBOL_PRINT_NAME (exp->elts[pc + 2].symbol));
3504 user_select_syms (candidates, n_candidates, 1);
3505 i = 0;
3506 }
3507
3508 exp->elts[pc + 1].block = candidates[i].block;
3509 exp->elts[pc + 2].symbol = candidates[i].symbol;
3510 if (innermost_block == NULL
3511 || contained_in (candidates[i].block, innermost_block))
3512 innermost_block = candidates[i].block;
3513 }
3514
3515 if (deprocedure_p
3516 && (TYPE_CODE (SYMBOL_TYPE (exp->elts[pc + 2].symbol))
3517 == TYPE_CODE_FUNC))
3518 {
3519 replace_operator_with_call (expp, pc, 0, 0,
3520 exp->elts[pc + 2].symbol,
3521 exp->elts[pc + 1].block);
3522 exp = expp->get ();
3523 }
3524 break;
3525
3526 case OP_FUNCALL:
3527 {
3528 if (exp->elts[pc + 3].opcode == OP_VAR_VALUE
3529 && SYMBOL_DOMAIN (exp->elts[pc + 5].symbol) == UNDEF_DOMAIN)
3530 {
3531 struct block_symbol *candidates;
3532 int n_candidates;
3533
3534 n_candidates =
3535 ada_lookup_symbol_list (SYMBOL_LINKAGE_NAME
3536 (exp->elts[pc + 5].symbol),
3537 exp->elts[pc + 4].block, VAR_DOMAIN,
3538 &candidates);
3539 make_cleanup (xfree, candidates);
3540
3541 if (n_candidates == 1)
3542 i = 0;
3543 else
3544 {
3545 i = ada_resolve_function
3546 (candidates, n_candidates,
3547 argvec, nargs,
3548 SYMBOL_LINKAGE_NAME (exp->elts[pc + 5].symbol),
3549 context_type);
3550 if (i < 0)
3551 error (_("Could not find a match for %s"),
3552 SYMBOL_PRINT_NAME (exp->elts[pc + 5].symbol));
3553 }
3554
3555 exp->elts[pc + 4].block = candidates[i].block;
3556 exp->elts[pc + 5].symbol = candidates[i].symbol;
3557 if (innermost_block == NULL
3558 || contained_in (candidates[i].block, innermost_block))
3559 innermost_block = candidates[i].block;
3560 }
3561 }
3562 break;
3563 case BINOP_ADD:
3564 case BINOP_SUB:
3565 case BINOP_MUL:
3566 case BINOP_DIV:
3567 case BINOP_REM:
3568 case BINOP_MOD:
3569 case BINOP_CONCAT:
3570 case BINOP_BITWISE_AND:
3571 case BINOP_BITWISE_IOR:
3572 case BINOP_BITWISE_XOR:
3573 case BINOP_EQUAL:
3574 case BINOP_NOTEQUAL:
3575 case BINOP_LESS:
3576 case BINOP_GTR:
3577 case BINOP_LEQ:
3578 case BINOP_GEQ:
3579 case BINOP_EXP:
3580 case UNOP_NEG:
3581 case UNOP_PLUS:
3582 case UNOP_LOGICAL_NOT:
3583 case UNOP_ABS:
3584 if (possible_user_operator_p (op, argvec))
3585 {
3586 struct block_symbol *candidates;
3587 int n_candidates;
3588
3589 n_candidates =
3590 ada_lookup_symbol_list (ada_decoded_op_name (op),
3591 (struct block *) NULL, VAR_DOMAIN,
3592 &candidates);
3593 make_cleanup (xfree, candidates);
3594
3595 i = ada_resolve_function (candidates, n_candidates, argvec, nargs,
3596 ada_decoded_op_name (op), NULL);
3597 if (i < 0)
3598 break;
3599
3600 replace_operator_with_call (expp, pc, nargs, 1,
3601 candidates[i].symbol,
3602 candidates[i].block);
3603 exp = expp->get ();
3604 }
3605 break;
3606
3607 case OP_TYPE:
3608 case OP_REGISTER:
3609 do_cleanups (old_chain);
3610 return NULL;
3611 }
3612
3613 *pos = pc;
3614 do_cleanups (old_chain);
3615 if (exp->elts[pc].opcode == OP_VAR_MSYM_VALUE)
3616 return evaluate_var_msym_value (EVAL_AVOID_SIDE_EFFECTS,
3617 exp->elts[pc + 1].objfile,
3618 exp->elts[pc + 2].msymbol);
3619 else
3620 return evaluate_subexp_type (exp, pos);
3621 }
3622
3623 /* Return non-zero if formal type FTYPE matches actual type ATYPE. If
3624 MAY_DEREF is non-zero, the formal may be a pointer and the actual
3625 a non-pointer. */
3626 /* The term "match" here is rather loose. The match is heuristic and
3627 liberal. */
3628
3629 static int
3630 ada_type_match (struct type *ftype, struct type *atype, int may_deref)
3631 {
3632 ftype = ada_check_typedef (ftype);
3633 atype = ada_check_typedef (atype);
3634
3635 if (TYPE_CODE (ftype) == TYPE_CODE_REF)
3636 ftype = TYPE_TARGET_TYPE (ftype);
3637 if (TYPE_CODE (atype) == TYPE_CODE_REF)
3638 atype = TYPE_TARGET_TYPE (atype);
3639
3640 switch (TYPE_CODE (ftype))
3641 {
3642 default:
3643 return TYPE_CODE (ftype) == TYPE_CODE (atype);
3644 case TYPE_CODE_PTR:
3645 if (TYPE_CODE (atype) == TYPE_CODE_PTR)
3646 return ada_type_match (TYPE_TARGET_TYPE (ftype),
3647 TYPE_TARGET_TYPE (atype), 0);
3648 else
3649 return (may_deref
3650 && ada_type_match (TYPE_TARGET_TYPE (ftype), atype, 0));
3651 case TYPE_CODE_INT:
3652 case TYPE_CODE_ENUM:
3653 case TYPE_CODE_RANGE:
3654 switch (TYPE_CODE (atype))
3655 {
3656 case TYPE_CODE_INT:
3657 case TYPE_CODE_ENUM:
3658 case TYPE_CODE_RANGE:
3659 return 1;
3660 default:
3661 return 0;
3662 }
3663
3664 case TYPE_CODE_ARRAY:
3665 return (TYPE_CODE (atype) == TYPE_CODE_ARRAY
3666 || ada_is_array_descriptor_type (atype));
3667
3668 case TYPE_CODE_STRUCT:
3669 if (ada_is_array_descriptor_type (ftype))
3670 return (TYPE_CODE (atype) == TYPE_CODE_ARRAY
3671 || ada_is_array_descriptor_type (atype));
3672 else
3673 return (TYPE_CODE (atype) == TYPE_CODE_STRUCT
3674 && !ada_is_array_descriptor_type (atype));
3675
3676 case TYPE_CODE_UNION:
3677 case TYPE_CODE_FLT:
3678 return (TYPE_CODE (atype) == TYPE_CODE (ftype));
3679 }
3680 }
3681
3682 /* Return non-zero if the formals of FUNC "sufficiently match" the
3683 vector of actual argument types ACTUALS of size N_ACTUALS. FUNC
3684 may also be an enumeral, in which case it is treated as a 0-
3685 argument function. */
3686
3687 static int
3688 ada_args_match (struct symbol *func, struct value **actuals, int n_actuals)
3689 {
3690 int i;
3691 struct type *func_type = SYMBOL_TYPE (func);
3692
3693 if (SYMBOL_CLASS (func) == LOC_CONST
3694 && TYPE_CODE (func_type) == TYPE_CODE_ENUM)
3695 return (n_actuals == 0);
3696 else if (func_type == NULL || TYPE_CODE (func_type) != TYPE_CODE_FUNC)
3697 return 0;
3698
3699 if (TYPE_NFIELDS (func_type) != n_actuals)
3700 return 0;
3701
3702 for (i = 0; i < n_actuals; i += 1)
3703 {
3704 if (actuals[i] == NULL)
3705 return 0;
3706 else
3707 {
3708 struct type *ftype = ada_check_typedef (TYPE_FIELD_TYPE (func_type,
3709 i));
3710 struct type *atype = ada_check_typedef (value_type (actuals[i]));
3711
3712 if (!ada_type_match (ftype, atype, 1))
3713 return 0;
3714 }
3715 }
3716 return 1;
3717 }
3718
3719 /* False iff function type FUNC_TYPE definitely does not produce a value
3720 compatible with type CONTEXT_TYPE. Conservatively returns 1 if
3721 FUNC_TYPE is not a valid function type with a non-null return type
3722 or an enumerated type. A null CONTEXT_TYPE indicates any non-void type. */
3723
3724 static int
3725 return_match (struct type *func_type, struct type *context_type)
3726 {
3727 struct type *return_type;
3728
3729 if (func_type == NULL)
3730 return 1;
3731
3732 if (TYPE_CODE (func_type) == TYPE_CODE_FUNC)
3733 return_type = get_base_type (TYPE_TARGET_TYPE (func_type));
3734 else
3735 return_type = get_base_type (func_type);
3736 if (return_type == NULL)
3737 return 1;
3738
3739 context_type = get_base_type (context_type);
3740
3741 if (TYPE_CODE (return_type) == TYPE_CODE_ENUM)
3742 return context_type == NULL || return_type == context_type;
3743 else if (context_type == NULL)
3744 return TYPE_CODE (return_type) != TYPE_CODE_VOID;
3745 else
3746 return TYPE_CODE (return_type) == TYPE_CODE (context_type);
3747 }
3748
3749
3750 /* Returns the index in SYMS[0..NSYMS-1] that contains the symbol for the
3751 function (if any) that matches the types of the NARGS arguments in
3752 ARGS. If CONTEXT_TYPE is non-null and there is at least one match
3753 that returns that type, then eliminate matches that don't. If
3754 CONTEXT_TYPE is void and there is at least one match that does not
3755 return void, eliminate all matches that do.
3756
3757 Asks the user if there is more than one match remaining. Returns -1
3758 if there is no such symbol or none is selected. NAME is used
3759 solely for messages. May re-arrange and modify SYMS in
3760 the process; the index returned is for the modified vector. */
3761
3762 static int
3763 ada_resolve_function (struct block_symbol syms[],
3764 int nsyms, struct value **args, int nargs,
3765 const char *name, struct type *context_type)
3766 {
3767 int fallback;
3768 int k;
3769 int m; /* Number of hits */
3770
3771 m = 0;
3772 /* In the first pass of the loop, we only accept functions matching
3773 context_type. If none are found, we add a second pass of the loop
3774 where every function is accepted. */
3775 for (fallback = 0; m == 0 && fallback < 2; fallback++)
3776 {
3777 for (k = 0; k < nsyms; k += 1)
3778 {
3779 struct type *type = ada_check_typedef (SYMBOL_TYPE (syms[k].symbol));
3780
3781 if (ada_args_match (syms[k].symbol, args, nargs)
3782 && (fallback || return_match (type, context_type)))
3783 {
3784 syms[m] = syms[k];
3785 m += 1;
3786 }
3787 }
3788 }
3789
3790 /* If we got multiple matches, ask the user which one to use. Don't do this
3791 interactive thing during completion, though, as the purpose of the
3792 completion is providing a list of all possible matches. Prompting the
3793 user to filter it down would be completely unexpected in this case. */
3794 if (m == 0)
3795 return -1;
3796 else if (m > 1 && !parse_completion)
3797 {
3798 printf_filtered (_("Multiple matches for %s\n"), name);
3799 user_select_syms (syms, m, 1);
3800 return 0;
3801 }
3802 return 0;
3803 }
3804
3805 /* Returns true (non-zero) iff decoded name N0 should appear before N1
3806 in a listing of choices during disambiguation (see sort_choices, below).
3807 The idea is that overloadings of a subprogram name from the
3808 same package should sort in their source order. We settle for ordering
3809 such symbols by their trailing number (__N or $N). */
3810
3811 static int
3812 encoded_ordered_before (const char *N0, const char *N1)
3813 {
3814 if (N1 == NULL)
3815 return 0;
3816 else if (N0 == NULL)
3817 return 1;
3818 else
3819 {
3820 int k0, k1;
3821
3822 for (k0 = strlen (N0) - 1; k0 > 0 && isdigit (N0[k0]); k0 -= 1)
3823 ;
3824 for (k1 = strlen (N1) - 1; k1 > 0 && isdigit (N1[k1]); k1 -= 1)
3825 ;
3826 if ((N0[k0] == '_' || N0[k0] == '$') && N0[k0 + 1] != '\000'
3827 && (N1[k1] == '_' || N1[k1] == '$') && N1[k1 + 1] != '\000')
3828 {
3829 int n0, n1;
3830
3831 n0 = k0;
3832 while (N0[n0] == '_' && n0 > 0 && N0[n0 - 1] == '_')
3833 n0 -= 1;
3834 n1 = k1;
3835 while (N1[n1] == '_' && n1 > 0 && N1[n1 - 1] == '_')
3836 n1 -= 1;
3837 if (n0 == n1 && strncmp (N0, N1, n0) == 0)
3838 return (atoi (N0 + k0 + 1) < atoi (N1 + k1 + 1));
3839 }
3840 return (strcmp (N0, N1) < 0);
3841 }
3842 }
3843
3844 /* Sort SYMS[0..NSYMS-1] to put the choices in a canonical order by the
3845 encoded names. */
3846
3847 static void
3848 sort_choices (struct block_symbol syms[], int nsyms)
3849 {
3850 int i;
3851
3852 for (i = 1; i < nsyms; i += 1)
3853 {
3854 struct block_symbol sym = syms[i];
3855 int j;
3856
3857 for (j = i - 1; j >= 0; j -= 1)
3858 {
3859 if (encoded_ordered_before (SYMBOL_LINKAGE_NAME (syms[j].symbol),
3860 SYMBOL_LINKAGE_NAME (sym.symbol)))
3861 break;
3862 syms[j + 1] = syms[j];
3863 }
3864 syms[j + 1] = sym;
3865 }
3866 }
3867
3868 /* Whether GDB should display formals and return types for functions in the
3869 overloads selection menu. */
3870 static int print_signatures = 1;
3871
3872 /* Print the signature for SYM on STREAM according to the FLAGS options. For
3873 all but functions, the signature is just the name of the symbol. For
3874 functions, this is the name of the function, the list of types for formals
3875 and the return type (if any). */
3876
3877 static void
3878 ada_print_symbol_signature (struct ui_file *stream, struct symbol *sym,
3879 const struct type_print_options *flags)
3880 {
3881 struct type *type = SYMBOL_TYPE (sym);
3882
3883 fprintf_filtered (stream, "%s", SYMBOL_PRINT_NAME (sym));
3884 if (!print_signatures
3885 || type == NULL
3886 || TYPE_CODE (type) != TYPE_CODE_FUNC)
3887 return;
3888
3889 if (TYPE_NFIELDS (type) > 0)
3890 {
3891 int i;
3892
3893 fprintf_filtered (stream, " (");
3894 for (i = 0; i < TYPE_NFIELDS (type); ++i)
3895 {
3896 if (i > 0)
3897 fprintf_filtered (stream, "; ");
3898 ada_print_type (TYPE_FIELD_TYPE (type, i), NULL, stream, -1, 0,
3899 flags);
3900 }
3901 fprintf_filtered (stream, ")");
3902 }
3903 if (TYPE_TARGET_TYPE (type) != NULL
3904 && TYPE_CODE (TYPE_TARGET_TYPE (type)) != TYPE_CODE_VOID)
3905 {
3906 fprintf_filtered (stream, " return ");
3907 ada_print_type (TYPE_TARGET_TYPE (type), NULL, stream, -1, 0, flags);
3908 }
3909 }
3910
3911 /* Given a list of NSYMS symbols in SYMS, select up to MAX_RESULTS>0
3912 by asking the user (if necessary), returning the number selected,
3913 and setting the first elements of SYMS items. Error if no symbols
3914 selected. */
3915
3916 /* NOTE: Adapted from decode_line_2 in symtab.c, with which it ought
3917 to be re-integrated one of these days. */
3918
3919 int
3920 user_select_syms (struct block_symbol *syms, int nsyms, int max_results)
3921 {
3922 int i;
3923 int *chosen = XALLOCAVEC (int , nsyms);
3924 int n_chosen;
3925 int first_choice = (max_results == 1) ? 1 : 2;
3926 const char *select_mode = multiple_symbols_select_mode ();
3927
3928 if (max_results < 1)
3929 error (_("Request to select 0 symbols!"));
3930 if (nsyms <= 1)
3931 return nsyms;
3932
3933 if (select_mode == multiple_symbols_cancel)
3934 error (_("\
3935 canceled because the command is ambiguous\n\
3936 See set/show multiple-symbol."));
3937
3938 /* If select_mode is "all", then return all possible symbols.
3939 Only do that if more than one symbol can be selected, of course.
3940 Otherwise, display the menu as usual. */
3941 if (select_mode == multiple_symbols_all && max_results > 1)
3942 return nsyms;
3943
3944 printf_unfiltered (_("[0] cancel\n"));
3945 if (max_results > 1)
3946 printf_unfiltered (_("[1] all\n"));
3947
3948 sort_choices (syms, nsyms);
3949
3950 for (i = 0; i < nsyms; i += 1)
3951 {
3952 if (syms[i].symbol == NULL)
3953 continue;
3954
3955 if (SYMBOL_CLASS (syms[i].symbol) == LOC_BLOCK)
3956 {
3957 struct symtab_and_line sal =
3958 find_function_start_sal (syms[i].symbol, 1);
3959
3960 printf_unfiltered ("[%d] ", i + first_choice);
3961 ada_print_symbol_signature (gdb_stdout, syms[i].symbol,
3962 &type_print_raw_options);
3963 if (sal.symtab == NULL)
3964 printf_unfiltered (_(" at <no source file available>:%d\n"),
3965 sal.line);
3966 else
3967 printf_unfiltered (_(" at %s:%d\n"),
3968 symtab_to_filename_for_display (sal.symtab),
3969 sal.line);
3970 continue;
3971 }
3972 else
3973 {
3974 int is_enumeral =
3975 (SYMBOL_CLASS (syms[i].symbol) == LOC_CONST
3976 && SYMBOL_TYPE (syms[i].symbol) != NULL
3977 && TYPE_CODE (SYMBOL_TYPE (syms[i].symbol)) == TYPE_CODE_ENUM);
3978 struct symtab *symtab = NULL;
3979
3980 if (SYMBOL_OBJFILE_OWNED (syms[i].symbol))
3981 symtab = symbol_symtab (syms[i].symbol);
3982
3983 if (SYMBOL_LINE (syms[i].symbol) != 0 && symtab != NULL)
3984 {
3985 printf_unfiltered ("[%d] ", i + first_choice);
3986 ada_print_symbol_signature (gdb_stdout, syms[i].symbol,
3987 &type_print_raw_options);
3988 printf_unfiltered (_(" at %s:%d\n"),
3989 symtab_to_filename_for_display (symtab),
3990 SYMBOL_LINE (syms[i].symbol));
3991 }
3992 else if (is_enumeral
3993 && TYPE_NAME (SYMBOL_TYPE (syms[i].symbol)) != NULL)
3994 {
3995 printf_unfiltered (("[%d] "), i + first_choice);
3996 ada_print_type (SYMBOL_TYPE (syms[i].symbol), NULL,
3997 gdb_stdout, -1, 0, &type_print_raw_options);
3998 printf_unfiltered (_("'(%s) (enumeral)\n"),
3999 SYMBOL_PRINT_NAME (syms[i].symbol));
4000 }
4001 else
4002 {
4003 printf_unfiltered ("[%d] ", i + first_choice);
4004 ada_print_symbol_signature (gdb_stdout, syms[i].symbol,
4005 &type_print_raw_options);
4006
4007 if (symtab != NULL)
4008 printf_unfiltered (is_enumeral
4009 ? _(" in %s (enumeral)\n")
4010 : _(" at %s:?\n"),
4011 symtab_to_filename_for_display (symtab));
4012 else
4013 printf_unfiltered (is_enumeral
4014 ? _(" (enumeral)\n")
4015 : _(" at ?\n"));
4016 }
4017 }
4018 }
4019
4020 n_chosen = get_selections (chosen, nsyms, max_results, max_results > 1,
4021 "overload-choice");
4022
4023 for (i = 0; i < n_chosen; i += 1)
4024 syms[i] = syms[chosen[i]];
4025
4026 return n_chosen;
4027 }
4028
4029 /* Read and validate a set of numeric choices from the user in the
4030 range 0 .. N_CHOICES-1. Place the results in increasing
4031 order in CHOICES[0 .. N-1], and return N.
4032
4033 The user types choices as a sequence of numbers on one line
4034 separated by blanks, encoding them as follows:
4035
4036 + A choice of 0 means to cancel the selection, throwing an error.
4037 + If IS_ALL_CHOICE, a choice of 1 selects the entire set 0 .. N_CHOICES-1.
4038 + The user chooses k by typing k+IS_ALL_CHOICE+1.
4039
4040 The user is not allowed to choose more than MAX_RESULTS values.
4041
4042 ANNOTATION_SUFFIX, if present, is used to annotate the input
4043 prompts (for use with the -f switch). */
4044
4045 int
4046 get_selections (int *choices, int n_choices, int max_results,
4047 int is_all_choice, const char *annotation_suffix)
4048 {
4049 char *args;
4050 const char *prompt;
4051 int n_chosen;
4052 int first_choice = is_all_choice ? 2 : 1;
4053
4054 prompt = getenv ("PS2");
4055 if (prompt == NULL)
4056 prompt = "> ";
4057
4058 args = command_line_input (prompt, 0, annotation_suffix);
4059
4060 if (args == NULL)
4061 error_no_arg (_("one or more choice numbers"));
4062
4063 n_chosen = 0;
4064
4065 /* Set choices[0 .. n_chosen-1] to the users' choices in ascending
4066 order, as given in args. Choices are validated. */
4067 while (1)
4068 {
4069 char *args2;
4070 int choice, j;
4071
4072 args = skip_spaces (args);
4073 if (*args == '\0' && n_chosen == 0)
4074 error_no_arg (_("one or more choice numbers"));
4075 else if (*args == '\0')
4076 break;
4077
4078 choice = strtol (args, &args2, 10);
4079 if (args == args2 || choice < 0
4080 || choice > n_choices + first_choice - 1)
4081 error (_("Argument must be choice number"));
4082 args = args2;
4083
4084 if (choice == 0)
4085 error (_("cancelled"));
4086
4087 if (choice < first_choice)
4088 {
4089 n_chosen = n_choices;
4090 for (j = 0; j < n_choices; j += 1)
4091 choices[j] = j;
4092 break;
4093 }
4094 choice -= first_choice;
4095
4096 for (j = n_chosen - 1; j >= 0 && choice < choices[j]; j -= 1)
4097 {
4098 }
4099
4100 if (j < 0 || choice != choices[j])
4101 {
4102 int k;
4103
4104 for (k = n_chosen - 1; k > j; k -= 1)
4105 choices[k + 1] = choices[k];
4106 choices[j + 1] = choice;
4107 n_chosen += 1;
4108 }
4109 }
4110
4111 if (n_chosen > max_results)
4112 error (_("Select no more than %d of the above"), max_results);
4113
4114 return n_chosen;
4115 }
4116
4117 /* Replace the operator of length OPLEN at position PC in *EXPP with a call
4118 on the function identified by SYM and BLOCK, and taking NARGS
4119 arguments. Update *EXPP as needed to hold more space. */
4120
4121 static void
4122 replace_operator_with_call (expression_up *expp, int pc, int nargs,
4123 int oplen, struct symbol *sym,
4124 const struct block *block)
4125 {
4126 /* A new expression, with 6 more elements (3 for funcall, 4 for function
4127 symbol, -oplen for operator being replaced). */
4128 struct expression *newexp = (struct expression *)
4129 xzalloc (sizeof (struct expression)
4130 + EXP_ELEM_TO_BYTES ((*expp)->nelts + 7 - oplen));
4131 struct expression *exp = expp->get ();
4132
4133 newexp->nelts = exp->nelts + 7 - oplen;
4134 newexp->language_defn = exp->language_defn;
4135 newexp->gdbarch = exp->gdbarch;
4136 memcpy (newexp->elts, exp->elts, EXP_ELEM_TO_BYTES (pc));
4137 memcpy (newexp->elts + pc + 7, exp->elts + pc + oplen,
4138 EXP_ELEM_TO_BYTES (exp->nelts - pc - oplen));
4139
4140 newexp->elts[pc].opcode = newexp->elts[pc + 2].opcode = OP_FUNCALL;
4141 newexp->elts[pc + 1].longconst = (LONGEST) nargs;
4142
4143 newexp->elts[pc + 3].opcode = newexp->elts[pc + 6].opcode = OP_VAR_VALUE;
4144 newexp->elts[pc + 4].block = block;
4145 newexp->elts[pc + 5].symbol = sym;
4146
4147 expp->reset (newexp);
4148 }
4149
4150 /* Type-class predicates */
4151
4152 /* True iff TYPE is numeric (i.e., an INT, RANGE (of numeric type),
4153 or FLOAT). */
4154
4155 static int
4156 numeric_type_p (struct type *type)
4157 {
4158 if (type == NULL)
4159 return 0;
4160 else
4161 {
4162 switch (TYPE_CODE (type))
4163 {
4164 case TYPE_CODE_INT:
4165 case TYPE_CODE_FLT:
4166 return 1;
4167 case TYPE_CODE_RANGE:
4168 return (type == TYPE_TARGET_TYPE (type)
4169 || numeric_type_p (TYPE_TARGET_TYPE (type)));
4170 default:
4171 return 0;
4172 }
4173 }
4174 }
4175
4176 /* True iff TYPE is integral (an INT or RANGE of INTs). */
4177
4178 static int
4179 integer_type_p (struct type *type)
4180 {
4181 if (type == NULL)
4182 return 0;
4183 else
4184 {
4185 switch (TYPE_CODE (type))
4186 {
4187 case TYPE_CODE_INT:
4188 return 1;
4189 case TYPE_CODE_RANGE:
4190 return (type == TYPE_TARGET_TYPE (type)
4191 || integer_type_p (TYPE_TARGET_TYPE (type)));
4192 default:
4193 return 0;
4194 }
4195 }
4196 }
4197
4198 /* True iff TYPE is scalar (INT, RANGE, FLOAT, ENUM). */
4199
4200 static int
4201 scalar_type_p (struct type *type)
4202 {
4203 if (type == NULL)
4204 return 0;
4205 else
4206 {
4207 switch (TYPE_CODE (type))
4208 {
4209 case TYPE_CODE_INT:
4210 case TYPE_CODE_RANGE:
4211 case TYPE_CODE_ENUM:
4212 case TYPE_CODE_FLT:
4213 return 1;
4214 default:
4215 return 0;
4216 }
4217 }
4218 }
4219
4220 /* True iff TYPE is discrete (INT, RANGE, ENUM). */
4221
4222 static int
4223 discrete_type_p (struct type *type)
4224 {
4225 if (type == NULL)
4226 return 0;
4227 else
4228 {
4229 switch (TYPE_CODE (type))
4230 {
4231 case TYPE_CODE_INT:
4232 case TYPE_CODE_RANGE:
4233 case TYPE_CODE_ENUM:
4234 case TYPE_CODE_BOOL:
4235 return 1;
4236 default:
4237 return 0;
4238 }
4239 }
4240 }
4241
4242 /* Returns non-zero if OP with operands in the vector ARGS could be
4243 a user-defined function. Errs on the side of pre-defined operators
4244 (i.e., result 0). */
4245
4246 static int
4247 possible_user_operator_p (enum exp_opcode op, struct value *args[])
4248 {
4249 struct type *type0 =
4250 (args[0] == NULL) ? NULL : ada_check_typedef (value_type (args[0]));
4251 struct type *type1 =
4252 (args[1] == NULL) ? NULL : ada_check_typedef (value_type (args[1]));
4253
4254 if (type0 == NULL)
4255 return 0;
4256
4257 switch (op)
4258 {
4259 default:
4260 return 0;
4261
4262 case BINOP_ADD:
4263 case BINOP_SUB:
4264 case BINOP_MUL:
4265 case BINOP_DIV:
4266 return (!(numeric_type_p (type0) && numeric_type_p (type1)));
4267
4268 case BINOP_REM:
4269 case BINOP_MOD:
4270 case BINOP_BITWISE_AND:
4271 case BINOP_BITWISE_IOR:
4272 case BINOP_BITWISE_XOR:
4273 return (!(integer_type_p (type0) && integer_type_p (type1)));
4274
4275 case BINOP_EQUAL:
4276 case BINOP_NOTEQUAL:
4277 case BINOP_LESS:
4278 case BINOP_GTR:
4279 case BINOP_LEQ:
4280 case BINOP_GEQ:
4281 return (!(scalar_type_p (type0) && scalar_type_p (type1)));
4282
4283 case BINOP_CONCAT:
4284 return !ada_is_array_type (type0) || !ada_is_array_type (type1);
4285
4286 case BINOP_EXP:
4287 return (!(numeric_type_p (type0) && integer_type_p (type1)));
4288
4289 case UNOP_NEG:
4290 case UNOP_PLUS:
4291 case UNOP_LOGICAL_NOT:
4292 case UNOP_ABS:
4293 return (!numeric_type_p (type0));
4294
4295 }
4296 }
4297 \f
4298 /* Renaming */
4299
4300 /* NOTES:
4301
4302 1. In the following, we assume that a renaming type's name may
4303 have an ___XD suffix. It would be nice if this went away at some
4304 point.
4305 2. We handle both the (old) purely type-based representation of
4306 renamings and the (new) variable-based encoding. At some point,
4307 it is devoutly to be hoped that the former goes away
4308 (FIXME: hilfinger-2007-07-09).
4309 3. Subprogram renamings are not implemented, although the XRS
4310 suffix is recognized (FIXME: hilfinger-2007-07-09). */
4311
4312 /* If SYM encodes a renaming,
4313
4314 <renaming> renames <renamed entity>,
4315
4316 sets *LEN to the length of the renamed entity's name,
4317 *RENAMED_ENTITY to that name (not null-terminated), and *RENAMING_EXPR to
4318 the string describing the subcomponent selected from the renamed
4319 entity. Returns ADA_NOT_RENAMING if SYM does not encode a renaming
4320 (in which case, the values of *RENAMED_ENTITY, *LEN, and *RENAMING_EXPR
4321 are undefined). Otherwise, returns a value indicating the category
4322 of entity renamed: an object (ADA_OBJECT_RENAMING), exception
4323 (ADA_EXCEPTION_RENAMING), package (ADA_PACKAGE_RENAMING), or
4324 subprogram (ADA_SUBPROGRAM_RENAMING). Does no allocation; the
4325 strings returned in *RENAMED_ENTITY and *RENAMING_EXPR should not be
4326 deallocated. The values of RENAMED_ENTITY, LEN, or RENAMING_EXPR
4327 may be NULL, in which case they are not assigned.
4328
4329 [Currently, however, GCC does not generate subprogram renamings.] */
4330
4331 enum ada_renaming_category
4332 ada_parse_renaming (struct symbol *sym,
4333 const char **renamed_entity, int *len,
4334 const char **renaming_expr)
4335 {
4336 enum ada_renaming_category kind;
4337 const char *info;
4338 const char *suffix;
4339
4340 if (sym == NULL)
4341 return ADA_NOT_RENAMING;
4342 switch (SYMBOL_CLASS (sym))
4343 {
4344 default:
4345 return ADA_NOT_RENAMING;
4346 case LOC_TYPEDEF:
4347 return parse_old_style_renaming (SYMBOL_TYPE (sym),
4348 renamed_entity, len, renaming_expr);
4349 case LOC_LOCAL:
4350 case LOC_STATIC:
4351 case LOC_COMPUTED:
4352 case LOC_OPTIMIZED_OUT:
4353 info = strstr (SYMBOL_LINKAGE_NAME (sym), "___XR");
4354 if (info == NULL)
4355 return ADA_NOT_RENAMING;
4356 switch (info[5])
4357 {
4358 case '_':
4359 kind = ADA_OBJECT_RENAMING;
4360 info += 6;
4361 break;
4362 case 'E':
4363 kind = ADA_EXCEPTION_RENAMING;
4364 info += 7;
4365 break;
4366 case 'P':
4367 kind = ADA_PACKAGE_RENAMING;
4368 info += 7;
4369 break;
4370 case 'S':
4371 kind = ADA_SUBPROGRAM_RENAMING;
4372 info += 7;
4373 break;
4374 default:
4375 return ADA_NOT_RENAMING;
4376 }
4377 }
4378
4379 if (renamed_entity != NULL)
4380 *renamed_entity = info;
4381 suffix = strstr (info, "___XE");
4382 if (suffix == NULL || suffix == info)
4383 return ADA_NOT_RENAMING;
4384 if (len != NULL)
4385 *len = strlen (info) - strlen (suffix);
4386 suffix += 5;
4387 if (renaming_expr != NULL)
4388 *renaming_expr = suffix;
4389 return kind;
4390 }
4391
4392 /* Assuming TYPE encodes a renaming according to the old encoding in
4393 exp_dbug.ads, returns details of that renaming in *RENAMED_ENTITY,
4394 *LEN, and *RENAMING_EXPR, as for ada_parse_renaming, above. Returns
4395 ADA_NOT_RENAMING otherwise. */
4396 static enum ada_renaming_category
4397 parse_old_style_renaming (struct type *type,
4398 const char **renamed_entity, int *len,
4399 const char **renaming_expr)
4400 {
4401 enum ada_renaming_category kind;
4402 const char *name;
4403 const char *info;
4404 const char *suffix;
4405
4406 if (type == NULL || TYPE_CODE (type) != TYPE_CODE_ENUM
4407 || TYPE_NFIELDS (type) != 1)
4408 return ADA_NOT_RENAMING;
4409
4410 name = type_name_no_tag (type);
4411 if (name == NULL)
4412 return ADA_NOT_RENAMING;
4413
4414 name = strstr (name, "___XR");
4415 if (name == NULL)
4416 return ADA_NOT_RENAMING;
4417 switch (name[5])
4418 {
4419 case '\0':
4420 case '_':
4421 kind = ADA_OBJECT_RENAMING;
4422 break;
4423 case 'E':
4424 kind = ADA_EXCEPTION_RENAMING;
4425 break;
4426 case 'P':
4427 kind = ADA_PACKAGE_RENAMING;
4428 break;
4429 case 'S':
4430 kind = ADA_SUBPROGRAM_RENAMING;
4431 break;
4432 default:
4433 return ADA_NOT_RENAMING;
4434 }
4435
4436 info = TYPE_FIELD_NAME (type, 0);
4437 if (info == NULL)
4438 return ADA_NOT_RENAMING;
4439 if (renamed_entity != NULL)
4440 *renamed_entity = info;
4441 suffix = strstr (info, "___XE");
4442 if (renaming_expr != NULL)
4443 *renaming_expr = suffix + 5;
4444 if (suffix == NULL || suffix == info)
4445 return ADA_NOT_RENAMING;
4446 if (len != NULL)
4447 *len = suffix - info;
4448 return kind;
4449 }
4450
4451 /* Compute the value of the given RENAMING_SYM, which is expected to
4452 be a symbol encoding a renaming expression. BLOCK is the block
4453 used to evaluate the renaming. */
4454
4455 static struct value *
4456 ada_read_renaming_var_value (struct symbol *renaming_sym,
4457 const struct block *block)
4458 {
4459 const char *sym_name;
4460
4461 sym_name = SYMBOL_LINKAGE_NAME (renaming_sym);
4462 expression_up expr = parse_exp_1 (&sym_name, 0, block, 0);
4463 return evaluate_expression (expr.get ());
4464 }
4465 \f
4466
4467 /* Evaluation: Function Calls */
4468
4469 /* Return an lvalue containing the value VAL. This is the identity on
4470 lvalues, and otherwise has the side-effect of allocating memory
4471 in the inferior where a copy of the value contents is copied. */
4472
4473 static struct value *
4474 ensure_lval (struct value *val)
4475 {
4476 if (VALUE_LVAL (val) == not_lval
4477 || VALUE_LVAL (val) == lval_internalvar)
4478 {
4479 int len = TYPE_LENGTH (ada_check_typedef (value_type (val)));
4480 const CORE_ADDR addr =
4481 value_as_long (value_allocate_space_in_inferior (len));
4482
4483 VALUE_LVAL (val) = lval_memory;
4484 set_value_address (val, addr);
4485 write_memory (addr, value_contents (val), len);
4486 }
4487
4488 return val;
4489 }
4490
4491 /* Return the value ACTUAL, converted to be an appropriate value for a
4492 formal of type FORMAL_TYPE. Use *SP as a stack pointer for
4493 allocating any necessary descriptors (fat pointers), or copies of
4494 values not residing in memory, updating it as needed. */
4495
4496 struct value *
4497 ada_convert_actual (struct value *actual, struct type *formal_type0)
4498 {
4499 struct type *actual_type = ada_check_typedef (value_type (actual));
4500 struct type *formal_type = ada_check_typedef (formal_type0);
4501 struct type *formal_target =
4502 TYPE_CODE (formal_type) == TYPE_CODE_PTR
4503 ? ada_check_typedef (TYPE_TARGET_TYPE (formal_type)) : formal_type;
4504 struct type *actual_target =
4505 TYPE_CODE (actual_type) == TYPE_CODE_PTR
4506 ? ada_check_typedef (TYPE_TARGET_TYPE (actual_type)) : actual_type;
4507
4508 if (ada_is_array_descriptor_type (formal_target)
4509 && TYPE_CODE (actual_target) == TYPE_CODE_ARRAY)
4510 return make_array_descriptor (formal_type, actual);
4511 else if (TYPE_CODE (formal_type) == TYPE_CODE_PTR
4512 || TYPE_CODE (formal_type) == TYPE_CODE_REF)
4513 {
4514 struct value *result;
4515
4516 if (TYPE_CODE (formal_target) == TYPE_CODE_ARRAY
4517 && ada_is_array_descriptor_type (actual_target))
4518 result = desc_data (actual);
4519 else if (TYPE_CODE (formal_type) != TYPE_CODE_PTR)
4520 {
4521 if (VALUE_LVAL (actual) != lval_memory)
4522 {
4523 struct value *val;
4524
4525 actual_type = ada_check_typedef (value_type (actual));
4526 val = allocate_value (actual_type);
4527 memcpy ((char *) value_contents_raw (val),
4528 (char *) value_contents (actual),
4529 TYPE_LENGTH (actual_type));
4530 actual = ensure_lval (val);
4531 }
4532 result = value_addr (actual);
4533 }
4534 else
4535 return actual;
4536 return value_cast_pointers (formal_type, result, 0);
4537 }
4538 else if (TYPE_CODE (actual_type) == TYPE_CODE_PTR)
4539 return ada_value_ind (actual);
4540 else if (ada_is_aligner_type (formal_type))
4541 {
4542 /* We need to turn this parameter into an aligner type
4543 as well. */
4544 struct value *aligner = allocate_value (formal_type);
4545 struct value *component = ada_value_struct_elt (aligner, "F", 0);
4546
4547 value_assign_to_component (aligner, component, actual);
4548 return aligner;
4549 }
4550
4551 return actual;
4552 }
4553
4554 /* Convert VALUE (which must be an address) to a CORE_ADDR that is a pointer of
4555 type TYPE. This is usually an inefficient no-op except on some targets
4556 (such as AVR) where the representation of a pointer and an address
4557 differs. */
4558
4559 static CORE_ADDR
4560 value_pointer (struct value *value, struct type *type)
4561 {
4562 struct gdbarch *gdbarch = get_type_arch (type);
4563 unsigned len = TYPE_LENGTH (type);
4564 gdb_byte *buf = (gdb_byte *) alloca (len);
4565 CORE_ADDR addr;
4566
4567 addr = value_address (value);
4568 gdbarch_address_to_pointer (gdbarch, type, buf, addr);
4569 addr = extract_unsigned_integer (buf, len, gdbarch_byte_order (gdbarch));
4570 return addr;
4571 }
4572
4573
4574 /* Push a descriptor of type TYPE for array value ARR on the stack at
4575 *SP, updating *SP to reflect the new descriptor. Return either
4576 an lvalue representing the new descriptor, or (if TYPE is a pointer-
4577 to-descriptor type rather than a descriptor type), a struct value *
4578 representing a pointer to this descriptor. */
4579
4580 static struct value *
4581 make_array_descriptor (struct type *type, struct value *arr)
4582 {
4583 struct type *bounds_type = desc_bounds_type (type);
4584 struct type *desc_type = desc_base_type (type);
4585 struct value *descriptor = allocate_value (desc_type);
4586 struct value *bounds = allocate_value (bounds_type);
4587 int i;
4588
4589 for (i = ada_array_arity (ada_check_typedef (value_type (arr)));
4590 i > 0; i -= 1)
4591 {
4592 modify_field (value_type (bounds), value_contents_writeable (bounds),
4593 ada_array_bound (arr, i, 0),
4594 desc_bound_bitpos (bounds_type, i, 0),
4595 desc_bound_bitsize (bounds_type, i, 0));
4596 modify_field (value_type (bounds), value_contents_writeable (bounds),
4597 ada_array_bound (arr, i, 1),
4598 desc_bound_bitpos (bounds_type, i, 1),
4599 desc_bound_bitsize (bounds_type, i, 1));
4600 }
4601
4602 bounds = ensure_lval (bounds);
4603
4604 modify_field (value_type (descriptor),
4605 value_contents_writeable (descriptor),
4606 value_pointer (ensure_lval (arr),
4607 TYPE_FIELD_TYPE (desc_type, 0)),
4608 fat_pntr_data_bitpos (desc_type),
4609 fat_pntr_data_bitsize (desc_type));
4610
4611 modify_field (value_type (descriptor),
4612 value_contents_writeable (descriptor),
4613 value_pointer (bounds,
4614 TYPE_FIELD_TYPE (desc_type, 1)),
4615 fat_pntr_bounds_bitpos (desc_type),
4616 fat_pntr_bounds_bitsize (desc_type));
4617
4618 descriptor = ensure_lval (descriptor);
4619
4620 if (TYPE_CODE (type) == TYPE_CODE_PTR)
4621 return value_addr (descriptor);
4622 else
4623 return descriptor;
4624 }
4625 \f
4626 /* Symbol Cache Module */
4627
4628 /* Performance measurements made as of 2010-01-15 indicate that
4629 this cache does bring some noticeable improvements. Depending
4630 on the type of entity being printed, the cache can make it as much
4631 as an order of magnitude faster than without it.
4632
4633 The descriptive type DWARF extension has significantly reduced
4634 the need for this cache, at least when DWARF is being used. However,
4635 even in this case, some expensive name-based symbol searches are still
4636 sometimes necessary - to find an XVZ variable, mostly. */
4637
4638 /* Initialize the contents of SYM_CACHE. */
4639
4640 static void
4641 ada_init_symbol_cache (struct ada_symbol_cache *sym_cache)
4642 {
4643 obstack_init (&sym_cache->cache_space);
4644 memset (sym_cache->root, '\000', sizeof (sym_cache->root));
4645 }
4646
4647 /* Free the memory used by SYM_CACHE. */
4648
4649 static void
4650 ada_free_symbol_cache (struct ada_symbol_cache *sym_cache)
4651 {
4652 obstack_free (&sym_cache->cache_space, NULL);
4653 xfree (sym_cache);
4654 }
4655
4656 /* Return the symbol cache associated to the given program space PSPACE.
4657 If not allocated for this PSPACE yet, allocate and initialize one. */
4658
4659 static struct ada_symbol_cache *
4660 ada_get_symbol_cache (struct program_space *pspace)
4661 {
4662 struct ada_pspace_data *pspace_data = get_ada_pspace_data (pspace);
4663
4664 if (pspace_data->sym_cache == NULL)
4665 {
4666 pspace_data->sym_cache = XCNEW (struct ada_symbol_cache);
4667 ada_init_symbol_cache (pspace_data->sym_cache);
4668 }
4669
4670 return pspace_data->sym_cache;
4671 }
4672
4673 /* Clear all entries from the symbol cache. */
4674
4675 static void
4676 ada_clear_symbol_cache (void)
4677 {
4678 struct ada_symbol_cache *sym_cache
4679 = ada_get_symbol_cache (current_program_space);
4680
4681 obstack_free (&sym_cache->cache_space, NULL);
4682 ada_init_symbol_cache (sym_cache);
4683 }
4684
4685 /* Search our cache for an entry matching NAME and DOMAIN.
4686 Return it if found, or NULL otherwise. */
4687
4688 static struct cache_entry **
4689 find_entry (const char *name, domain_enum domain)
4690 {
4691 struct ada_symbol_cache *sym_cache
4692 = ada_get_symbol_cache (current_program_space);
4693 int h = msymbol_hash (name) % HASH_SIZE;
4694 struct cache_entry **e;
4695
4696 for (e = &sym_cache->root[h]; *e != NULL; e = &(*e)->next)
4697 {
4698 if (domain == (*e)->domain && strcmp (name, (*e)->name) == 0)
4699 return e;
4700 }
4701 return NULL;
4702 }
4703
4704 /* Search the symbol cache for an entry matching NAME and DOMAIN.
4705 Return 1 if found, 0 otherwise.
4706
4707 If an entry was found and SYM is not NULL, set *SYM to the entry's
4708 SYM. Same principle for BLOCK if not NULL. */
4709
4710 static int
4711 lookup_cached_symbol (const char *name, domain_enum domain,
4712 struct symbol **sym, const struct block **block)
4713 {
4714 struct cache_entry **e = find_entry (name, domain);
4715
4716 if (e == NULL)
4717 return 0;
4718 if (sym != NULL)
4719 *sym = (*e)->sym;
4720 if (block != NULL)
4721 *block = (*e)->block;
4722 return 1;
4723 }
4724
4725 /* Assuming that (SYM, BLOCK) is the result of the lookup of NAME
4726 in domain DOMAIN, save this result in our symbol cache. */
4727
4728 static void
4729 cache_symbol (const char *name, domain_enum domain, struct symbol *sym,
4730 const struct block *block)
4731 {
4732 struct ada_symbol_cache *sym_cache
4733 = ada_get_symbol_cache (current_program_space);
4734 int h;
4735 char *copy;
4736 struct cache_entry *e;
4737
4738 /* Symbols for builtin types don't have a block.
4739 For now don't cache such symbols. */
4740 if (sym != NULL && !SYMBOL_OBJFILE_OWNED (sym))
4741 return;
4742
4743 /* If the symbol is a local symbol, then do not cache it, as a search
4744 for that symbol depends on the context. To determine whether
4745 the symbol is local or not, we check the block where we found it
4746 against the global and static blocks of its associated symtab. */
4747 if (sym
4748 && BLOCKVECTOR_BLOCK (SYMTAB_BLOCKVECTOR (symbol_symtab (sym)),
4749 GLOBAL_BLOCK) != block
4750 && BLOCKVECTOR_BLOCK (SYMTAB_BLOCKVECTOR (symbol_symtab (sym)),
4751 STATIC_BLOCK) != block)
4752 return;
4753
4754 h = msymbol_hash (name) % HASH_SIZE;
4755 e = (struct cache_entry *) obstack_alloc (&sym_cache->cache_space,
4756 sizeof (*e));
4757 e->next = sym_cache->root[h];
4758 sym_cache->root[h] = e;
4759 e->name = copy
4760 = (char *) obstack_alloc (&sym_cache->cache_space, strlen (name) + 1);
4761 strcpy (copy, name);
4762 e->sym = sym;
4763 e->domain = domain;
4764 e->block = block;
4765 }
4766 \f
4767 /* Symbol Lookup */
4768
4769 /* Return the symbol name match type that should be used used when
4770 searching for all symbols matching LOOKUP_NAME.
4771
4772 LOOKUP_NAME is expected to be a symbol name after transformation
4773 for Ada lookups (see ada_name_for_lookup). */
4774
4775 static symbol_name_match_type
4776 name_match_type_from_name (const char *lookup_name)
4777 {
4778 return (strstr (lookup_name, "__") == NULL
4779 ? symbol_name_match_type::WILD
4780 : symbol_name_match_type::FULL);
4781 }
4782
4783 /* Return the result of a standard (literal, C-like) lookup of NAME in
4784 given DOMAIN, visible from lexical block BLOCK. */
4785
4786 static struct symbol *
4787 standard_lookup (const char *name, const struct block *block,
4788 domain_enum domain)
4789 {
4790 /* Initialize it just to avoid a GCC false warning. */
4791 struct block_symbol sym = {NULL, NULL};
4792
4793 if (lookup_cached_symbol (name, domain, &sym.symbol, NULL))
4794 return sym.symbol;
4795 sym = lookup_symbol_in_language (name, block, domain, language_c, 0);
4796 cache_symbol (name, domain, sym.symbol, sym.block);
4797 return sym.symbol;
4798 }
4799
4800
4801 /* Non-zero iff there is at least one non-function/non-enumeral symbol
4802 in the symbol fields of SYMS[0..N-1]. We treat enumerals as functions,
4803 since they contend in overloading in the same way. */
4804 static int
4805 is_nonfunction (struct block_symbol syms[], int n)
4806 {
4807 int i;
4808
4809 for (i = 0; i < n; i += 1)
4810 if (TYPE_CODE (SYMBOL_TYPE (syms[i].symbol)) != TYPE_CODE_FUNC
4811 && (TYPE_CODE (SYMBOL_TYPE (syms[i].symbol)) != TYPE_CODE_ENUM
4812 || SYMBOL_CLASS (syms[i].symbol) != LOC_CONST))
4813 return 1;
4814
4815 return 0;
4816 }
4817
4818 /* If true (non-zero), then TYPE0 and TYPE1 represent equivalent
4819 struct types. Otherwise, they may not. */
4820
4821 static int
4822 equiv_types (struct type *type0, struct type *type1)
4823 {
4824 if (type0 == type1)
4825 return 1;
4826 if (type0 == NULL || type1 == NULL
4827 || TYPE_CODE (type0) != TYPE_CODE (type1))
4828 return 0;
4829 if ((TYPE_CODE (type0) == TYPE_CODE_STRUCT
4830 || TYPE_CODE (type0) == TYPE_CODE_ENUM)
4831 && ada_type_name (type0) != NULL && ada_type_name (type1) != NULL
4832 && strcmp (ada_type_name (type0), ada_type_name (type1)) == 0)
4833 return 1;
4834
4835 return 0;
4836 }
4837
4838 /* True iff SYM0 represents the same entity as SYM1, or one that is
4839 no more defined than that of SYM1. */
4840
4841 static int
4842 lesseq_defined_than (struct symbol *sym0, struct symbol *sym1)
4843 {
4844 if (sym0 == sym1)
4845 return 1;
4846 if (SYMBOL_DOMAIN (sym0) != SYMBOL_DOMAIN (sym1)
4847 || SYMBOL_CLASS (sym0) != SYMBOL_CLASS (sym1))
4848 return 0;
4849
4850 switch (SYMBOL_CLASS (sym0))
4851 {
4852 case LOC_UNDEF:
4853 return 1;
4854 case LOC_TYPEDEF:
4855 {
4856 struct type *type0 = SYMBOL_TYPE (sym0);
4857 struct type *type1 = SYMBOL_TYPE (sym1);
4858 const char *name0 = SYMBOL_LINKAGE_NAME (sym0);
4859 const char *name1 = SYMBOL_LINKAGE_NAME (sym1);
4860 int len0 = strlen (name0);
4861
4862 return
4863 TYPE_CODE (type0) == TYPE_CODE (type1)
4864 && (equiv_types (type0, type1)
4865 || (len0 < strlen (name1) && strncmp (name0, name1, len0) == 0
4866 && startswith (name1 + len0, "___XV")));
4867 }
4868 case LOC_CONST:
4869 return SYMBOL_VALUE (sym0) == SYMBOL_VALUE (sym1)
4870 && equiv_types (SYMBOL_TYPE (sym0), SYMBOL_TYPE (sym1));
4871 default:
4872 return 0;
4873 }
4874 }
4875
4876 /* Append (SYM,BLOCK,SYMTAB) to the end of the array of struct block_symbol
4877 records in OBSTACKP. Do nothing if SYM is a duplicate. */
4878
4879 static void
4880 add_defn_to_vec (struct obstack *obstackp,
4881 struct symbol *sym,
4882 const struct block *block)
4883 {
4884 int i;
4885 struct block_symbol *prevDefns = defns_collected (obstackp, 0);
4886
4887 /* Do not try to complete stub types, as the debugger is probably
4888 already scanning all symbols matching a certain name at the
4889 time when this function is called. Trying to replace the stub
4890 type by its associated full type will cause us to restart a scan
4891 which may lead to an infinite recursion. Instead, the client
4892 collecting the matching symbols will end up collecting several
4893 matches, with at least one of them complete. It can then filter
4894 out the stub ones if needed. */
4895
4896 for (i = num_defns_collected (obstackp) - 1; i >= 0; i -= 1)
4897 {
4898 if (lesseq_defined_than (sym, prevDefns[i].symbol))
4899 return;
4900 else if (lesseq_defined_than (prevDefns[i].symbol, sym))
4901 {
4902 prevDefns[i].symbol = sym;
4903 prevDefns[i].block = block;
4904 return;
4905 }
4906 }
4907
4908 {
4909 struct block_symbol info;
4910
4911 info.symbol = sym;
4912 info.block = block;
4913 obstack_grow (obstackp, &info, sizeof (struct block_symbol));
4914 }
4915 }
4916
4917 /* Number of block_symbol structures currently collected in current vector in
4918 OBSTACKP. */
4919
4920 static int
4921 num_defns_collected (struct obstack *obstackp)
4922 {
4923 return obstack_object_size (obstackp) / sizeof (struct block_symbol);
4924 }
4925
4926 /* Vector of block_symbol structures currently collected in current vector in
4927 OBSTACKP. If FINISH, close off the vector and return its final address. */
4928
4929 static struct block_symbol *
4930 defns_collected (struct obstack *obstackp, int finish)
4931 {
4932 if (finish)
4933 return (struct block_symbol *) obstack_finish (obstackp);
4934 else
4935 return (struct block_symbol *) obstack_base (obstackp);
4936 }
4937
4938 /* Return a bound minimal symbol matching NAME according to Ada
4939 decoding rules. Returns an invalid symbol if there is no such
4940 minimal symbol. Names prefixed with "standard__" are handled
4941 specially: "standard__" is first stripped off, and only static and
4942 global symbols are searched. */
4943
4944 struct bound_minimal_symbol
4945 ada_lookup_simple_minsym (const char *name)
4946 {
4947 struct bound_minimal_symbol result;
4948 struct objfile *objfile;
4949 struct minimal_symbol *msymbol;
4950
4951 memset (&result, 0, sizeof (result));
4952
4953 symbol_name_match_type match_type = name_match_type_from_name (name);
4954 lookup_name_info lookup_name (name, match_type);
4955
4956 symbol_name_matcher_ftype *match_name
4957 = ada_get_symbol_name_matcher (lookup_name);
4958
4959 ALL_MSYMBOLS (objfile, msymbol)
4960 {
4961 if (match_name (MSYMBOL_LINKAGE_NAME (msymbol), lookup_name, NULL)
4962 && MSYMBOL_TYPE (msymbol) != mst_solib_trampoline)
4963 {
4964 result.minsym = msymbol;
4965 result.objfile = objfile;
4966 break;
4967 }
4968 }
4969
4970 return result;
4971 }
4972
4973 /* For all subprograms that statically enclose the subprogram of the
4974 selected frame, add symbols matching identifier NAME in DOMAIN
4975 and their blocks to the list of data in OBSTACKP, as for
4976 ada_add_block_symbols (q.v.). If WILD_MATCH_P, treat as NAME
4977 with a wildcard prefix. */
4978
4979 static void
4980 add_symbols_from_enclosing_procs (struct obstack *obstackp,
4981 const lookup_name_info &lookup_name,
4982 domain_enum domain)
4983 {
4984 }
4985
4986 /* True if TYPE is definitely an artificial type supplied to a symbol
4987 for which no debugging information was given in the symbol file. */
4988
4989 static int
4990 is_nondebugging_type (struct type *type)
4991 {
4992 const char *name = ada_type_name (type);
4993
4994 return (name != NULL && strcmp (name, "<variable, no debug info>") == 0);
4995 }
4996
4997 /* Return nonzero if TYPE1 and TYPE2 are two enumeration types
4998 that are deemed "identical" for practical purposes.
4999
5000 This function assumes that TYPE1 and TYPE2 are both TYPE_CODE_ENUM
5001 types and that their number of enumerals is identical (in other
5002 words, TYPE_NFIELDS (type1) == TYPE_NFIELDS (type2)). */
5003
5004 static int
5005 ada_identical_enum_types_p (struct type *type1, struct type *type2)
5006 {
5007 int i;
5008
5009 /* The heuristic we use here is fairly conservative. We consider
5010 that 2 enumerate types are identical if they have the same
5011 number of enumerals and that all enumerals have the same
5012 underlying value and name. */
5013
5014 /* All enums in the type should have an identical underlying value. */
5015 for (i = 0; i < TYPE_NFIELDS (type1); i++)
5016 if (TYPE_FIELD_ENUMVAL (type1, i) != TYPE_FIELD_ENUMVAL (type2, i))
5017 return 0;
5018
5019 /* All enumerals should also have the same name (modulo any numerical
5020 suffix). */
5021 for (i = 0; i < TYPE_NFIELDS (type1); i++)
5022 {
5023 const char *name_1 = TYPE_FIELD_NAME (type1, i);
5024 const char *name_2 = TYPE_FIELD_NAME (type2, i);
5025 int len_1 = strlen (name_1);
5026 int len_2 = strlen (name_2);
5027
5028 ada_remove_trailing_digits (TYPE_FIELD_NAME (type1, i), &len_1);
5029 ada_remove_trailing_digits (TYPE_FIELD_NAME (type2, i), &len_2);
5030 if (len_1 != len_2
5031 || strncmp (TYPE_FIELD_NAME (type1, i),
5032 TYPE_FIELD_NAME (type2, i),
5033 len_1) != 0)
5034 return 0;
5035 }
5036
5037 return 1;
5038 }
5039
5040 /* Return nonzero if all the symbols in SYMS are all enumeral symbols
5041 that are deemed "identical" for practical purposes. Sometimes,
5042 enumerals are not strictly identical, but their types are so similar
5043 that they can be considered identical.
5044
5045 For instance, consider the following code:
5046
5047 type Color is (Black, Red, Green, Blue, White);
5048 type RGB_Color is new Color range Red .. Blue;
5049
5050 Type RGB_Color is a subrange of an implicit type which is a copy
5051 of type Color. If we call that implicit type RGB_ColorB ("B" is
5052 for "Base Type"), then type RGB_ColorB is a copy of type Color.
5053 As a result, when an expression references any of the enumeral
5054 by name (Eg. "print green"), the expression is technically
5055 ambiguous and the user should be asked to disambiguate. But
5056 doing so would only hinder the user, since it wouldn't matter
5057 what choice he makes, the outcome would always be the same.
5058 So, for practical purposes, we consider them as the same. */
5059
5060 static int
5061 symbols_are_identical_enums (struct block_symbol *syms, int nsyms)
5062 {
5063 int i;
5064
5065 /* Before performing a thorough comparison check of each type,
5066 we perform a series of inexpensive checks. We expect that these
5067 checks will quickly fail in the vast majority of cases, and thus
5068 help prevent the unnecessary use of a more expensive comparison.
5069 Said comparison also expects us to make some of these checks
5070 (see ada_identical_enum_types_p). */
5071
5072 /* Quick check: All symbols should have an enum type. */
5073 for (i = 0; i < nsyms; i++)
5074 if (TYPE_CODE (SYMBOL_TYPE (syms[i].symbol)) != TYPE_CODE_ENUM)
5075 return 0;
5076
5077 /* Quick check: They should all have the same value. */
5078 for (i = 1; i < nsyms; i++)
5079 if (SYMBOL_VALUE (syms[i].symbol) != SYMBOL_VALUE (syms[0].symbol))
5080 return 0;
5081
5082 /* Quick check: They should all have the same number of enumerals. */
5083 for (i = 1; i < nsyms; i++)
5084 if (TYPE_NFIELDS (SYMBOL_TYPE (syms[i].symbol))
5085 != TYPE_NFIELDS (SYMBOL_TYPE (syms[0].symbol)))
5086 return 0;
5087
5088 /* All the sanity checks passed, so we might have a set of
5089 identical enumeration types. Perform a more complete
5090 comparison of the type of each symbol. */
5091 for (i = 1; i < nsyms; i++)
5092 if (!ada_identical_enum_types_p (SYMBOL_TYPE (syms[i].symbol),
5093 SYMBOL_TYPE (syms[0].symbol)))
5094 return 0;
5095
5096 return 1;
5097 }
5098
5099 /* Remove any non-debugging symbols in SYMS[0 .. NSYMS-1] that definitely
5100 duplicate other symbols in the list (The only case I know of where
5101 this happens is when object files containing stabs-in-ecoff are
5102 linked with files containing ordinary ecoff debugging symbols (or no
5103 debugging symbols)). Modifies SYMS to squeeze out deleted entries.
5104 Returns the number of items in the modified list. */
5105
5106 static int
5107 remove_extra_symbols (struct block_symbol *syms, int nsyms)
5108 {
5109 int i, j;
5110
5111 /* We should never be called with less than 2 symbols, as there
5112 cannot be any extra symbol in that case. But it's easy to
5113 handle, since we have nothing to do in that case. */
5114 if (nsyms < 2)
5115 return nsyms;
5116
5117 i = 0;
5118 while (i < nsyms)
5119 {
5120 int remove_p = 0;
5121
5122 /* If two symbols have the same name and one of them is a stub type,
5123 the get rid of the stub. */
5124
5125 if (TYPE_STUB (SYMBOL_TYPE (syms[i].symbol))
5126 && SYMBOL_LINKAGE_NAME (syms[i].symbol) != NULL)
5127 {
5128 for (j = 0; j < nsyms; j++)
5129 {
5130 if (j != i
5131 && !TYPE_STUB (SYMBOL_TYPE (syms[j].symbol))
5132 && SYMBOL_LINKAGE_NAME (syms[j].symbol) != NULL
5133 && strcmp (SYMBOL_LINKAGE_NAME (syms[i].symbol),
5134 SYMBOL_LINKAGE_NAME (syms[j].symbol)) == 0)
5135 remove_p = 1;
5136 }
5137 }
5138
5139 /* Two symbols with the same name, same class and same address
5140 should be identical. */
5141
5142 else if (SYMBOL_LINKAGE_NAME (syms[i].symbol) != NULL
5143 && SYMBOL_CLASS (syms[i].symbol) == LOC_STATIC
5144 && is_nondebugging_type (SYMBOL_TYPE (syms[i].symbol)))
5145 {
5146 for (j = 0; j < nsyms; j += 1)
5147 {
5148 if (i != j
5149 && SYMBOL_LINKAGE_NAME (syms[j].symbol) != NULL
5150 && strcmp (SYMBOL_LINKAGE_NAME (syms[i].symbol),
5151 SYMBOL_LINKAGE_NAME (syms[j].symbol)) == 0
5152 && SYMBOL_CLASS (syms[i].symbol)
5153 == SYMBOL_CLASS (syms[j].symbol)
5154 && SYMBOL_VALUE_ADDRESS (syms[i].symbol)
5155 == SYMBOL_VALUE_ADDRESS (syms[j].symbol))
5156 remove_p = 1;
5157 }
5158 }
5159
5160 if (remove_p)
5161 {
5162 for (j = i + 1; j < nsyms; j += 1)
5163 syms[j - 1] = syms[j];
5164 nsyms -= 1;
5165 }
5166
5167 i += 1;
5168 }
5169
5170 /* If all the remaining symbols are identical enumerals, then
5171 just keep the first one and discard the rest.
5172
5173 Unlike what we did previously, we do not discard any entry
5174 unless they are ALL identical. This is because the symbol
5175 comparison is not a strict comparison, but rather a practical
5176 comparison. If all symbols are considered identical, then
5177 we can just go ahead and use the first one and discard the rest.
5178 But if we cannot reduce the list to a single element, we have
5179 to ask the user to disambiguate anyways. And if we have to
5180 present a multiple-choice menu, it's less confusing if the list
5181 isn't missing some choices that were identical and yet distinct. */
5182 if (symbols_are_identical_enums (syms, nsyms))
5183 nsyms = 1;
5184
5185 return nsyms;
5186 }
5187
5188 /* Given a type that corresponds to a renaming entity, use the type name
5189 to extract the scope (package name or function name, fully qualified,
5190 and following the GNAT encoding convention) where this renaming has been
5191 defined. The string returned needs to be deallocated after use. */
5192
5193 static char *
5194 xget_renaming_scope (struct type *renaming_type)
5195 {
5196 /* The renaming types adhere to the following convention:
5197 <scope>__<rename>___<XR extension>.
5198 So, to extract the scope, we search for the "___XR" extension,
5199 and then backtrack until we find the first "__". */
5200
5201 const char *name = type_name_no_tag (renaming_type);
5202 const char *suffix = strstr (name, "___XR");
5203 const char *last;
5204 int scope_len;
5205 char *scope;
5206
5207 /* Now, backtrack a bit until we find the first "__". Start looking
5208 at suffix - 3, as the <rename> part is at least one character long. */
5209
5210 for (last = suffix - 3; last > name; last--)
5211 if (last[0] == '_' && last[1] == '_')
5212 break;
5213
5214 /* Make a copy of scope and return it. */
5215
5216 scope_len = last - name;
5217 scope = (char *) xmalloc ((scope_len + 1) * sizeof (char));
5218
5219 strncpy (scope, name, scope_len);
5220 scope[scope_len] = '\0';
5221
5222 return scope;
5223 }
5224
5225 /* Return nonzero if NAME corresponds to a package name. */
5226
5227 static int
5228 is_package_name (const char *name)
5229 {
5230 /* Here, We take advantage of the fact that no symbols are generated
5231 for packages, while symbols are generated for each function.
5232 So the condition for NAME represent a package becomes equivalent
5233 to NAME not existing in our list of symbols. There is only one
5234 small complication with library-level functions (see below). */
5235
5236 char *fun_name;
5237
5238 /* If it is a function that has not been defined at library level,
5239 then we should be able to look it up in the symbols. */
5240 if (standard_lookup (name, NULL, VAR_DOMAIN) != NULL)
5241 return 0;
5242
5243 /* Library-level function names start with "_ada_". See if function
5244 "_ada_" followed by NAME can be found. */
5245
5246 /* Do a quick check that NAME does not contain "__", since library-level
5247 functions names cannot contain "__" in them. */
5248 if (strstr (name, "__") != NULL)
5249 return 0;
5250
5251 fun_name = xstrprintf ("_ada_%s", name);
5252
5253 return (standard_lookup (fun_name, NULL, VAR_DOMAIN) == NULL);
5254 }
5255
5256 /* Return nonzero if SYM corresponds to a renaming entity that is
5257 not visible from FUNCTION_NAME. */
5258
5259 static int
5260 old_renaming_is_invisible (const struct symbol *sym, const char *function_name)
5261 {
5262 char *scope;
5263 struct cleanup *old_chain;
5264
5265 if (SYMBOL_CLASS (sym) != LOC_TYPEDEF)
5266 return 0;
5267
5268 scope = xget_renaming_scope (SYMBOL_TYPE (sym));
5269 old_chain = make_cleanup (xfree, scope);
5270
5271 /* If the rename has been defined in a package, then it is visible. */
5272 if (is_package_name (scope))
5273 {
5274 do_cleanups (old_chain);
5275 return 0;
5276 }
5277
5278 /* Check that the rename is in the current function scope by checking
5279 that its name starts with SCOPE. */
5280
5281 /* If the function name starts with "_ada_", it means that it is
5282 a library-level function. Strip this prefix before doing the
5283 comparison, as the encoding for the renaming does not contain
5284 this prefix. */
5285 if (startswith (function_name, "_ada_"))
5286 function_name += 5;
5287
5288 {
5289 int is_invisible = !startswith (function_name, scope);
5290
5291 do_cleanups (old_chain);
5292 return is_invisible;
5293 }
5294 }
5295
5296 /* Remove entries from SYMS that corresponds to a renaming entity that
5297 is not visible from the function associated with CURRENT_BLOCK or
5298 that is superfluous due to the presence of more specific renaming
5299 information. Places surviving symbols in the initial entries of
5300 SYMS and returns the number of surviving symbols.
5301
5302 Rationale:
5303 First, in cases where an object renaming is implemented as a
5304 reference variable, GNAT may produce both the actual reference
5305 variable and the renaming encoding. In this case, we discard the
5306 latter.
5307
5308 Second, GNAT emits a type following a specified encoding for each renaming
5309 entity. Unfortunately, STABS currently does not support the definition
5310 of types that are local to a given lexical block, so all renamings types
5311 are emitted at library level. As a consequence, if an application
5312 contains two renaming entities using the same name, and a user tries to
5313 print the value of one of these entities, the result of the ada symbol
5314 lookup will also contain the wrong renaming type.
5315
5316 This function partially covers for this limitation by attempting to
5317 remove from the SYMS list renaming symbols that should be visible
5318 from CURRENT_BLOCK. However, there does not seem be a 100% reliable
5319 method with the current information available. The implementation
5320 below has a couple of limitations (FIXME: brobecker-2003-05-12):
5321
5322 - When the user tries to print a rename in a function while there
5323 is another rename entity defined in a package: Normally, the
5324 rename in the function has precedence over the rename in the
5325 package, so the latter should be removed from the list. This is
5326 currently not the case.
5327
5328 - This function will incorrectly remove valid renames if
5329 the CURRENT_BLOCK corresponds to a function which symbol name
5330 has been changed by an "Export" pragma. As a consequence,
5331 the user will be unable to print such rename entities. */
5332
5333 static int
5334 remove_irrelevant_renamings (struct block_symbol *syms,
5335 int nsyms, const struct block *current_block)
5336 {
5337 struct symbol *current_function;
5338 const char *current_function_name;
5339 int i;
5340 int is_new_style_renaming;
5341
5342 /* If there is both a renaming foo___XR... encoded as a variable and
5343 a simple variable foo in the same block, discard the latter.
5344 First, zero out such symbols, then compress. */
5345 is_new_style_renaming = 0;
5346 for (i = 0; i < nsyms; i += 1)
5347 {
5348 struct symbol *sym = syms[i].symbol;
5349 const struct block *block = syms[i].block;
5350 const char *name;
5351 const char *suffix;
5352
5353 if (sym == NULL || SYMBOL_CLASS (sym) == LOC_TYPEDEF)
5354 continue;
5355 name = SYMBOL_LINKAGE_NAME (sym);
5356 suffix = strstr (name, "___XR");
5357
5358 if (suffix != NULL)
5359 {
5360 int name_len = suffix - name;
5361 int j;
5362
5363 is_new_style_renaming = 1;
5364 for (j = 0; j < nsyms; j += 1)
5365 if (i != j && syms[j].symbol != NULL
5366 && strncmp (name, SYMBOL_LINKAGE_NAME (syms[j].symbol),
5367 name_len) == 0
5368 && block == syms[j].block)
5369 syms[j].symbol = NULL;
5370 }
5371 }
5372 if (is_new_style_renaming)
5373 {
5374 int j, k;
5375
5376 for (j = k = 0; j < nsyms; j += 1)
5377 if (syms[j].symbol != NULL)
5378 {
5379 syms[k] = syms[j];
5380 k += 1;
5381 }
5382 return k;
5383 }
5384
5385 /* Extract the function name associated to CURRENT_BLOCK.
5386 Abort if unable to do so. */
5387
5388 if (current_block == NULL)
5389 return nsyms;
5390
5391 current_function = block_linkage_function (current_block);
5392 if (current_function == NULL)
5393 return nsyms;
5394
5395 current_function_name = SYMBOL_LINKAGE_NAME (current_function);
5396 if (current_function_name == NULL)
5397 return nsyms;
5398
5399 /* Check each of the symbols, and remove it from the list if it is
5400 a type corresponding to a renaming that is out of the scope of
5401 the current block. */
5402
5403 i = 0;
5404 while (i < nsyms)
5405 {
5406 if (ada_parse_renaming (syms[i].symbol, NULL, NULL, NULL)
5407 == ADA_OBJECT_RENAMING
5408 && old_renaming_is_invisible (syms[i].symbol, current_function_name))
5409 {
5410 int j;
5411
5412 for (j = i + 1; j < nsyms; j += 1)
5413 syms[j - 1] = syms[j];
5414 nsyms -= 1;
5415 }
5416 else
5417 i += 1;
5418 }
5419
5420 return nsyms;
5421 }
5422
5423 /* Add to OBSTACKP all symbols from BLOCK (and its super-blocks)
5424 whose name and domain match NAME and DOMAIN respectively.
5425 If no match was found, then extend the search to "enclosing"
5426 routines (in other words, if we're inside a nested function,
5427 search the symbols defined inside the enclosing functions).
5428 If WILD_MATCH_P is nonzero, perform the naming matching in
5429 "wild" mode (see function "wild_match" for more info).
5430
5431 Note: This function assumes that OBSTACKP has 0 (zero) element in it. */
5432
5433 static void
5434 ada_add_local_symbols (struct obstack *obstackp,
5435 const lookup_name_info &lookup_name,
5436 const struct block *block, domain_enum domain)
5437 {
5438 int block_depth = 0;
5439
5440 while (block != NULL)
5441 {
5442 block_depth += 1;
5443 ada_add_block_symbols (obstackp, block, lookup_name, domain, NULL);
5444
5445 /* If we found a non-function match, assume that's the one. */
5446 if (is_nonfunction (defns_collected (obstackp, 0),
5447 num_defns_collected (obstackp)))
5448 return;
5449
5450 block = BLOCK_SUPERBLOCK (block);
5451 }
5452
5453 /* If no luck so far, try to find NAME as a local symbol in some lexically
5454 enclosing subprogram. */
5455 if (num_defns_collected (obstackp) == 0 && block_depth > 2)
5456 add_symbols_from_enclosing_procs (obstackp, lookup_name, domain);
5457 }
5458
5459 /* An object of this type is used as the user_data argument when
5460 calling the map_matching_symbols method. */
5461
5462 struct match_data
5463 {
5464 struct objfile *objfile;
5465 struct obstack *obstackp;
5466 struct symbol *arg_sym;
5467 int found_sym;
5468 };
5469
5470 /* A callback for add_nonlocal_symbols that adds SYM, found in BLOCK,
5471 to a list of symbols. DATA0 is a pointer to a struct match_data *
5472 containing the obstack that collects the symbol list, the file that SYM
5473 must come from, a flag indicating whether a non-argument symbol has
5474 been found in the current block, and the last argument symbol
5475 passed in SYM within the current block (if any). When SYM is null,
5476 marking the end of a block, the argument symbol is added if no
5477 other has been found. */
5478
5479 static int
5480 aux_add_nonlocal_symbols (struct block *block, struct symbol *sym, void *data0)
5481 {
5482 struct match_data *data = (struct match_data *) data0;
5483
5484 if (sym == NULL)
5485 {
5486 if (!data->found_sym && data->arg_sym != NULL)
5487 add_defn_to_vec (data->obstackp,
5488 fixup_symbol_section (data->arg_sym, data->objfile),
5489 block);
5490 data->found_sym = 0;
5491 data->arg_sym = NULL;
5492 }
5493 else
5494 {
5495 if (SYMBOL_CLASS (sym) == LOC_UNRESOLVED)
5496 return 0;
5497 else if (SYMBOL_IS_ARGUMENT (sym))
5498 data->arg_sym = sym;
5499 else
5500 {
5501 data->found_sym = 1;
5502 add_defn_to_vec (data->obstackp,
5503 fixup_symbol_section (sym, data->objfile),
5504 block);
5505 }
5506 }
5507 return 0;
5508 }
5509
5510 /* Helper for add_nonlocal_symbols. Find symbols in DOMAIN which are
5511 targeted by renamings matching LOOKUP_NAME in BLOCK. Add these
5512 symbols to OBSTACKP. Return whether we found such symbols. */
5513
5514 static int
5515 ada_add_block_renamings (struct obstack *obstackp,
5516 const struct block *block,
5517 const lookup_name_info &lookup_name,
5518 domain_enum domain)
5519 {
5520 struct using_direct *renaming;
5521 int defns_mark = num_defns_collected (obstackp);
5522
5523 symbol_name_matcher_ftype *name_match
5524 = ada_get_symbol_name_matcher (lookup_name);
5525
5526 for (renaming = block_using (block);
5527 renaming != NULL;
5528 renaming = renaming->next)
5529 {
5530 const char *r_name;
5531
5532 /* Avoid infinite recursions: skip this renaming if we are actually
5533 already traversing it.
5534
5535 Currently, symbol lookup in Ada don't use the namespace machinery from
5536 C++/Fortran support: skip namespace imports that use them. */
5537 if (renaming->searched
5538 || (renaming->import_src != NULL
5539 && renaming->import_src[0] != '\0')
5540 || (renaming->import_dest != NULL
5541 && renaming->import_dest[0] != '\0'))
5542 continue;
5543 renaming->searched = 1;
5544
5545 /* TODO: here, we perform another name-based symbol lookup, which can
5546 pull its own multiple overloads. In theory, we should be able to do
5547 better in this case since, in DWARF, DW_AT_import is a DIE reference,
5548 not a simple name. But in order to do this, we would need to enhance
5549 the DWARF reader to associate a symbol to this renaming, instead of a
5550 name. So, for now, we do something simpler: re-use the C++/Fortran
5551 namespace machinery. */
5552 r_name = (renaming->alias != NULL
5553 ? renaming->alias
5554 : renaming->declaration);
5555 if (name_match (r_name, lookup_name, NULL))
5556 {
5557 lookup_name_info decl_lookup_name (renaming->declaration,
5558 lookup_name.match_type ());
5559 ada_add_all_symbols (obstackp, block, decl_lookup_name, domain,
5560 1, NULL);
5561 }
5562 renaming->searched = 0;
5563 }
5564 return num_defns_collected (obstackp) != defns_mark;
5565 }
5566
5567 /* Implements compare_names, but only applying the comparision using
5568 the given CASING. */
5569
5570 static int
5571 compare_names_with_case (const char *string1, const char *string2,
5572 enum case_sensitivity casing)
5573 {
5574 while (*string1 != '\0' && *string2 != '\0')
5575 {
5576 char c1, c2;
5577
5578 if (isspace (*string1) || isspace (*string2))
5579 return strcmp_iw_ordered (string1, string2);
5580
5581 if (casing == case_sensitive_off)
5582 {
5583 c1 = tolower (*string1);
5584 c2 = tolower (*string2);
5585 }
5586 else
5587 {
5588 c1 = *string1;
5589 c2 = *string2;
5590 }
5591 if (c1 != c2)
5592 break;
5593
5594 string1 += 1;
5595 string2 += 1;
5596 }
5597
5598 switch (*string1)
5599 {
5600 case '(':
5601 return strcmp_iw_ordered (string1, string2);
5602 case '_':
5603 if (*string2 == '\0')
5604 {
5605 if (is_name_suffix (string1))
5606 return 0;
5607 else
5608 return 1;
5609 }
5610 /* FALLTHROUGH */
5611 default:
5612 if (*string2 == '(')
5613 return strcmp_iw_ordered (string1, string2);
5614 else
5615 {
5616 if (casing == case_sensitive_off)
5617 return tolower (*string1) - tolower (*string2);
5618 else
5619 return *string1 - *string2;
5620 }
5621 }
5622 }
5623
5624 /* Compare STRING1 to STRING2, with results as for strcmp.
5625 Compatible with strcmp_iw_ordered in that...
5626
5627 strcmp_iw_ordered (STRING1, STRING2) <= 0
5628
5629 ... implies...
5630
5631 compare_names (STRING1, STRING2) <= 0
5632
5633 (they may differ as to what symbols compare equal). */
5634
5635 static int
5636 compare_names (const char *string1, const char *string2)
5637 {
5638 int result;
5639
5640 /* Similar to what strcmp_iw_ordered does, we need to perform
5641 a case-insensitive comparison first, and only resort to
5642 a second, case-sensitive, comparison if the first one was
5643 not sufficient to differentiate the two strings. */
5644
5645 result = compare_names_with_case (string1, string2, case_sensitive_off);
5646 if (result == 0)
5647 result = compare_names_with_case (string1, string2, case_sensitive_on);
5648
5649 return result;
5650 }
5651
5652 /* Convenience function to get at the Ada encoded lookup name for
5653 LOOKUP_NAME, as a C string. */
5654
5655 static const char *
5656 ada_lookup_name (const lookup_name_info &lookup_name)
5657 {
5658 return lookup_name.ada ().lookup_name ().c_str ();
5659 }
5660
5661 /* Add to OBSTACKP all non-local symbols whose name and domain match
5662 LOOKUP_NAME and DOMAIN respectively. The search is performed on
5663 GLOBAL_BLOCK symbols if GLOBAL is non-zero, or on STATIC_BLOCK
5664 symbols otherwise. */
5665
5666 static void
5667 add_nonlocal_symbols (struct obstack *obstackp,
5668 const lookup_name_info &lookup_name,
5669 domain_enum domain, int global)
5670 {
5671 struct objfile *objfile;
5672 struct compunit_symtab *cu;
5673 struct match_data data;
5674
5675 memset (&data, 0, sizeof data);
5676 data.obstackp = obstackp;
5677
5678 bool is_wild_match = lookup_name.ada ().wild_match_p ();
5679
5680 ALL_OBJFILES (objfile)
5681 {
5682 data.objfile = objfile;
5683
5684 if (is_wild_match)
5685 objfile->sf->qf->map_matching_symbols (objfile, lookup_name.name ().c_str (),
5686 domain, global,
5687 aux_add_nonlocal_symbols, &data,
5688 symbol_name_match_type::WILD,
5689 NULL);
5690 else
5691 objfile->sf->qf->map_matching_symbols (objfile, lookup_name.name ().c_str (),
5692 domain, global,
5693 aux_add_nonlocal_symbols, &data,
5694 symbol_name_match_type::FULL,
5695 compare_names);
5696
5697 ALL_OBJFILE_COMPUNITS (objfile, cu)
5698 {
5699 const struct block *global_block
5700 = BLOCKVECTOR_BLOCK (COMPUNIT_BLOCKVECTOR (cu), GLOBAL_BLOCK);
5701
5702 if (ada_add_block_renamings (obstackp, global_block, lookup_name,
5703 domain))
5704 data.found_sym = 1;
5705 }
5706 }
5707
5708 if (num_defns_collected (obstackp) == 0 && global && !is_wild_match)
5709 {
5710 const char *name = ada_lookup_name (lookup_name);
5711 std::string name1 = std::string ("<_ada_") + name + '>';
5712
5713 ALL_OBJFILES (objfile)
5714 {
5715 data.objfile = objfile;
5716 objfile->sf->qf->map_matching_symbols (objfile, name1.c_str (),
5717 domain, global,
5718 aux_add_nonlocal_symbols,
5719 &data,
5720 symbol_name_match_type::FULL,
5721 compare_names);
5722 }
5723 }
5724 }
5725
5726 /* Find symbols in DOMAIN matching LOOKUP_NAME, in BLOCK and, if
5727 FULL_SEARCH is non-zero, enclosing scope and in global scopes,
5728 returning the number of matches. Add these to OBSTACKP.
5729
5730 When FULL_SEARCH is non-zero, any non-function/non-enumeral
5731 symbol match within the nest of blocks whose innermost member is BLOCK,
5732 is the one match returned (no other matches in that or
5733 enclosing blocks is returned). If there are any matches in or
5734 surrounding BLOCK, then these alone are returned.
5735
5736 Names prefixed with "standard__" are handled specially:
5737 "standard__" is first stripped off (by the lookup_name
5738 constructor), and only static and global symbols are searched.
5739
5740 If MADE_GLOBAL_LOOKUP_P is non-null, set it before return to whether we had
5741 to lookup global symbols. */
5742
5743 static void
5744 ada_add_all_symbols (struct obstack *obstackp,
5745 const struct block *block,
5746 const lookup_name_info &lookup_name,
5747 domain_enum domain,
5748 int full_search,
5749 int *made_global_lookup_p)
5750 {
5751 struct symbol *sym;
5752
5753 if (made_global_lookup_p)
5754 *made_global_lookup_p = 0;
5755
5756 /* Special case: If the user specifies a symbol name inside package
5757 Standard, do a non-wild matching of the symbol name without
5758 the "standard__" prefix. This was primarily introduced in order
5759 to allow the user to specifically access the standard exceptions
5760 using, for instance, Standard.Constraint_Error when Constraint_Error
5761 is ambiguous (due to the user defining its own Constraint_Error
5762 entity inside its program). */
5763 if (lookup_name.ada ().standard_p ())
5764 block = NULL;
5765
5766 /* Check the non-global symbols. If we have ANY match, then we're done. */
5767
5768 if (block != NULL)
5769 {
5770 if (full_search)
5771 ada_add_local_symbols (obstackp, lookup_name, block, domain);
5772 else
5773 {
5774 /* In the !full_search case we're are being called by
5775 ada_iterate_over_symbols, and we don't want to search
5776 superblocks. */
5777 ada_add_block_symbols (obstackp, block, lookup_name, domain, NULL);
5778 }
5779 if (num_defns_collected (obstackp) > 0 || !full_search)
5780 return;
5781 }
5782
5783 /* No non-global symbols found. Check our cache to see if we have
5784 already performed this search before. If we have, then return
5785 the same result. */
5786
5787 if (lookup_cached_symbol (ada_lookup_name (lookup_name),
5788 domain, &sym, &block))
5789 {
5790 if (sym != NULL)
5791 add_defn_to_vec (obstackp, sym, block);
5792 return;
5793 }
5794
5795 if (made_global_lookup_p)
5796 *made_global_lookup_p = 1;
5797
5798 /* Search symbols from all global blocks. */
5799
5800 add_nonlocal_symbols (obstackp, lookup_name, domain, 1);
5801
5802 /* Now add symbols from all per-file blocks if we've gotten no hits
5803 (not strictly correct, but perhaps better than an error). */
5804
5805 if (num_defns_collected (obstackp) == 0)
5806 add_nonlocal_symbols (obstackp, lookup_name, domain, 0);
5807 }
5808
5809 /* Find symbols in DOMAIN matching LOOKUP_NAME, in BLOCK and, if FULL_SEARCH
5810 is non-zero, enclosing scope and in global scopes, returning the number of
5811 matches.
5812 Sets *RESULTS to point to a newly allocated vector of (SYM,BLOCK) tuples,
5813 indicating the symbols found and the blocks and symbol tables (if
5814 any) in which they were found. This vector should be freed when
5815 no longer useful.
5816
5817 When full_search is non-zero, any non-function/non-enumeral
5818 symbol match within the nest of blocks whose innermost member is BLOCK,
5819 is the one match returned (no other matches in that or
5820 enclosing blocks is returned). If there are any matches in or
5821 surrounding BLOCK, then these alone are returned.
5822
5823 Names prefixed with "standard__" are handled specially: "standard__"
5824 is first stripped off, and only static and global symbols are searched. */
5825
5826 static int
5827 ada_lookup_symbol_list_worker (const lookup_name_info &lookup_name,
5828 const struct block *block,
5829 domain_enum domain,
5830 struct block_symbol **results,
5831 int full_search)
5832 {
5833 int syms_from_global_search;
5834 int ndefns;
5835 int results_size;
5836 auto_obstack obstack;
5837
5838 ada_add_all_symbols (&obstack, block, lookup_name,
5839 domain, full_search, &syms_from_global_search);
5840
5841 ndefns = num_defns_collected (&obstack);
5842
5843 results_size = obstack_object_size (&obstack);
5844 *results = (struct block_symbol *) malloc (results_size);
5845 memcpy (*results, defns_collected (&obstack, 1), results_size);
5846
5847 ndefns = remove_extra_symbols (*results, ndefns);
5848
5849 if (ndefns == 0 && full_search && syms_from_global_search)
5850 cache_symbol (ada_lookup_name (lookup_name), domain, NULL, NULL);
5851
5852 if (ndefns == 1 && full_search && syms_from_global_search)
5853 cache_symbol (ada_lookup_name (lookup_name), domain,
5854 (*results)[0].symbol, (*results)[0].block);
5855
5856 ndefns = remove_irrelevant_renamings (*results, ndefns, block);
5857
5858 return ndefns;
5859 }
5860
5861 /* Find symbols in DOMAIN matching NAME, in BLOCK and enclosing scope and
5862 in global scopes, returning the number of matches, and setting *RESULTS
5863 to a newly-allocated vector of (SYM,BLOCK) tuples. This newly-allocated
5864 vector should be freed when no longer useful.
5865
5866 See ada_lookup_symbol_list_worker for further details. */
5867
5868 int
5869 ada_lookup_symbol_list (const char *name, const struct block *block,
5870 domain_enum domain, struct block_symbol **results)
5871 {
5872 symbol_name_match_type name_match_type = name_match_type_from_name (name);
5873 lookup_name_info lookup_name (name, name_match_type);
5874
5875 return ada_lookup_symbol_list_worker (lookup_name, block, domain, results, 1);
5876 }
5877
5878 /* Implementation of the la_iterate_over_symbols method. */
5879
5880 static void
5881 ada_iterate_over_symbols
5882 (const struct block *block, const lookup_name_info &name,
5883 domain_enum domain,
5884 gdb::function_view<symbol_found_callback_ftype> callback)
5885 {
5886 int ndefs, i;
5887 struct block_symbol *results;
5888 struct cleanup *old_chain;
5889
5890 ndefs = ada_lookup_symbol_list_worker (name, block, domain, &results, 0);
5891 old_chain = make_cleanup (xfree, results);
5892
5893 for (i = 0; i < ndefs; ++i)
5894 {
5895 if (!callback (results[i].symbol))
5896 break;
5897 }
5898
5899 do_cleanups (old_chain);
5900 }
5901
5902 /* The result is as for ada_lookup_symbol_list with FULL_SEARCH set
5903 to 1, but choosing the first symbol found if there are multiple
5904 choices.
5905
5906 The result is stored in *INFO, which must be non-NULL.
5907 If no match is found, INFO->SYM is set to NULL. */
5908
5909 void
5910 ada_lookup_encoded_symbol (const char *name, const struct block *block,
5911 domain_enum domain,
5912 struct block_symbol *info)
5913 {
5914 struct block_symbol *candidates;
5915 int n_candidates;
5916 struct cleanup *old_chain;
5917
5918 /* Since we already have an encoded name, wrap it in '<>' to force a
5919 verbatim match. Otherwise, if the name happens to not look like
5920 an encoded name (because it doesn't include a "__"),
5921 ada_lookup_name_info would re-encode/fold it again, and that
5922 would e.g., incorrectly lowercase object renaming names like
5923 "R28b" -> "r28b". */
5924 std::string verbatim = std::string ("<") + name + '>';
5925
5926 gdb_assert (info != NULL);
5927 memset (info, 0, sizeof (struct block_symbol));
5928
5929 n_candidates = ada_lookup_symbol_list (verbatim.c_str (), block,
5930 domain, &candidates);
5931 old_chain = make_cleanup (xfree, candidates);
5932
5933 if (n_candidates == 0)
5934 {
5935 do_cleanups (old_chain);
5936 return;
5937 }
5938
5939 *info = candidates[0];
5940 info->symbol = fixup_symbol_section (info->symbol, NULL);
5941
5942 do_cleanups (old_chain);
5943 }
5944
5945 /* Return a symbol in DOMAIN matching NAME, in BLOCK0 and enclosing
5946 scope and in global scopes, or NULL if none. NAME is folded and
5947 encoded first. Otherwise, the result is as for ada_lookup_symbol_list,
5948 choosing the first symbol if there are multiple choices.
5949 If IS_A_FIELD_OF_THIS is not NULL, it is set to zero. */
5950
5951 struct block_symbol
5952 ada_lookup_symbol (const char *name, const struct block *block0,
5953 domain_enum domain, int *is_a_field_of_this)
5954 {
5955 struct block_symbol info;
5956
5957 if (is_a_field_of_this != NULL)
5958 *is_a_field_of_this = 0;
5959
5960 ada_lookup_encoded_symbol (ada_encode (ada_fold_name (name)),
5961 block0, domain, &info);
5962 return info;
5963 }
5964
5965 static struct block_symbol
5966 ada_lookup_symbol_nonlocal (const struct language_defn *langdef,
5967 const char *name,
5968 const struct block *block,
5969 const domain_enum domain)
5970 {
5971 struct block_symbol sym;
5972
5973 sym = ada_lookup_symbol (name, block_static_block (block), domain, NULL);
5974 if (sym.symbol != NULL)
5975 return sym;
5976
5977 /* If we haven't found a match at this point, try the primitive
5978 types. In other languages, this search is performed before
5979 searching for global symbols in order to short-circuit that
5980 global-symbol search if it happens that the name corresponds
5981 to a primitive type. But we cannot do the same in Ada, because
5982 it is perfectly legitimate for a program to declare a type which
5983 has the same name as a standard type. If looking up a type in
5984 that situation, we have traditionally ignored the primitive type
5985 in favor of user-defined types. This is why, unlike most other
5986 languages, we search the primitive types this late and only after
5987 having searched the global symbols without success. */
5988
5989 if (domain == VAR_DOMAIN)
5990 {
5991 struct gdbarch *gdbarch;
5992
5993 if (block == NULL)
5994 gdbarch = target_gdbarch ();
5995 else
5996 gdbarch = block_gdbarch (block);
5997 sym.symbol = language_lookup_primitive_type_as_symbol (langdef, gdbarch, name);
5998 if (sym.symbol != NULL)
5999 return sym;
6000 }
6001
6002 return (struct block_symbol) {NULL, NULL};
6003 }
6004
6005
6006 /* True iff STR is a possible encoded suffix of a normal Ada name
6007 that is to be ignored for matching purposes. Suffixes of parallel
6008 names (e.g., XVE) are not included here. Currently, the possible suffixes
6009 are given by any of the regular expressions:
6010
6011 [.$][0-9]+ [nested subprogram suffix, on platforms such as GNU/Linux]
6012 ___[0-9]+ [nested subprogram suffix, on platforms such as HP/UX]
6013 TKB [subprogram suffix for task bodies]
6014 _E[0-9]+[bs]$ [protected object entry suffixes]
6015 (X[nb]*)?((\$|__)[0-9](_?[0-9]+)|___(JM|LJM|X([FDBUP].*|R[^T]?)))?$
6016
6017 Also, any leading "__[0-9]+" sequence is skipped before the suffix
6018 match is performed. This sequence is used to differentiate homonyms,
6019 is an optional part of a valid name suffix. */
6020
6021 static int
6022 is_name_suffix (const char *str)
6023 {
6024 int k;
6025 const char *matching;
6026 const int len = strlen (str);
6027
6028 /* Skip optional leading __[0-9]+. */
6029
6030 if (len > 3 && str[0] == '_' && str[1] == '_' && isdigit (str[2]))
6031 {
6032 str += 3;
6033 while (isdigit (str[0]))
6034 str += 1;
6035 }
6036
6037 /* [.$][0-9]+ */
6038
6039 if (str[0] == '.' || str[0] == '$')
6040 {
6041 matching = str + 1;
6042 while (isdigit (matching[0]))
6043 matching += 1;
6044 if (matching[0] == '\0')
6045 return 1;
6046 }
6047
6048 /* ___[0-9]+ */
6049
6050 if (len > 3 && str[0] == '_' && str[1] == '_' && str[2] == '_')
6051 {
6052 matching = str + 3;
6053 while (isdigit (matching[0]))
6054 matching += 1;
6055 if (matching[0] == '\0')
6056 return 1;
6057 }
6058
6059 /* "TKB" suffixes are used for subprograms implementing task bodies. */
6060
6061 if (strcmp (str, "TKB") == 0)
6062 return 1;
6063
6064 #if 0
6065 /* FIXME: brobecker/2005-09-23: Protected Object subprograms end
6066 with a N at the end. Unfortunately, the compiler uses the same
6067 convention for other internal types it creates. So treating
6068 all entity names that end with an "N" as a name suffix causes
6069 some regressions. For instance, consider the case of an enumerated
6070 type. To support the 'Image attribute, it creates an array whose
6071 name ends with N.
6072 Having a single character like this as a suffix carrying some
6073 information is a bit risky. Perhaps we should change the encoding
6074 to be something like "_N" instead. In the meantime, do not do
6075 the following check. */
6076 /* Protected Object Subprograms */
6077 if (len == 1 && str [0] == 'N')
6078 return 1;
6079 #endif
6080
6081 /* _E[0-9]+[bs]$ */
6082 if (len > 3 && str[0] == '_' && str [1] == 'E' && isdigit (str[2]))
6083 {
6084 matching = str + 3;
6085 while (isdigit (matching[0]))
6086 matching += 1;
6087 if ((matching[0] == 'b' || matching[0] == 's')
6088 && matching [1] == '\0')
6089 return 1;
6090 }
6091
6092 /* ??? We should not modify STR directly, as we are doing below. This
6093 is fine in this case, but may become problematic later if we find
6094 that this alternative did not work, and want to try matching
6095 another one from the begining of STR. Since we modified it, we
6096 won't be able to find the begining of the string anymore! */
6097 if (str[0] == 'X')
6098 {
6099 str += 1;
6100 while (str[0] != '_' && str[0] != '\0')
6101 {
6102 if (str[0] != 'n' && str[0] != 'b')
6103 return 0;
6104 str += 1;
6105 }
6106 }
6107
6108 if (str[0] == '\000')
6109 return 1;
6110
6111 if (str[0] == '_')
6112 {
6113 if (str[1] != '_' || str[2] == '\000')
6114 return 0;
6115 if (str[2] == '_')
6116 {
6117 if (strcmp (str + 3, "JM") == 0)
6118 return 1;
6119 /* FIXME: brobecker/2004-09-30: GNAT will soon stop using
6120 the LJM suffix in favor of the JM one. But we will
6121 still accept LJM as a valid suffix for a reasonable
6122 amount of time, just to allow ourselves to debug programs
6123 compiled using an older version of GNAT. */
6124 if (strcmp (str + 3, "LJM") == 0)
6125 return 1;
6126 if (str[3] != 'X')
6127 return 0;
6128 if (str[4] == 'F' || str[4] == 'D' || str[4] == 'B'
6129 || str[4] == 'U' || str[4] == 'P')
6130 return 1;
6131 if (str[4] == 'R' && str[5] != 'T')
6132 return 1;
6133 return 0;
6134 }
6135 if (!isdigit (str[2]))
6136 return 0;
6137 for (k = 3; str[k] != '\0'; k += 1)
6138 if (!isdigit (str[k]) && str[k] != '_')
6139 return 0;
6140 return 1;
6141 }
6142 if (str[0] == '$' && isdigit (str[1]))
6143 {
6144 for (k = 2; str[k] != '\0'; k += 1)
6145 if (!isdigit (str[k]) && str[k] != '_')
6146 return 0;
6147 return 1;
6148 }
6149 return 0;
6150 }
6151
6152 /* Return non-zero if the string starting at NAME and ending before
6153 NAME_END contains no capital letters. */
6154
6155 static int
6156 is_valid_name_for_wild_match (const char *name0)
6157 {
6158 const char *decoded_name = ada_decode (name0);
6159 int i;
6160
6161 /* If the decoded name starts with an angle bracket, it means that
6162 NAME0 does not follow the GNAT encoding format. It should then
6163 not be allowed as a possible wild match. */
6164 if (decoded_name[0] == '<')
6165 return 0;
6166
6167 for (i=0; decoded_name[i] != '\0'; i++)
6168 if (isalpha (decoded_name[i]) && !islower (decoded_name[i]))
6169 return 0;
6170
6171 return 1;
6172 }
6173
6174 /* Advance *NAMEP to next occurrence of TARGET0 in the string NAME0
6175 that could start a simple name. Assumes that *NAMEP points into
6176 the string beginning at NAME0. */
6177
6178 static int
6179 advance_wild_match (const char **namep, const char *name0, int target0)
6180 {
6181 const char *name = *namep;
6182
6183 while (1)
6184 {
6185 int t0, t1;
6186
6187 t0 = *name;
6188 if (t0 == '_')
6189 {
6190 t1 = name[1];
6191 if ((t1 >= 'a' && t1 <= 'z') || (t1 >= '0' && t1 <= '9'))
6192 {
6193 name += 1;
6194 if (name == name0 + 5 && startswith (name0, "_ada"))
6195 break;
6196 else
6197 name += 1;
6198 }
6199 else if (t1 == '_' && ((name[2] >= 'a' && name[2] <= 'z')
6200 || name[2] == target0))
6201 {
6202 name += 2;
6203 break;
6204 }
6205 else
6206 return 0;
6207 }
6208 else if ((t0 >= 'a' && t0 <= 'z') || (t0 >= '0' && t0 <= '9'))
6209 name += 1;
6210 else
6211 return 0;
6212 }
6213
6214 *namep = name;
6215 return 1;
6216 }
6217
6218 /* Return true iff NAME encodes a name of the form prefix.PATN.
6219 Ignores any informational suffixes of NAME (i.e., for which
6220 is_name_suffix is true). Assumes that PATN is a lower-cased Ada
6221 simple name. */
6222
6223 static bool
6224 wild_match (const char *name, const char *patn)
6225 {
6226 const char *p;
6227 const char *name0 = name;
6228
6229 while (1)
6230 {
6231 const char *match = name;
6232
6233 if (*name == *patn)
6234 {
6235 for (name += 1, p = patn + 1; *p != '\0'; name += 1, p += 1)
6236 if (*p != *name)
6237 break;
6238 if (*p == '\0' && is_name_suffix (name))
6239 return match == name0 || is_valid_name_for_wild_match (name0);
6240
6241 if (name[-1] == '_')
6242 name -= 1;
6243 }
6244 if (!advance_wild_match (&name, name0, *patn))
6245 return false;
6246 }
6247 }
6248
6249 /* Returns true iff symbol name SYM_NAME matches SEARCH_NAME, ignoring
6250 any trailing suffixes that encode debugging information or leading
6251 _ada_ on SYM_NAME (see is_name_suffix commentary for the debugging
6252 information that is ignored). */
6253
6254 static bool
6255 full_match (const char *sym_name, const char *search_name)
6256 {
6257 size_t search_name_len = strlen (search_name);
6258
6259 if (strncmp (sym_name, search_name, search_name_len) == 0
6260 && is_name_suffix (sym_name + search_name_len))
6261 return true;
6262
6263 if (startswith (sym_name, "_ada_")
6264 && strncmp (sym_name + 5, search_name, search_name_len) == 0
6265 && is_name_suffix (sym_name + search_name_len + 5))
6266 return true;
6267
6268 return false;
6269 }
6270
6271 /* Add symbols from BLOCK matching LOOKUP_NAME in DOMAIN to vector
6272 *defn_symbols, updating the list of symbols in OBSTACKP (if
6273 necessary). OBJFILE is the section containing BLOCK. */
6274
6275 static void
6276 ada_add_block_symbols (struct obstack *obstackp,
6277 const struct block *block,
6278 const lookup_name_info &lookup_name,
6279 domain_enum domain, struct objfile *objfile)
6280 {
6281 struct block_iterator iter;
6282 /* A matching argument symbol, if any. */
6283 struct symbol *arg_sym;
6284 /* Set true when we find a matching non-argument symbol. */
6285 int found_sym;
6286 struct symbol *sym;
6287
6288 arg_sym = NULL;
6289 found_sym = 0;
6290 for (sym = block_iter_match_first (block, lookup_name, &iter);
6291 sym != NULL;
6292 sym = block_iter_match_next (lookup_name, &iter))
6293 {
6294 if (symbol_matches_domain (SYMBOL_LANGUAGE (sym),
6295 SYMBOL_DOMAIN (sym), domain))
6296 {
6297 if (SYMBOL_CLASS (sym) != LOC_UNRESOLVED)
6298 {
6299 if (SYMBOL_IS_ARGUMENT (sym))
6300 arg_sym = sym;
6301 else
6302 {
6303 found_sym = 1;
6304 add_defn_to_vec (obstackp,
6305 fixup_symbol_section (sym, objfile),
6306 block);
6307 }
6308 }
6309 }
6310 }
6311
6312 /* Handle renamings. */
6313
6314 if (ada_add_block_renamings (obstackp, block, lookup_name, domain))
6315 found_sym = 1;
6316
6317 if (!found_sym && arg_sym != NULL)
6318 {
6319 add_defn_to_vec (obstackp,
6320 fixup_symbol_section (arg_sym, objfile),
6321 block);
6322 }
6323
6324 if (!lookup_name.ada ().wild_match_p ())
6325 {
6326 arg_sym = NULL;
6327 found_sym = 0;
6328 const std::string &ada_lookup_name = lookup_name.ada ().lookup_name ();
6329 const char *name = ada_lookup_name.c_str ();
6330 size_t name_len = ada_lookup_name.size ();
6331
6332 ALL_BLOCK_SYMBOLS (block, iter, sym)
6333 {
6334 if (symbol_matches_domain (SYMBOL_LANGUAGE (sym),
6335 SYMBOL_DOMAIN (sym), domain))
6336 {
6337 int cmp;
6338
6339 cmp = (int) '_' - (int) SYMBOL_LINKAGE_NAME (sym)[0];
6340 if (cmp == 0)
6341 {
6342 cmp = !startswith (SYMBOL_LINKAGE_NAME (sym), "_ada_");
6343 if (cmp == 0)
6344 cmp = strncmp (name, SYMBOL_LINKAGE_NAME (sym) + 5,
6345 name_len);
6346 }
6347
6348 if (cmp == 0
6349 && is_name_suffix (SYMBOL_LINKAGE_NAME (sym) + name_len + 5))
6350 {
6351 if (SYMBOL_CLASS (sym) != LOC_UNRESOLVED)
6352 {
6353 if (SYMBOL_IS_ARGUMENT (sym))
6354 arg_sym = sym;
6355 else
6356 {
6357 found_sym = 1;
6358 add_defn_to_vec (obstackp,
6359 fixup_symbol_section (sym, objfile),
6360 block);
6361 }
6362 }
6363 }
6364 }
6365 }
6366
6367 /* NOTE: This really shouldn't be needed for _ada_ symbols.
6368 They aren't parameters, right? */
6369 if (!found_sym && arg_sym != NULL)
6370 {
6371 add_defn_to_vec (obstackp,
6372 fixup_symbol_section (arg_sym, objfile),
6373 block);
6374 }
6375 }
6376 }
6377 \f
6378
6379 /* Symbol Completion */
6380
6381 /* See symtab.h. */
6382
6383 bool
6384 ada_lookup_name_info::matches
6385 (const char *sym_name,
6386 symbol_name_match_type match_type,
6387 completion_match_result *comp_match_res) const
6388 {
6389 bool match = false;
6390 const char *text = m_encoded_name.c_str ();
6391 size_t text_len = m_encoded_name.size ();
6392
6393 /* First, test against the fully qualified name of the symbol. */
6394
6395 if (strncmp (sym_name, text, text_len) == 0)
6396 match = true;
6397
6398 if (match && !m_encoded_p)
6399 {
6400 /* One needed check before declaring a positive match is to verify
6401 that iff we are doing a verbatim match, the decoded version
6402 of the symbol name starts with '<'. Otherwise, this symbol name
6403 is not a suitable completion. */
6404 const char *sym_name_copy = sym_name;
6405 bool has_angle_bracket;
6406
6407 sym_name = ada_decode (sym_name);
6408 has_angle_bracket = (sym_name[0] == '<');
6409 match = (has_angle_bracket == m_verbatim_p);
6410 sym_name = sym_name_copy;
6411 }
6412
6413 if (match && !m_verbatim_p)
6414 {
6415 /* When doing non-verbatim match, another check that needs to
6416 be done is to verify that the potentially matching symbol name
6417 does not include capital letters, because the ada-mode would
6418 not be able to understand these symbol names without the
6419 angle bracket notation. */
6420 const char *tmp;
6421
6422 for (tmp = sym_name; *tmp != '\0' && !isupper (*tmp); tmp++);
6423 if (*tmp != '\0')
6424 match = false;
6425 }
6426
6427 /* Second: Try wild matching... */
6428
6429 if (!match && m_wild_match_p)
6430 {
6431 /* Since we are doing wild matching, this means that TEXT
6432 may represent an unqualified symbol name. We therefore must
6433 also compare TEXT against the unqualified name of the symbol. */
6434 sym_name = ada_unqualified_name (ada_decode (sym_name));
6435
6436 if (strncmp (sym_name, text, text_len) == 0)
6437 match = true;
6438 }
6439
6440 /* Finally: If we found a match, prepare the result to return. */
6441
6442 if (!match)
6443 return false;
6444
6445 if (comp_match_res != NULL)
6446 {
6447 std::string &match_str = comp_match_res->match.storage ();
6448
6449 if (!m_encoded_p)
6450 match_str = ada_decode (sym_name);
6451 else
6452 {
6453 if (m_verbatim_p)
6454 match_str = add_angle_brackets (sym_name);
6455 else
6456 match_str = sym_name;
6457
6458 }
6459
6460 comp_match_res->set_match (match_str.c_str ());
6461 }
6462
6463 return true;
6464 }
6465
6466 /* Add the list of possible symbol names completing TEXT to TRACKER.
6467 WORD is the entire command on which completion is made. */
6468
6469 static void
6470 ada_collect_symbol_completion_matches (completion_tracker &tracker,
6471 complete_symbol_mode mode,
6472 symbol_name_match_type name_match_type,
6473 const char *text, const char *word,
6474 enum type_code code)
6475 {
6476 struct symbol *sym;
6477 struct compunit_symtab *s;
6478 struct minimal_symbol *msymbol;
6479 struct objfile *objfile;
6480 const struct block *b, *surrounding_static_block = 0;
6481 struct block_iterator iter;
6482 struct cleanup *old_chain = make_cleanup (null_cleanup, NULL);
6483
6484 gdb_assert (code == TYPE_CODE_UNDEF);
6485
6486 lookup_name_info lookup_name (text, name_match_type, true);
6487
6488 /* First, look at the partial symtab symbols. */
6489 expand_symtabs_matching (NULL,
6490 lookup_name,
6491 NULL,
6492 NULL,
6493 ALL_DOMAIN);
6494
6495 /* At this point scan through the misc symbol vectors and add each
6496 symbol you find to the list. Eventually we want to ignore
6497 anything that isn't a text symbol (everything else will be
6498 handled by the psymtab code above). */
6499
6500 ALL_MSYMBOLS (objfile, msymbol)
6501 {
6502 QUIT;
6503
6504 if (completion_skip_symbol (mode, msymbol))
6505 continue;
6506
6507 completion_list_add_name (tracker,
6508 MSYMBOL_LANGUAGE (msymbol),
6509 MSYMBOL_LINKAGE_NAME (msymbol),
6510 lookup_name, text, word);
6511 }
6512
6513 /* Search upwards from currently selected frame (so that we can
6514 complete on local vars. */
6515
6516 for (b = get_selected_block (0); b != NULL; b = BLOCK_SUPERBLOCK (b))
6517 {
6518 if (!BLOCK_SUPERBLOCK (b))
6519 surrounding_static_block = b; /* For elmin of dups */
6520
6521 ALL_BLOCK_SYMBOLS (b, iter, sym)
6522 {
6523 if (completion_skip_symbol (mode, sym))
6524 continue;
6525
6526 completion_list_add_name (tracker,
6527 SYMBOL_LANGUAGE (sym),
6528 SYMBOL_LINKAGE_NAME (sym),
6529 lookup_name, text, word);
6530 }
6531 }
6532
6533 /* Go through the symtabs and check the externs and statics for
6534 symbols which match. */
6535
6536 ALL_COMPUNITS (objfile, s)
6537 {
6538 QUIT;
6539 b = BLOCKVECTOR_BLOCK (COMPUNIT_BLOCKVECTOR (s), GLOBAL_BLOCK);
6540 ALL_BLOCK_SYMBOLS (b, iter, sym)
6541 {
6542 if (completion_skip_symbol (mode, sym))
6543 continue;
6544
6545 completion_list_add_name (tracker,
6546 SYMBOL_LANGUAGE (sym),
6547 SYMBOL_LINKAGE_NAME (sym),
6548 lookup_name, text, word);
6549 }
6550 }
6551
6552 ALL_COMPUNITS (objfile, s)
6553 {
6554 QUIT;
6555 b = BLOCKVECTOR_BLOCK (COMPUNIT_BLOCKVECTOR (s), STATIC_BLOCK);
6556 /* Don't do this block twice. */
6557 if (b == surrounding_static_block)
6558 continue;
6559 ALL_BLOCK_SYMBOLS (b, iter, sym)
6560 {
6561 if (completion_skip_symbol (mode, sym))
6562 continue;
6563
6564 completion_list_add_name (tracker,
6565 SYMBOL_LANGUAGE (sym),
6566 SYMBOL_LINKAGE_NAME (sym),
6567 lookup_name, text, word);
6568 }
6569 }
6570
6571 do_cleanups (old_chain);
6572 }
6573
6574 /* Field Access */
6575
6576 /* Return non-zero if TYPE is a pointer to the GNAT dispatch table used
6577 for tagged types. */
6578
6579 static int
6580 ada_is_dispatch_table_ptr_type (struct type *type)
6581 {
6582 const char *name;
6583
6584 if (TYPE_CODE (type) != TYPE_CODE_PTR)
6585 return 0;
6586
6587 name = TYPE_NAME (TYPE_TARGET_TYPE (type));
6588 if (name == NULL)
6589 return 0;
6590
6591 return (strcmp (name, "ada__tags__dispatch_table") == 0);
6592 }
6593
6594 /* Return non-zero if TYPE is an interface tag. */
6595
6596 static int
6597 ada_is_interface_tag (struct type *type)
6598 {
6599 const char *name = TYPE_NAME (type);
6600
6601 if (name == NULL)
6602 return 0;
6603
6604 return (strcmp (name, "ada__tags__interface_tag") == 0);
6605 }
6606
6607 /* True if field number FIELD_NUM in struct or union type TYPE is supposed
6608 to be invisible to users. */
6609
6610 int
6611 ada_is_ignored_field (struct type *type, int field_num)
6612 {
6613 if (field_num < 0 || field_num > TYPE_NFIELDS (type))
6614 return 1;
6615
6616 /* Check the name of that field. */
6617 {
6618 const char *name = TYPE_FIELD_NAME (type, field_num);
6619
6620 /* Anonymous field names should not be printed.
6621 brobecker/2007-02-20: I don't think this can actually happen
6622 but we don't want to print the value of annonymous fields anyway. */
6623 if (name == NULL)
6624 return 1;
6625
6626 /* Normally, fields whose name start with an underscore ("_")
6627 are fields that have been internally generated by the compiler,
6628 and thus should not be printed. The "_parent" field is special,
6629 however: This is a field internally generated by the compiler
6630 for tagged types, and it contains the components inherited from
6631 the parent type. This field should not be printed as is, but
6632 should not be ignored either. */
6633 if (name[0] == '_' && !startswith (name, "_parent"))
6634 return 1;
6635 }
6636
6637 /* If this is the dispatch table of a tagged type or an interface tag,
6638 then ignore. */
6639 if (ada_is_tagged_type (type, 1)
6640 && (ada_is_dispatch_table_ptr_type (TYPE_FIELD_TYPE (type, field_num))
6641 || ada_is_interface_tag (TYPE_FIELD_TYPE (type, field_num))))
6642 return 1;
6643
6644 /* Not a special field, so it should not be ignored. */
6645 return 0;
6646 }
6647
6648 /* True iff TYPE has a tag field. If REFOK, then TYPE may also be a
6649 pointer or reference type whose ultimate target has a tag field. */
6650
6651 int
6652 ada_is_tagged_type (struct type *type, int refok)
6653 {
6654 return (ada_lookup_struct_elt_type (type, "_tag", refok, 1) != NULL);
6655 }
6656
6657 /* True iff TYPE represents the type of X'Tag */
6658
6659 int
6660 ada_is_tag_type (struct type *type)
6661 {
6662 type = ada_check_typedef (type);
6663
6664 if (type == NULL || TYPE_CODE (type) != TYPE_CODE_PTR)
6665 return 0;
6666 else
6667 {
6668 const char *name = ada_type_name (TYPE_TARGET_TYPE (type));
6669
6670 return (name != NULL
6671 && strcmp (name, "ada__tags__dispatch_table") == 0);
6672 }
6673 }
6674
6675 /* The type of the tag on VAL. */
6676
6677 struct type *
6678 ada_tag_type (struct value *val)
6679 {
6680 return ada_lookup_struct_elt_type (value_type (val), "_tag", 1, 0);
6681 }
6682
6683 /* Return 1 if TAG follows the old scheme for Ada tags (used for Ada 95,
6684 retired at Ada 05). */
6685
6686 static int
6687 is_ada95_tag (struct value *tag)
6688 {
6689 return ada_value_struct_elt (tag, "tsd", 1) != NULL;
6690 }
6691
6692 /* The value of the tag on VAL. */
6693
6694 struct value *
6695 ada_value_tag (struct value *val)
6696 {
6697 return ada_value_struct_elt (val, "_tag", 0);
6698 }
6699
6700 /* The value of the tag on the object of type TYPE whose contents are
6701 saved at VALADDR, if it is non-null, or is at memory address
6702 ADDRESS. */
6703
6704 static struct value *
6705 value_tag_from_contents_and_address (struct type *type,
6706 const gdb_byte *valaddr,
6707 CORE_ADDR address)
6708 {
6709 int tag_byte_offset;
6710 struct type *tag_type;
6711
6712 if (find_struct_field ("_tag", type, 0, &tag_type, &tag_byte_offset,
6713 NULL, NULL, NULL))
6714 {
6715 const gdb_byte *valaddr1 = ((valaddr == NULL)
6716 ? NULL
6717 : valaddr + tag_byte_offset);
6718 CORE_ADDR address1 = (address == 0) ? 0 : address + tag_byte_offset;
6719
6720 return value_from_contents_and_address (tag_type, valaddr1, address1);
6721 }
6722 return NULL;
6723 }
6724
6725 static struct type *
6726 type_from_tag (struct value *tag)
6727 {
6728 const char *type_name = ada_tag_name (tag);
6729
6730 if (type_name != NULL)
6731 return ada_find_any_type (ada_encode (type_name));
6732 return NULL;
6733 }
6734
6735 /* Given a value OBJ of a tagged type, return a value of this
6736 type at the base address of the object. The base address, as
6737 defined in Ada.Tags, it is the address of the primary tag of
6738 the object, and therefore where the field values of its full
6739 view can be fetched. */
6740
6741 struct value *
6742 ada_tag_value_at_base_address (struct value *obj)
6743 {
6744 struct value *val;
6745 LONGEST offset_to_top = 0;
6746 struct type *ptr_type, *obj_type;
6747 struct value *tag;
6748 CORE_ADDR base_address;
6749
6750 obj_type = value_type (obj);
6751
6752 /* It is the responsability of the caller to deref pointers. */
6753
6754 if (TYPE_CODE (obj_type) == TYPE_CODE_PTR
6755 || TYPE_CODE (obj_type) == TYPE_CODE_REF)
6756 return obj;
6757
6758 tag = ada_value_tag (obj);
6759 if (!tag)
6760 return obj;
6761
6762 /* Base addresses only appeared with Ada 05 and multiple inheritance. */
6763
6764 if (is_ada95_tag (tag))
6765 return obj;
6766
6767 ptr_type = language_lookup_primitive_type
6768 (language_def (language_ada), target_gdbarch(), "storage_offset");
6769 ptr_type = lookup_pointer_type (ptr_type);
6770 val = value_cast (ptr_type, tag);
6771 if (!val)
6772 return obj;
6773
6774 /* It is perfectly possible that an exception be raised while
6775 trying to determine the base address, just like for the tag;
6776 see ada_tag_name for more details. We do not print the error
6777 message for the same reason. */
6778
6779 TRY
6780 {
6781 offset_to_top = value_as_long (value_ind (value_ptradd (val, -2)));
6782 }
6783
6784 CATCH (e, RETURN_MASK_ERROR)
6785 {
6786 return obj;
6787 }
6788 END_CATCH
6789
6790 /* If offset is null, nothing to do. */
6791
6792 if (offset_to_top == 0)
6793 return obj;
6794
6795 /* -1 is a special case in Ada.Tags; however, what should be done
6796 is not quite clear from the documentation. So do nothing for
6797 now. */
6798
6799 if (offset_to_top == -1)
6800 return obj;
6801
6802 /* OFFSET_TO_TOP used to be a positive value to be subtracted
6803 from the base address. This was however incompatible with
6804 C++ dispatch table: C++ uses a *negative* value to *add*
6805 to the base address. Ada's convention has therefore been
6806 changed in GNAT 19.0w 20171023: since then, C++ and Ada
6807 use the same convention. Here, we support both cases by
6808 checking the sign of OFFSET_TO_TOP. */
6809
6810 if (offset_to_top > 0)
6811 offset_to_top = -offset_to_top;
6812
6813 base_address = value_address (obj) + offset_to_top;
6814 tag = value_tag_from_contents_and_address (obj_type, NULL, base_address);
6815
6816 /* Make sure that we have a proper tag at the new address.
6817 Otherwise, offset_to_top is bogus (which can happen when
6818 the object is not initialized yet). */
6819
6820 if (!tag)
6821 return obj;
6822
6823 obj_type = type_from_tag (tag);
6824
6825 if (!obj_type)
6826 return obj;
6827
6828 return value_from_contents_and_address (obj_type, NULL, base_address);
6829 }
6830
6831 /* Return the "ada__tags__type_specific_data" type. */
6832
6833 static struct type *
6834 ada_get_tsd_type (struct inferior *inf)
6835 {
6836 struct ada_inferior_data *data = get_ada_inferior_data (inf);
6837
6838 if (data->tsd_type == 0)
6839 data->tsd_type = ada_find_any_type ("ada__tags__type_specific_data");
6840 return data->tsd_type;
6841 }
6842
6843 /* Return the TSD (type-specific data) associated to the given TAG.
6844 TAG is assumed to be the tag of a tagged-type entity.
6845
6846 May return NULL if we are unable to get the TSD. */
6847
6848 static struct value *
6849 ada_get_tsd_from_tag (struct value *tag)
6850 {
6851 struct value *val;
6852 struct type *type;
6853
6854 /* First option: The TSD is simply stored as a field of our TAG.
6855 Only older versions of GNAT would use this format, but we have
6856 to test it first, because there are no visible markers for
6857 the current approach except the absence of that field. */
6858
6859 val = ada_value_struct_elt (tag, "tsd", 1);
6860 if (val)
6861 return val;
6862
6863 /* Try the second representation for the dispatch table (in which
6864 there is no explicit 'tsd' field in the referent of the tag pointer,
6865 and instead the tsd pointer is stored just before the dispatch
6866 table. */
6867
6868 type = ada_get_tsd_type (current_inferior());
6869 if (type == NULL)
6870 return NULL;
6871 type = lookup_pointer_type (lookup_pointer_type (type));
6872 val = value_cast (type, tag);
6873 if (val == NULL)
6874 return NULL;
6875 return value_ind (value_ptradd (val, -1));
6876 }
6877
6878 /* Given the TSD of a tag (type-specific data), return a string
6879 containing the name of the associated type.
6880
6881 The returned value is good until the next call. May return NULL
6882 if we are unable to determine the tag name. */
6883
6884 static char *
6885 ada_tag_name_from_tsd (struct value *tsd)
6886 {
6887 static char name[1024];
6888 char *p;
6889 struct value *val;
6890
6891 val = ada_value_struct_elt (tsd, "expanded_name", 1);
6892 if (val == NULL)
6893 return NULL;
6894 read_memory_string (value_as_address (val), name, sizeof (name) - 1);
6895 for (p = name; *p != '\0'; p += 1)
6896 if (isalpha (*p))
6897 *p = tolower (*p);
6898 return name;
6899 }
6900
6901 /* The type name of the dynamic type denoted by the 'tag value TAG, as
6902 a C string.
6903
6904 Return NULL if the TAG is not an Ada tag, or if we were unable to
6905 determine the name of that tag. The result is good until the next
6906 call. */
6907
6908 const char *
6909 ada_tag_name (struct value *tag)
6910 {
6911 char *name = NULL;
6912
6913 if (!ada_is_tag_type (value_type (tag)))
6914 return NULL;
6915
6916 /* It is perfectly possible that an exception be raised while trying
6917 to determine the TAG's name, even under normal circumstances:
6918 The associated variable may be uninitialized or corrupted, for
6919 instance. We do not let any exception propagate past this point.
6920 instead we return NULL.
6921
6922 We also do not print the error message either (which often is very
6923 low-level (Eg: "Cannot read memory at 0x[...]"), but instead let
6924 the caller print a more meaningful message if necessary. */
6925 TRY
6926 {
6927 struct value *tsd = ada_get_tsd_from_tag (tag);
6928
6929 if (tsd != NULL)
6930 name = ada_tag_name_from_tsd (tsd);
6931 }
6932 CATCH (e, RETURN_MASK_ERROR)
6933 {
6934 }
6935 END_CATCH
6936
6937 return name;
6938 }
6939
6940 /* The parent type of TYPE, or NULL if none. */
6941
6942 struct type *
6943 ada_parent_type (struct type *type)
6944 {
6945 int i;
6946
6947 type = ada_check_typedef (type);
6948
6949 if (type == NULL || TYPE_CODE (type) != TYPE_CODE_STRUCT)
6950 return NULL;
6951
6952 for (i = 0; i < TYPE_NFIELDS (type); i += 1)
6953 if (ada_is_parent_field (type, i))
6954 {
6955 struct type *parent_type = TYPE_FIELD_TYPE (type, i);
6956
6957 /* If the _parent field is a pointer, then dereference it. */
6958 if (TYPE_CODE (parent_type) == TYPE_CODE_PTR)
6959 parent_type = TYPE_TARGET_TYPE (parent_type);
6960 /* If there is a parallel XVS type, get the actual base type. */
6961 parent_type = ada_get_base_type (parent_type);
6962
6963 return ada_check_typedef (parent_type);
6964 }
6965
6966 return NULL;
6967 }
6968
6969 /* True iff field number FIELD_NUM of structure type TYPE contains the
6970 parent-type (inherited) fields of a derived type. Assumes TYPE is
6971 a structure type with at least FIELD_NUM+1 fields. */
6972
6973 int
6974 ada_is_parent_field (struct type *type, int field_num)
6975 {
6976 const char *name = TYPE_FIELD_NAME (ada_check_typedef (type), field_num);
6977
6978 return (name != NULL
6979 && (startswith (name, "PARENT")
6980 || startswith (name, "_parent")));
6981 }
6982
6983 /* True iff field number FIELD_NUM of structure type TYPE is a
6984 transparent wrapper field (which should be silently traversed when doing
6985 field selection and flattened when printing). Assumes TYPE is a
6986 structure type with at least FIELD_NUM+1 fields. Such fields are always
6987 structures. */
6988
6989 int
6990 ada_is_wrapper_field (struct type *type, int field_num)
6991 {
6992 const char *name = TYPE_FIELD_NAME (type, field_num);
6993
6994 if (name != NULL && strcmp (name, "RETVAL") == 0)
6995 {
6996 /* This happens in functions with "out" or "in out" parameters
6997 which are passed by copy. For such functions, GNAT describes
6998 the function's return type as being a struct where the return
6999 value is in a field called RETVAL, and where the other "out"
7000 or "in out" parameters are fields of that struct. This is not
7001 a wrapper. */
7002 return 0;
7003 }
7004
7005 return (name != NULL
7006 && (startswith (name, "PARENT")
7007 || strcmp (name, "REP") == 0
7008 || startswith (name, "_parent")
7009 || name[0] == 'S' || name[0] == 'R' || name[0] == 'O'));
7010 }
7011
7012 /* True iff field number FIELD_NUM of structure or union type TYPE
7013 is a variant wrapper. Assumes TYPE is a structure type with at least
7014 FIELD_NUM+1 fields. */
7015
7016 int
7017 ada_is_variant_part (struct type *type, int field_num)
7018 {
7019 struct type *field_type = TYPE_FIELD_TYPE (type, field_num);
7020
7021 return (TYPE_CODE (field_type) == TYPE_CODE_UNION
7022 || (is_dynamic_field (type, field_num)
7023 && (TYPE_CODE (TYPE_TARGET_TYPE (field_type))
7024 == TYPE_CODE_UNION)));
7025 }
7026
7027 /* Assuming that VAR_TYPE is a variant wrapper (type of the variant part)
7028 whose discriminants are contained in the record type OUTER_TYPE,
7029 returns the type of the controlling discriminant for the variant.
7030 May return NULL if the type could not be found. */
7031
7032 struct type *
7033 ada_variant_discrim_type (struct type *var_type, struct type *outer_type)
7034 {
7035 const char *name = ada_variant_discrim_name (var_type);
7036
7037 return ada_lookup_struct_elt_type (outer_type, name, 1, 1);
7038 }
7039
7040 /* Assuming that TYPE is the type of a variant wrapper, and FIELD_NUM is a
7041 valid field number within it, returns 1 iff field FIELD_NUM of TYPE
7042 represents a 'when others' clause; otherwise 0. */
7043
7044 int
7045 ada_is_others_clause (struct type *type, int field_num)
7046 {
7047 const char *name = TYPE_FIELD_NAME (type, field_num);
7048
7049 return (name != NULL && name[0] == 'O');
7050 }
7051
7052 /* Assuming that TYPE0 is the type of the variant part of a record,
7053 returns the name of the discriminant controlling the variant.
7054 The value is valid until the next call to ada_variant_discrim_name. */
7055
7056 const char *
7057 ada_variant_discrim_name (struct type *type0)
7058 {
7059 static char *result = NULL;
7060 static size_t result_len = 0;
7061 struct type *type;
7062 const char *name;
7063 const char *discrim_end;
7064 const char *discrim_start;
7065
7066 if (TYPE_CODE (type0) == TYPE_CODE_PTR)
7067 type = TYPE_TARGET_TYPE (type0);
7068 else
7069 type = type0;
7070
7071 name = ada_type_name (type);
7072
7073 if (name == NULL || name[0] == '\000')
7074 return "";
7075
7076 for (discrim_end = name + strlen (name) - 6; discrim_end != name;
7077 discrim_end -= 1)
7078 {
7079 if (startswith (discrim_end, "___XVN"))
7080 break;
7081 }
7082 if (discrim_end == name)
7083 return "";
7084
7085 for (discrim_start = discrim_end; discrim_start != name + 3;
7086 discrim_start -= 1)
7087 {
7088 if (discrim_start == name + 1)
7089 return "";
7090 if ((discrim_start > name + 3
7091 && startswith (discrim_start - 3, "___"))
7092 || discrim_start[-1] == '.')
7093 break;
7094 }
7095
7096 GROW_VECT (result, result_len, discrim_end - discrim_start + 1);
7097 strncpy (result, discrim_start, discrim_end - discrim_start);
7098 result[discrim_end - discrim_start] = '\0';
7099 return result;
7100 }
7101
7102 /* Scan STR for a subtype-encoded number, beginning at position K.
7103 Put the position of the character just past the number scanned in
7104 *NEW_K, if NEW_K!=NULL. Put the scanned number in *R, if R!=NULL.
7105 Return 1 if there was a valid number at the given position, and 0
7106 otherwise. A "subtype-encoded" number consists of the absolute value
7107 in decimal, followed by the letter 'm' to indicate a negative number.
7108 Assumes 0m does not occur. */
7109
7110 int
7111 ada_scan_number (const char str[], int k, LONGEST * R, int *new_k)
7112 {
7113 ULONGEST RU;
7114
7115 if (!isdigit (str[k]))
7116 return 0;
7117
7118 /* Do it the hard way so as not to make any assumption about
7119 the relationship of unsigned long (%lu scan format code) and
7120 LONGEST. */
7121 RU = 0;
7122 while (isdigit (str[k]))
7123 {
7124 RU = RU * 10 + (str[k] - '0');
7125 k += 1;
7126 }
7127
7128 if (str[k] == 'm')
7129 {
7130 if (R != NULL)
7131 *R = (-(LONGEST) (RU - 1)) - 1;
7132 k += 1;
7133 }
7134 else if (R != NULL)
7135 *R = (LONGEST) RU;
7136
7137 /* NOTE on the above: Technically, C does not say what the results of
7138 - (LONGEST) RU or (LONGEST) -RU are for RU == largest positive
7139 number representable as a LONGEST (although either would probably work
7140 in most implementations). When RU>0, the locution in the then branch
7141 above is always equivalent to the negative of RU. */
7142
7143 if (new_k != NULL)
7144 *new_k = k;
7145 return 1;
7146 }
7147
7148 /* Assuming that TYPE is a variant part wrapper type (a VARIANTS field),
7149 and FIELD_NUM is a valid field number within it, returns 1 iff VAL is
7150 in the range encoded by field FIELD_NUM of TYPE; otherwise 0. */
7151
7152 int
7153 ada_in_variant (LONGEST val, struct type *type, int field_num)
7154 {
7155 const char *name = TYPE_FIELD_NAME (type, field_num);
7156 int p;
7157
7158 p = 0;
7159 while (1)
7160 {
7161 switch (name[p])
7162 {
7163 case '\0':
7164 return 0;
7165 case 'S':
7166 {
7167 LONGEST W;
7168
7169 if (!ada_scan_number (name, p + 1, &W, &p))
7170 return 0;
7171 if (val == W)
7172 return 1;
7173 break;
7174 }
7175 case 'R':
7176 {
7177 LONGEST L, U;
7178
7179 if (!ada_scan_number (name, p + 1, &L, &p)
7180 || name[p] != 'T' || !ada_scan_number (name, p + 1, &U, &p))
7181 return 0;
7182 if (val >= L && val <= U)
7183 return 1;
7184 break;
7185 }
7186 case 'O':
7187 return 1;
7188 default:
7189 return 0;
7190 }
7191 }
7192 }
7193
7194 /* FIXME: Lots of redundancy below. Try to consolidate. */
7195
7196 /* Given a value ARG1 (offset by OFFSET bytes) of a struct or union type
7197 ARG_TYPE, extract and return the value of one of its (non-static)
7198 fields. FIELDNO says which field. Differs from value_primitive_field
7199 only in that it can handle packed values of arbitrary type. */
7200
7201 static struct value *
7202 ada_value_primitive_field (struct value *arg1, int offset, int fieldno,
7203 struct type *arg_type)
7204 {
7205 struct type *type;
7206
7207 arg_type = ada_check_typedef (arg_type);
7208 type = TYPE_FIELD_TYPE (arg_type, fieldno);
7209
7210 /* Handle packed fields. */
7211
7212 if (TYPE_FIELD_BITSIZE (arg_type, fieldno) != 0)
7213 {
7214 int bit_pos = TYPE_FIELD_BITPOS (arg_type, fieldno);
7215 int bit_size = TYPE_FIELD_BITSIZE (arg_type, fieldno);
7216
7217 return ada_value_primitive_packed_val (arg1, value_contents (arg1),
7218 offset + bit_pos / 8,
7219 bit_pos % 8, bit_size, type);
7220 }
7221 else
7222 return value_primitive_field (arg1, offset, fieldno, arg_type);
7223 }
7224
7225 /* Find field with name NAME in object of type TYPE. If found,
7226 set the following for each argument that is non-null:
7227 - *FIELD_TYPE_P to the field's type;
7228 - *BYTE_OFFSET_P to OFFSET + the byte offset of the field within
7229 an object of that type;
7230 - *BIT_OFFSET_P to the bit offset modulo byte size of the field;
7231 - *BIT_SIZE_P to its size in bits if the field is packed, and
7232 0 otherwise;
7233 If INDEX_P is non-null, increment *INDEX_P by the number of source-visible
7234 fields up to but not including the desired field, or by the total
7235 number of fields if not found. A NULL value of NAME never
7236 matches; the function just counts visible fields in this case.
7237
7238 Notice that we need to handle when a tagged record hierarchy
7239 has some components with the same name, like in this scenario:
7240
7241 type Top_T is tagged record
7242 N : Integer := 1;
7243 U : Integer := 974;
7244 A : Integer := 48;
7245 end record;
7246
7247 type Middle_T is new Top.Top_T with record
7248 N : Character := 'a';
7249 C : Integer := 3;
7250 end record;
7251
7252 type Bottom_T is new Middle.Middle_T with record
7253 N : Float := 4.0;
7254 C : Character := '5';
7255 X : Integer := 6;
7256 A : Character := 'J';
7257 end record;
7258
7259 Let's say we now have a variable declared and initialized as follow:
7260
7261 TC : Top_A := new Bottom_T;
7262
7263 And then we use this variable to call this function
7264
7265 procedure Assign (Obj: in out Top_T; TV : Integer);
7266
7267 as follow:
7268
7269 Assign (Top_T (B), 12);
7270
7271 Now, we're in the debugger, and we're inside that procedure
7272 then and we want to print the value of obj.c:
7273
7274 Usually, the tagged record or one of the parent type owns the
7275 component to print and there's no issue but in this particular
7276 case, what does it mean to ask for Obj.C? Since the actual
7277 type for object is type Bottom_T, it could mean two things: type
7278 component C from the Middle_T view, but also component C from
7279 Bottom_T. So in that "undefined" case, when the component is
7280 not found in the non-resolved type (which includes all the
7281 components of the parent type), then resolve it and see if we
7282 get better luck once expanded.
7283
7284 In the case of homonyms in the derived tagged type, we don't
7285 guaranty anything, and pick the one that's easiest for us
7286 to program.
7287
7288 Returns 1 if found, 0 otherwise. */
7289
7290 static int
7291 find_struct_field (const char *name, struct type *type, int offset,
7292 struct type **field_type_p,
7293 int *byte_offset_p, int *bit_offset_p, int *bit_size_p,
7294 int *index_p)
7295 {
7296 int i;
7297 int parent_offset = -1;
7298
7299 type = ada_check_typedef (type);
7300
7301 if (field_type_p != NULL)
7302 *field_type_p = NULL;
7303 if (byte_offset_p != NULL)
7304 *byte_offset_p = 0;
7305 if (bit_offset_p != NULL)
7306 *bit_offset_p = 0;
7307 if (bit_size_p != NULL)
7308 *bit_size_p = 0;
7309
7310 for (i = 0; i < TYPE_NFIELDS (type); i += 1)
7311 {
7312 int bit_pos = TYPE_FIELD_BITPOS (type, i);
7313 int fld_offset = offset + bit_pos / 8;
7314 const char *t_field_name = TYPE_FIELD_NAME (type, i);
7315
7316 if (t_field_name == NULL)
7317 continue;
7318
7319 else if (ada_is_parent_field (type, i))
7320 {
7321 /* This is a field pointing us to the parent type of a tagged
7322 type. As hinted in this function's documentation, we give
7323 preference to fields in the current record first, so what
7324 we do here is just record the index of this field before
7325 we skip it. If it turns out we couldn't find our field
7326 in the current record, then we'll get back to it and search
7327 inside it whether the field might exist in the parent. */
7328
7329 parent_offset = i;
7330 continue;
7331 }
7332
7333 else if (name != NULL && field_name_match (t_field_name, name))
7334 {
7335 int bit_size = TYPE_FIELD_BITSIZE (type, i);
7336
7337 if (field_type_p != NULL)
7338 *field_type_p = TYPE_FIELD_TYPE (type, i);
7339 if (byte_offset_p != NULL)
7340 *byte_offset_p = fld_offset;
7341 if (bit_offset_p != NULL)
7342 *bit_offset_p = bit_pos % 8;
7343 if (bit_size_p != NULL)
7344 *bit_size_p = bit_size;
7345 return 1;
7346 }
7347 else if (ada_is_wrapper_field (type, i))
7348 {
7349 if (find_struct_field (name, TYPE_FIELD_TYPE (type, i), fld_offset,
7350 field_type_p, byte_offset_p, bit_offset_p,
7351 bit_size_p, index_p))
7352 return 1;
7353 }
7354 else if (ada_is_variant_part (type, i))
7355 {
7356 /* PNH: Wait. Do we ever execute this section, or is ARG always of
7357 fixed type?? */
7358 int j;
7359 struct type *field_type
7360 = ada_check_typedef (TYPE_FIELD_TYPE (type, i));
7361
7362 for (j = 0; j < TYPE_NFIELDS (field_type); j += 1)
7363 {
7364 if (find_struct_field (name, TYPE_FIELD_TYPE (field_type, j),
7365 fld_offset
7366 + TYPE_FIELD_BITPOS (field_type, j) / 8,
7367 field_type_p, byte_offset_p,
7368 bit_offset_p, bit_size_p, index_p))
7369 return 1;
7370 }
7371 }
7372 else if (index_p != NULL)
7373 *index_p += 1;
7374 }
7375
7376 /* Field not found so far. If this is a tagged type which
7377 has a parent, try finding that field in the parent now. */
7378
7379 if (parent_offset != -1)
7380 {
7381 int bit_pos = TYPE_FIELD_BITPOS (type, parent_offset);
7382 int fld_offset = offset + bit_pos / 8;
7383
7384 if (find_struct_field (name, TYPE_FIELD_TYPE (type, parent_offset),
7385 fld_offset, field_type_p, byte_offset_p,
7386 bit_offset_p, bit_size_p, index_p))
7387 return 1;
7388 }
7389
7390 return 0;
7391 }
7392
7393 /* Number of user-visible fields in record type TYPE. */
7394
7395 static int
7396 num_visible_fields (struct type *type)
7397 {
7398 int n;
7399
7400 n = 0;
7401 find_struct_field (NULL, type, 0, NULL, NULL, NULL, NULL, &n);
7402 return n;
7403 }
7404
7405 /* Look for a field NAME in ARG. Adjust the address of ARG by OFFSET bytes,
7406 and search in it assuming it has (class) type TYPE.
7407 If found, return value, else return NULL.
7408
7409 Searches recursively through wrapper fields (e.g., '_parent').
7410
7411 In the case of homonyms in the tagged types, please refer to the
7412 long explanation in find_struct_field's function documentation. */
7413
7414 static struct value *
7415 ada_search_struct_field (const char *name, struct value *arg, int offset,
7416 struct type *type)
7417 {
7418 int i;
7419 int parent_offset = -1;
7420
7421 type = ada_check_typedef (type);
7422 for (i = 0; i < TYPE_NFIELDS (type); i += 1)
7423 {
7424 const char *t_field_name = TYPE_FIELD_NAME (type, i);
7425
7426 if (t_field_name == NULL)
7427 continue;
7428
7429 else if (ada_is_parent_field (type, i))
7430 {
7431 /* This is a field pointing us to the parent type of a tagged
7432 type. As hinted in this function's documentation, we give
7433 preference to fields in the current record first, so what
7434 we do here is just record the index of this field before
7435 we skip it. If it turns out we couldn't find our field
7436 in the current record, then we'll get back to it and search
7437 inside it whether the field might exist in the parent. */
7438
7439 parent_offset = i;
7440 continue;
7441 }
7442
7443 else if (field_name_match (t_field_name, name))
7444 return ada_value_primitive_field (arg, offset, i, type);
7445
7446 else if (ada_is_wrapper_field (type, i))
7447 {
7448 struct value *v = /* Do not let indent join lines here. */
7449 ada_search_struct_field (name, arg,
7450 offset + TYPE_FIELD_BITPOS (type, i) / 8,
7451 TYPE_FIELD_TYPE (type, i));
7452
7453 if (v != NULL)
7454 return v;
7455 }
7456
7457 else if (ada_is_variant_part (type, i))
7458 {
7459 /* PNH: Do we ever get here? See find_struct_field. */
7460 int j;
7461 struct type *field_type = ada_check_typedef (TYPE_FIELD_TYPE (type,
7462 i));
7463 int var_offset = offset + TYPE_FIELD_BITPOS (type, i) / 8;
7464
7465 for (j = 0; j < TYPE_NFIELDS (field_type); j += 1)
7466 {
7467 struct value *v = ada_search_struct_field /* Force line
7468 break. */
7469 (name, arg,
7470 var_offset + TYPE_FIELD_BITPOS (field_type, j) / 8,
7471 TYPE_FIELD_TYPE (field_type, j));
7472
7473 if (v != NULL)
7474 return v;
7475 }
7476 }
7477 }
7478
7479 /* Field not found so far. If this is a tagged type which
7480 has a parent, try finding that field in the parent now. */
7481
7482 if (parent_offset != -1)
7483 {
7484 struct value *v = ada_search_struct_field (
7485 name, arg, offset + TYPE_FIELD_BITPOS (type, parent_offset) / 8,
7486 TYPE_FIELD_TYPE (type, parent_offset));
7487
7488 if (v != NULL)
7489 return v;
7490 }
7491
7492 return NULL;
7493 }
7494
7495 static struct value *ada_index_struct_field_1 (int *, struct value *,
7496 int, struct type *);
7497
7498
7499 /* Return field #INDEX in ARG, where the index is that returned by
7500 * find_struct_field through its INDEX_P argument. Adjust the address
7501 * of ARG by OFFSET bytes, and search in it assuming it has (class) type TYPE.
7502 * If found, return value, else return NULL. */
7503
7504 static struct value *
7505 ada_index_struct_field (int index, struct value *arg, int offset,
7506 struct type *type)
7507 {
7508 return ada_index_struct_field_1 (&index, arg, offset, type);
7509 }
7510
7511
7512 /* Auxiliary function for ada_index_struct_field. Like
7513 * ada_index_struct_field, but takes index from *INDEX_P and modifies
7514 * *INDEX_P. */
7515
7516 static struct value *
7517 ada_index_struct_field_1 (int *index_p, struct value *arg, int offset,
7518 struct type *type)
7519 {
7520 int i;
7521 type = ada_check_typedef (type);
7522
7523 for (i = 0; i < TYPE_NFIELDS (type); i += 1)
7524 {
7525 if (TYPE_FIELD_NAME (type, i) == NULL)
7526 continue;
7527 else if (ada_is_wrapper_field (type, i))
7528 {
7529 struct value *v = /* Do not let indent join lines here. */
7530 ada_index_struct_field_1 (index_p, arg,
7531 offset + TYPE_FIELD_BITPOS (type, i) / 8,
7532 TYPE_FIELD_TYPE (type, i));
7533
7534 if (v != NULL)
7535 return v;
7536 }
7537
7538 else if (ada_is_variant_part (type, i))
7539 {
7540 /* PNH: Do we ever get here? See ada_search_struct_field,
7541 find_struct_field. */
7542 error (_("Cannot assign this kind of variant record"));
7543 }
7544 else if (*index_p == 0)
7545 return ada_value_primitive_field (arg, offset, i, type);
7546 else
7547 *index_p -= 1;
7548 }
7549 return NULL;
7550 }
7551
7552 /* Given ARG, a value of type (pointer or reference to a)*
7553 structure/union, extract the component named NAME from the ultimate
7554 target structure/union and return it as a value with its
7555 appropriate type.
7556
7557 The routine searches for NAME among all members of the structure itself
7558 and (recursively) among all members of any wrapper members
7559 (e.g., '_parent').
7560
7561 If NO_ERR, then simply return NULL in case of error, rather than
7562 calling error. */
7563
7564 struct value *
7565 ada_value_struct_elt (struct value *arg, const char *name, int no_err)
7566 {
7567 struct type *t, *t1;
7568 struct value *v;
7569
7570 v = NULL;
7571 t1 = t = ada_check_typedef (value_type (arg));
7572 if (TYPE_CODE (t) == TYPE_CODE_REF)
7573 {
7574 t1 = TYPE_TARGET_TYPE (t);
7575 if (t1 == NULL)
7576 goto BadValue;
7577 t1 = ada_check_typedef (t1);
7578 if (TYPE_CODE (t1) == TYPE_CODE_PTR)
7579 {
7580 arg = coerce_ref (arg);
7581 t = t1;
7582 }
7583 }
7584
7585 while (TYPE_CODE (t) == TYPE_CODE_PTR)
7586 {
7587 t1 = TYPE_TARGET_TYPE (t);
7588 if (t1 == NULL)
7589 goto BadValue;
7590 t1 = ada_check_typedef (t1);
7591 if (TYPE_CODE (t1) == TYPE_CODE_PTR)
7592 {
7593 arg = value_ind (arg);
7594 t = t1;
7595 }
7596 else
7597 break;
7598 }
7599
7600 if (TYPE_CODE (t1) != TYPE_CODE_STRUCT && TYPE_CODE (t1) != TYPE_CODE_UNION)
7601 goto BadValue;
7602
7603 if (t1 == t)
7604 v = ada_search_struct_field (name, arg, 0, t);
7605 else
7606 {
7607 int bit_offset, bit_size, byte_offset;
7608 struct type *field_type;
7609 CORE_ADDR address;
7610
7611 if (TYPE_CODE (t) == TYPE_CODE_PTR)
7612 address = value_address (ada_value_ind (arg));
7613 else
7614 address = value_address (ada_coerce_ref (arg));
7615
7616 /* Check to see if this is a tagged type. We also need to handle
7617 the case where the type is a reference to a tagged type, but
7618 we have to be careful to exclude pointers to tagged types.
7619 The latter should be shown as usual (as a pointer), whereas
7620 a reference should mostly be transparent to the user. */
7621
7622 if (ada_is_tagged_type (t1, 0)
7623 || (TYPE_CODE (t1) == TYPE_CODE_REF
7624 && ada_is_tagged_type (TYPE_TARGET_TYPE (t1), 0)))
7625 {
7626 /* We first try to find the searched field in the current type.
7627 If not found then let's look in the fixed type. */
7628
7629 if (!find_struct_field (name, t1, 0,
7630 &field_type, &byte_offset, &bit_offset,
7631 &bit_size, NULL))
7632 t1 = ada_to_fixed_type (ada_get_base_type (t1), NULL,
7633 address, NULL, 1);
7634 }
7635 else
7636 t1 = ada_to_fixed_type (ada_get_base_type (t1), NULL,
7637 address, NULL, 1);
7638
7639 if (find_struct_field (name, t1, 0,
7640 &field_type, &byte_offset, &bit_offset,
7641 &bit_size, NULL))
7642 {
7643 if (bit_size != 0)
7644 {
7645 if (TYPE_CODE (t) == TYPE_CODE_REF)
7646 arg = ada_coerce_ref (arg);
7647 else
7648 arg = ada_value_ind (arg);
7649 v = ada_value_primitive_packed_val (arg, NULL, byte_offset,
7650 bit_offset, bit_size,
7651 field_type);
7652 }
7653 else
7654 v = value_at_lazy (field_type, address + byte_offset);
7655 }
7656 }
7657
7658 if (v != NULL || no_err)
7659 return v;
7660 else
7661 error (_("There is no member named %s."), name);
7662
7663 BadValue:
7664 if (no_err)
7665 return NULL;
7666 else
7667 error (_("Attempt to extract a component of "
7668 "a value that is not a record."));
7669 }
7670
7671 /* Return a string representation of type TYPE. */
7672
7673 static std::string
7674 type_as_string (struct type *type)
7675 {
7676 string_file tmp_stream;
7677
7678 type_print (type, "", &tmp_stream, -1);
7679
7680 return std::move (tmp_stream.string ());
7681 }
7682
7683 /* Given a type TYPE, look up the type of the component of type named NAME.
7684 If DISPP is non-null, add its byte displacement from the beginning of a
7685 structure (pointed to by a value) of type TYPE to *DISPP (does not
7686 work for packed fields).
7687
7688 Matches any field whose name has NAME as a prefix, possibly
7689 followed by "___".
7690
7691 TYPE can be either a struct or union. If REFOK, TYPE may also
7692 be a (pointer or reference)+ to a struct or union, and the
7693 ultimate target type will be searched.
7694
7695 Looks recursively into variant clauses and parent types.
7696
7697 In the case of homonyms in the tagged types, please refer to the
7698 long explanation in find_struct_field's function documentation.
7699
7700 If NOERR is nonzero, return NULL if NAME is not suitably defined or
7701 TYPE is not a type of the right kind. */
7702
7703 static struct type *
7704 ada_lookup_struct_elt_type (struct type *type, const char *name, int refok,
7705 int noerr)
7706 {
7707 int i;
7708 int parent_offset = -1;
7709
7710 if (name == NULL)
7711 goto BadName;
7712
7713 if (refok && type != NULL)
7714 while (1)
7715 {
7716 type = ada_check_typedef (type);
7717 if (TYPE_CODE (type) != TYPE_CODE_PTR
7718 && TYPE_CODE (type) != TYPE_CODE_REF)
7719 break;
7720 type = TYPE_TARGET_TYPE (type);
7721 }
7722
7723 if (type == NULL
7724 || (TYPE_CODE (type) != TYPE_CODE_STRUCT
7725 && TYPE_CODE (type) != TYPE_CODE_UNION))
7726 {
7727 if (noerr)
7728 return NULL;
7729
7730 error (_("Type %s is not a structure or union type"),
7731 type != NULL ? type_as_string (type).c_str () : _("(null)"));
7732 }
7733
7734 type = to_static_fixed_type (type);
7735
7736 for (i = 0; i < TYPE_NFIELDS (type); i += 1)
7737 {
7738 const char *t_field_name = TYPE_FIELD_NAME (type, i);
7739 struct type *t;
7740
7741 if (t_field_name == NULL)
7742 continue;
7743
7744 else if (ada_is_parent_field (type, i))
7745 {
7746 /* This is a field pointing us to the parent type of a tagged
7747 type. As hinted in this function's documentation, we give
7748 preference to fields in the current record first, so what
7749 we do here is just record the index of this field before
7750 we skip it. If it turns out we couldn't find our field
7751 in the current record, then we'll get back to it and search
7752 inside it whether the field might exist in the parent. */
7753
7754 parent_offset = i;
7755 continue;
7756 }
7757
7758 else if (field_name_match (t_field_name, name))
7759 return TYPE_FIELD_TYPE (type, i);
7760
7761 else if (ada_is_wrapper_field (type, i))
7762 {
7763 t = ada_lookup_struct_elt_type (TYPE_FIELD_TYPE (type, i), name,
7764 0, 1);
7765 if (t != NULL)
7766 return t;
7767 }
7768
7769 else if (ada_is_variant_part (type, i))
7770 {
7771 int j;
7772 struct type *field_type = ada_check_typedef (TYPE_FIELD_TYPE (type,
7773 i));
7774
7775 for (j = TYPE_NFIELDS (field_type) - 1; j >= 0; j -= 1)
7776 {
7777 /* FIXME pnh 2008/01/26: We check for a field that is
7778 NOT wrapped in a struct, since the compiler sometimes
7779 generates these for unchecked variant types. Revisit
7780 if the compiler changes this practice. */
7781 const char *v_field_name = TYPE_FIELD_NAME (field_type, j);
7782
7783 if (v_field_name != NULL
7784 && field_name_match (v_field_name, name))
7785 t = TYPE_FIELD_TYPE (field_type, j);
7786 else
7787 t = ada_lookup_struct_elt_type (TYPE_FIELD_TYPE (field_type,
7788 j),
7789 name, 0, 1);
7790
7791 if (t != NULL)
7792 return t;
7793 }
7794 }
7795
7796 }
7797
7798 /* Field not found so far. If this is a tagged type which
7799 has a parent, try finding that field in the parent now. */
7800
7801 if (parent_offset != -1)
7802 {
7803 struct type *t;
7804
7805 t = ada_lookup_struct_elt_type (TYPE_FIELD_TYPE (type, parent_offset),
7806 name, 0, 1);
7807 if (t != NULL)
7808 return t;
7809 }
7810
7811 BadName:
7812 if (!noerr)
7813 {
7814 const char *name_str = name != NULL ? name : _("<null>");
7815
7816 error (_("Type %s has no component named %s"),
7817 type_as_string (type).c_str (), name_str);
7818 }
7819
7820 return NULL;
7821 }
7822
7823 /* Assuming that VAR_TYPE is the type of a variant part of a record (a union),
7824 within a value of type OUTER_TYPE, return true iff VAR_TYPE
7825 represents an unchecked union (that is, the variant part of a
7826 record that is named in an Unchecked_Union pragma). */
7827
7828 static int
7829 is_unchecked_variant (struct type *var_type, struct type *outer_type)
7830 {
7831 const char *discrim_name = ada_variant_discrim_name (var_type);
7832
7833 return (ada_lookup_struct_elt_type (outer_type, discrim_name, 0, 1) == NULL);
7834 }
7835
7836
7837 /* Assuming that VAR_TYPE is the type of a variant part of a record (a union),
7838 within a value of type OUTER_TYPE that is stored in GDB at
7839 OUTER_VALADDR, determine which variant clause (field number in VAR_TYPE,
7840 numbering from 0) is applicable. Returns -1 if none are. */
7841
7842 int
7843 ada_which_variant_applies (struct type *var_type, struct type *outer_type,
7844 const gdb_byte *outer_valaddr)
7845 {
7846 int others_clause;
7847 int i;
7848 const char *discrim_name = ada_variant_discrim_name (var_type);
7849 struct value *outer;
7850 struct value *discrim;
7851 LONGEST discrim_val;
7852
7853 /* Using plain value_from_contents_and_address here causes problems
7854 because we will end up trying to resolve a type that is currently
7855 being constructed. */
7856 outer = value_from_contents_and_address_unresolved (outer_type,
7857 outer_valaddr, 0);
7858 discrim = ada_value_struct_elt (outer, discrim_name, 1);
7859 if (discrim == NULL)
7860 return -1;
7861 discrim_val = value_as_long (discrim);
7862
7863 others_clause = -1;
7864 for (i = 0; i < TYPE_NFIELDS (var_type); i += 1)
7865 {
7866 if (ada_is_others_clause (var_type, i))
7867 others_clause = i;
7868 else if (ada_in_variant (discrim_val, var_type, i))
7869 return i;
7870 }
7871
7872 return others_clause;
7873 }
7874 \f
7875
7876
7877 /* Dynamic-Sized Records */
7878
7879 /* Strategy: The type ostensibly attached to a value with dynamic size
7880 (i.e., a size that is not statically recorded in the debugging
7881 data) does not accurately reflect the size or layout of the value.
7882 Our strategy is to convert these values to values with accurate,
7883 conventional types that are constructed on the fly. */
7884
7885 /* There is a subtle and tricky problem here. In general, we cannot
7886 determine the size of dynamic records without its data. However,
7887 the 'struct value' data structure, which GDB uses to represent
7888 quantities in the inferior process (the target), requires the size
7889 of the type at the time of its allocation in order to reserve space
7890 for GDB's internal copy of the data. That's why the
7891 'to_fixed_xxx_type' routines take (target) addresses as parameters,
7892 rather than struct value*s.
7893
7894 However, GDB's internal history variables ($1, $2, etc.) are
7895 struct value*s containing internal copies of the data that are not, in
7896 general, the same as the data at their corresponding addresses in
7897 the target. Fortunately, the types we give to these values are all
7898 conventional, fixed-size types (as per the strategy described
7899 above), so that we don't usually have to perform the
7900 'to_fixed_xxx_type' conversions to look at their values.
7901 Unfortunately, there is one exception: if one of the internal
7902 history variables is an array whose elements are unconstrained
7903 records, then we will need to create distinct fixed types for each
7904 element selected. */
7905
7906 /* The upshot of all of this is that many routines take a (type, host
7907 address, target address) triple as arguments to represent a value.
7908 The host address, if non-null, is supposed to contain an internal
7909 copy of the relevant data; otherwise, the program is to consult the
7910 target at the target address. */
7911
7912 /* Assuming that VAL0 represents a pointer value, the result of
7913 dereferencing it. Differs from value_ind in its treatment of
7914 dynamic-sized types. */
7915
7916 struct value *
7917 ada_value_ind (struct value *val0)
7918 {
7919 struct value *val = value_ind (val0);
7920
7921 if (ada_is_tagged_type (value_type (val), 0))
7922 val = ada_tag_value_at_base_address (val);
7923
7924 return ada_to_fixed_value (val);
7925 }
7926
7927 /* The value resulting from dereferencing any "reference to"
7928 qualifiers on VAL0. */
7929
7930 static struct value *
7931 ada_coerce_ref (struct value *val0)
7932 {
7933 if (TYPE_CODE (value_type (val0)) == TYPE_CODE_REF)
7934 {
7935 struct value *val = val0;
7936
7937 val = coerce_ref (val);
7938
7939 if (ada_is_tagged_type (value_type (val), 0))
7940 val = ada_tag_value_at_base_address (val);
7941
7942 return ada_to_fixed_value (val);
7943 }
7944 else
7945 return val0;
7946 }
7947
7948 /* Return OFF rounded upward if necessary to a multiple of
7949 ALIGNMENT (a power of 2). */
7950
7951 static unsigned int
7952 align_value (unsigned int off, unsigned int alignment)
7953 {
7954 return (off + alignment - 1) & ~(alignment - 1);
7955 }
7956
7957 /* Return the bit alignment required for field #F of template type TYPE. */
7958
7959 static unsigned int
7960 field_alignment (struct type *type, int f)
7961 {
7962 const char *name = TYPE_FIELD_NAME (type, f);
7963 int len;
7964 int align_offset;
7965
7966 /* The field name should never be null, unless the debugging information
7967 is somehow malformed. In this case, we assume the field does not
7968 require any alignment. */
7969 if (name == NULL)
7970 return 1;
7971
7972 len = strlen (name);
7973
7974 if (!isdigit (name[len - 1]))
7975 return 1;
7976
7977 if (isdigit (name[len - 2]))
7978 align_offset = len - 2;
7979 else
7980 align_offset = len - 1;
7981
7982 if (align_offset < 7 || !startswith (name + align_offset - 6, "___XV"))
7983 return TARGET_CHAR_BIT;
7984
7985 return atoi (name + align_offset) * TARGET_CHAR_BIT;
7986 }
7987
7988 /* Find a typedef or tag symbol named NAME. Ignores ambiguity. */
7989
7990 static struct symbol *
7991 ada_find_any_type_symbol (const char *name)
7992 {
7993 struct symbol *sym;
7994
7995 sym = standard_lookup (name, get_selected_block (NULL), VAR_DOMAIN);
7996 if (sym != NULL && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
7997 return sym;
7998
7999 sym = standard_lookup (name, NULL, STRUCT_DOMAIN);
8000 return sym;
8001 }
8002
8003 /* Find a type named NAME. Ignores ambiguity. This routine will look
8004 solely for types defined by debug info, it will not search the GDB
8005 primitive types. */
8006
8007 static struct type *
8008 ada_find_any_type (const char *name)
8009 {
8010 struct symbol *sym = ada_find_any_type_symbol (name);
8011
8012 if (sym != NULL)
8013 return SYMBOL_TYPE (sym);
8014
8015 return NULL;
8016 }
8017
8018 /* Given NAME_SYM and an associated BLOCK, find a "renaming" symbol
8019 associated with NAME_SYM's name. NAME_SYM may itself be a renaming
8020 symbol, in which case it is returned. Otherwise, this looks for
8021 symbols whose name is that of NAME_SYM suffixed with "___XR".
8022 Return symbol if found, and NULL otherwise. */
8023
8024 struct symbol *
8025 ada_find_renaming_symbol (struct symbol *name_sym, const struct block *block)
8026 {
8027 const char *name = SYMBOL_LINKAGE_NAME (name_sym);
8028 struct symbol *sym;
8029
8030 if (strstr (name, "___XR") != NULL)
8031 return name_sym;
8032
8033 sym = find_old_style_renaming_symbol (name, block);
8034
8035 if (sym != NULL)
8036 return sym;
8037
8038 /* Not right yet. FIXME pnh 7/20/2007. */
8039 sym = ada_find_any_type_symbol (name);
8040 if (sym != NULL && strstr (SYMBOL_LINKAGE_NAME (sym), "___XR") != NULL)
8041 return sym;
8042 else
8043 return NULL;
8044 }
8045
8046 static struct symbol *
8047 find_old_style_renaming_symbol (const char *name, const struct block *block)
8048 {
8049 const struct symbol *function_sym = block_linkage_function (block);
8050 char *rename;
8051
8052 if (function_sym != NULL)
8053 {
8054 /* If the symbol is defined inside a function, NAME is not fully
8055 qualified. This means we need to prepend the function name
8056 as well as adding the ``___XR'' suffix to build the name of
8057 the associated renaming symbol. */
8058 const char *function_name = SYMBOL_LINKAGE_NAME (function_sym);
8059 /* Function names sometimes contain suffixes used
8060 for instance to qualify nested subprograms. When building
8061 the XR type name, we need to make sure that this suffix is
8062 not included. So do not include any suffix in the function
8063 name length below. */
8064 int function_name_len = ada_name_prefix_len (function_name);
8065 const int rename_len = function_name_len + 2 /* "__" */
8066 + strlen (name) + 6 /* "___XR\0" */ ;
8067
8068 /* Strip the suffix if necessary. */
8069 ada_remove_trailing_digits (function_name, &function_name_len);
8070 ada_remove_po_subprogram_suffix (function_name, &function_name_len);
8071 ada_remove_Xbn_suffix (function_name, &function_name_len);
8072
8073 /* Library-level functions are a special case, as GNAT adds
8074 a ``_ada_'' prefix to the function name to avoid namespace
8075 pollution. However, the renaming symbols themselves do not
8076 have this prefix, so we need to skip this prefix if present. */
8077 if (function_name_len > 5 /* "_ada_" */
8078 && strstr (function_name, "_ada_") == function_name)
8079 {
8080 function_name += 5;
8081 function_name_len -= 5;
8082 }
8083
8084 rename = (char *) alloca (rename_len * sizeof (char));
8085 strncpy (rename, function_name, function_name_len);
8086 xsnprintf (rename + function_name_len, rename_len - function_name_len,
8087 "__%s___XR", name);
8088 }
8089 else
8090 {
8091 const int rename_len = strlen (name) + 6;
8092
8093 rename = (char *) alloca (rename_len * sizeof (char));
8094 xsnprintf (rename, rename_len * sizeof (char), "%s___XR", name);
8095 }
8096
8097 return ada_find_any_type_symbol (rename);
8098 }
8099
8100 /* Because of GNAT encoding conventions, several GDB symbols may match a
8101 given type name. If the type denoted by TYPE0 is to be preferred to
8102 that of TYPE1 for purposes of type printing, return non-zero;
8103 otherwise return 0. */
8104
8105 int
8106 ada_prefer_type (struct type *type0, struct type *type1)
8107 {
8108 if (type1 == NULL)
8109 return 1;
8110 else if (type0 == NULL)
8111 return 0;
8112 else if (TYPE_CODE (type1) == TYPE_CODE_VOID)
8113 return 1;
8114 else if (TYPE_CODE (type0) == TYPE_CODE_VOID)
8115 return 0;
8116 else if (TYPE_NAME (type1) == NULL && TYPE_NAME (type0) != NULL)
8117 return 1;
8118 else if (ada_is_constrained_packed_array_type (type0))
8119 return 1;
8120 else if (ada_is_array_descriptor_type (type0)
8121 && !ada_is_array_descriptor_type (type1))
8122 return 1;
8123 else
8124 {
8125 const char *type0_name = type_name_no_tag (type0);
8126 const char *type1_name = type_name_no_tag (type1);
8127
8128 if (type0_name != NULL && strstr (type0_name, "___XR") != NULL
8129 && (type1_name == NULL || strstr (type1_name, "___XR") == NULL))
8130 return 1;
8131 }
8132 return 0;
8133 }
8134
8135 /* The name of TYPE, which is either its TYPE_NAME, or, if that is
8136 null, its TYPE_TAG_NAME. Null if TYPE is null. */
8137
8138 const char *
8139 ada_type_name (struct type *type)
8140 {
8141 if (type == NULL)
8142 return NULL;
8143 else if (TYPE_NAME (type) != NULL)
8144 return TYPE_NAME (type);
8145 else
8146 return TYPE_TAG_NAME (type);
8147 }
8148
8149 /* Search the list of "descriptive" types associated to TYPE for a type
8150 whose name is NAME. */
8151
8152 static struct type *
8153 find_parallel_type_by_descriptive_type (struct type *type, const char *name)
8154 {
8155 struct type *result, *tmp;
8156
8157 if (ada_ignore_descriptive_types_p)
8158 return NULL;
8159
8160 /* If there no descriptive-type info, then there is no parallel type
8161 to be found. */
8162 if (!HAVE_GNAT_AUX_INFO (type))
8163 return NULL;
8164
8165 result = TYPE_DESCRIPTIVE_TYPE (type);
8166 while (result != NULL)
8167 {
8168 const char *result_name = ada_type_name (result);
8169
8170 if (result_name == NULL)
8171 {
8172 warning (_("unexpected null name on descriptive type"));
8173 return NULL;
8174 }
8175
8176 /* If the names match, stop. */
8177 if (strcmp (result_name, name) == 0)
8178 break;
8179
8180 /* Otherwise, look at the next item on the list, if any. */
8181 if (HAVE_GNAT_AUX_INFO (result))
8182 tmp = TYPE_DESCRIPTIVE_TYPE (result);
8183 else
8184 tmp = NULL;
8185
8186 /* If not found either, try after having resolved the typedef. */
8187 if (tmp != NULL)
8188 result = tmp;
8189 else
8190 {
8191 result = check_typedef (result);
8192 if (HAVE_GNAT_AUX_INFO (result))
8193 result = TYPE_DESCRIPTIVE_TYPE (result);
8194 else
8195 result = NULL;
8196 }
8197 }
8198
8199 /* If we didn't find a match, see whether this is a packed array. With
8200 older compilers, the descriptive type information is either absent or
8201 irrelevant when it comes to packed arrays so the above lookup fails.
8202 Fall back to using a parallel lookup by name in this case. */
8203 if (result == NULL && ada_is_constrained_packed_array_type (type))
8204 return ada_find_any_type (name);
8205
8206 return result;
8207 }
8208
8209 /* Find a parallel type to TYPE with the specified NAME, using the
8210 descriptive type taken from the debugging information, if available,
8211 and otherwise using the (slower) name-based method. */
8212
8213 static struct type *
8214 ada_find_parallel_type_with_name (struct type *type, const char *name)
8215 {
8216 struct type *result = NULL;
8217
8218 if (HAVE_GNAT_AUX_INFO (type))
8219 result = find_parallel_type_by_descriptive_type (type, name);
8220 else
8221 result = ada_find_any_type (name);
8222
8223 return result;
8224 }
8225
8226 /* Same as above, but specify the name of the parallel type by appending
8227 SUFFIX to the name of TYPE. */
8228
8229 struct type *
8230 ada_find_parallel_type (struct type *type, const char *suffix)
8231 {
8232 char *name;
8233 const char *type_name = ada_type_name (type);
8234 int len;
8235
8236 if (type_name == NULL)
8237 return NULL;
8238
8239 len = strlen (type_name);
8240
8241 name = (char *) alloca (len + strlen (suffix) + 1);
8242
8243 strcpy (name, type_name);
8244 strcpy (name + len, suffix);
8245
8246 return ada_find_parallel_type_with_name (type, name);
8247 }
8248
8249 /* If TYPE is a variable-size record type, return the corresponding template
8250 type describing its fields. Otherwise, return NULL. */
8251
8252 static struct type *
8253 dynamic_template_type (struct type *type)
8254 {
8255 type = ada_check_typedef (type);
8256
8257 if (type == NULL || TYPE_CODE (type) != TYPE_CODE_STRUCT
8258 || ada_type_name (type) == NULL)
8259 return NULL;
8260 else
8261 {
8262 int len = strlen (ada_type_name (type));
8263
8264 if (len > 6 && strcmp (ada_type_name (type) + len - 6, "___XVE") == 0)
8265 return type;
8266 else
8267 return ada_find_parallel_type (type, "___XVE");
8268 }
8269 }
8270
8271 /* Assuming that TEMPL_TYPE is a union or struct type, returns
8272 non-zero iff field FIELD_NUM of TEMPL_TYPE has dynamic size. */
8273
8274 static int
8275 is_dynamic_field (struct type *templ_type, int field_num)
8276 {
8277 const char *name = TYPE_FIELD_NAME (templ_type, field_num);
8278
8279 return name != NULL
8280 && TYPE_CODE (TYPE_FIELD_TYPE (templ_type, field_num)) == TYPE_CODE_PTR
8281 && strstr (name, "___XVL") != NULL;
8282 }
8283
8284 /* The index of the variant field of TYPE, or -1 if TYPE does not
8285 represent a variant record type. */
8286
8287 static int
8288 variant_field_index (struct type *type)
8289 {
8290 int f;
8291
8292 if (type == NULL || TYPE_CODE (type) != TYPE_CODE_STRUCT)
8293 return -1;
8294
8295 for (f = 0; f < TYPE_NFIELDS (type); f += 1)
8296 {
8297 if (ada_is_variant_part (type, f))
8298 return f;
8299 }
8300 return -1;
8301 }
8302
8303 /* A record type with no fields. */
8304
8305 static struct type *
8306 empty_record (struct type *templ)
8307 {
8308 struct type *type = alloc_type_copy (templ);
8309
8310 TYPE_CODE (type) = TYPE_CODE_STRUCT;
8311 TYPE_NFIELDS (type) = 0;
8312 TYPE_FIELDS (type) = NULL;
8313 INIT_CPLUS_SPECIFIC (type);
8314 TYPE_NAME (type) = "<empty>";
8315 TYPE_TAG_NAME (type) = NULL;
8316 TYPE_LENGTH (type) = 0;
8317 return type;
8318 }
8319
8320 /* An ordinary record type (with fixed-length fields) that describes
8321 the value of type TYPE at VALADDR or ADDRESS (see comments at
8322 the beginning of this section) VAL according to GNAT conventions.
8323 DVAL0 should describe the (portion of a) record that contains any
8324 necessary discriminants. It should be NULL if value_type (VAL) is
8325 an outer-level type (i.e., as opposed to a branch of a variant.) A
8326 variant field (unless unchecked) is replaced by a particular branch
8327 of the variant.
8328
8329 If not KEEP_DYNAMIC_FIELDS, then all fields whose position or
8330 length are not statically known are discarded. As a consequence,
8331 VALADDR, ADDRESS and DVAL0 are ignored.
8332
8333 NOTE: Limitations: For now, we assume that dynamic fields and
8334 variants occupy whole numbers of bytes. However, they need not be
8335 byte-aligned. */
8336
8337 struct type *
8338 ada_template_to_fixed_record_type_1 (struct type *type,
8339 const gdb_byte *valaddr,
8340 CORE_ADDR address, struct value *dval0,
8341 int keep_dynamic_fields)
8342 {
8343 struct value *mark = value_mark ();
8344 struct value *dval;
8345 struct type *rtype;
8346 int nfields, bit_len;
8347 int variant_field;
8348 long off;
8349 int fld_bit_len;
8350 int f;
8351
8352 /* Compute the number of fields in this record type that are going
8353 to be processed: unless keep_dynamic_fields, this includes only
8354 fields whose position and length are static will be processed. */
8355 if (keep_dynamic_fields)
8356 nfields = TYPE_NFIELDS (type);
8357 else
8358 {
8359 nfields = 0;
8360 while (nfields < TYPE_NFIELDS (type)
8361 && !ada_is_variant_part (type, nfields)
8362 && !is_dynamic_field (type, nfields))
8363 nfields++;
8364 }
8365
8366 rtype = alloc_type_copy (type);
8367 TYPE_CODE (rtype) = TYPE_CODE_STRUCT;
8368 INIT_CPLUS_SPECIFIC (rtype);
8369 TYPE_NFIELDS (rtype) = nfields;
8370 TYPE_FIELDS (rtype) = (struct field *)
8371 TYPE_ALLOC (rtype, nfields * sizeof (struct field));
8372 memset (TYPE_FIELDS (rtype), 0, sizeof (struct field) * nfields);
8373 TYPE_NAME (rtype) = ada_type_name (type);
8374 TYPE_TAG_NAME (rtype) = NULL;
8375 TYPE_FIXED_INSTANCE (rtype) = 1;
8376
8377 off = 0;
8378 bit_len = 0;
8379 variant_field = -1;
8380
8381 for (f = 0; f < nfields; f += 1)
8382 {
8383 off = align_value (off, field_alignment (type, f))
8384 + TYPE_FIELD_BITPOS (type, f);
8385 SET_FIELD_BITPOS (TYPE_FIELD (rtype, f), off);
8386 TYPE_FIELD_BITSIZE (rtype, f) = 0;
8387
8388 if (ada_is_variant_part (type, f))
8389 {
8390 variant_field = f;
8391 fld_bit_len = 0;
8392 }
8393 else if (is_dynamic_field (type, f))
8394 {
8395 const gdb_byte *field_valaddr = valaddr;
8396 CORE_ADDR field_address = address;
8397 struct type *field_type =
8398 TYPE_TARGET_TYPE (TYPE_FIELD_TYPE (type, f));
8399
8400 if (dval0 == NULL)
8401 {
8402 /* rtype's length is computed based on the run-time
8403 value of discriminants. If the discriminants are not
8404 initialized, the type size may be completely bogus and
8405 GDB may fail to allocate a value for it. So check the
8406 size first before creating the value. */
8407 ada_ensure_varsize_limit (rtype);
8408 /* Using plain value_from_contents_and_address here
8409 causes problems because we will end up trying to
8410 resolve a type that is currently being
8411 constructed. */
8412 dval = value_from_contents_and_address_unresolved (rtype,
8413 valaddr,
8414 address);
8415 rtype = value_type (dval);
8416 }
8417 else
8418 dval = dval0;
8419
8420 /* If the type referenced by this field is an aligner type, we need
8421 to unwrap that aligner type, because its size might not be set.
8422 Keeping the aligner type would cause us to compute the wrong
8423 size for this field, impacting the offset of the all the fields
8424 that follow this one. */
8425 if (ada_is_aligner_type (field_type))
8426 {
8427 long field_offset = TYPE_FIELD_BITPOS (field_type, f);
8428
8429 field_valaddr = cond_offset_host (field_valaddr, field_offset);
8430 field_address = cond_offset_target (field_address, field_offset);
8431 field_type = ada_aligned_type (field_type);
8432 }
8433
8434 field_valaddr = cond_offset_host (field_valaddr,
8435 off / TARGET_CHAR_BIT);
8436 field_address = cond_offset_target (field_address,
8437 off / TARGET_CHAR_BIT);
8438
8439 /* Get the fixed type of the field. Note that, in this case,
8440 we do not want to get the real type out of the tag: if
8441 the current field is the parent part of a tagged record,
8442 we will get the tag of the object. Clearly wrong: the real
8443 type of the parent is not the real type of the child. We
8444 would end up in an infinite loop. */
8445 field_type = ada_get_base_type (field_type);
8446 field_type = ada_to_fixed_type (field_type, field_valaddr,
8447 field_address, dval, 0);
8448 /* If the field size is already larger than the maximum
8449 object size, then the record itself will necessarily
8450 be larger than the maximum object size. We need to make
8451 this check now, because the size might be so ridiculously
8452 large (due to an uninitialized variable in the inferior)
8453 that it would cause an overflow when adding it to the
8454 record size. */
8455 ada_ensure_varsize_limit (field_type);
8456
8457 TYPE_FIELD_TYPE (rtype, f) = field_type;
8458 TYPE_FIELD_NAME (rtype, f) = TYPE_FIELD_NAME (type, f);
8459 /* The multiplication can potentially overflow. But because
8460 the field length has been size-checked just above, and
8461 assuming that the maximum size is a reasonable value,
8462 an overflow should not happen in practice. So rather than
8463 adding overflow recovery code to this already complex code,
8464 we just assume that it's not going to happen. */
8465 fld_bit_len =
8466 TYPE_LENGTH (TYPE_FIELD_TYPE (rtype, f)) * TARGET_CHAR_BIT;
8467 }
8468 else
8469 {
8470 /* Note: If this field's type is a typedef, it is important
8471 to preserve the typedef layer.
8472
8473 Otherwise, we might be transforming a typedef to a fat
8474 pointer (encoding a pointer to an unconstrained array),
8475 into a basic fat pointer (encoding an unconstrained
8476 array). As both types are implemented using the same
8477 structure, the typedef is the only clue which allows us
8478 to distinguish between the two options. Stripping it
8479 would prevent us from printing this field appropriately. */
8480 TYPE_FIELD_TYPE (rtype, f) = TYPE_FIELD_TYPE (type, f);
8481 TYPE_FIELD_NAME (rtype, f) = TYPE_FIELD_NAME (type, f);
8482 if (TYPE_FIELD_BITSIZE (type, f) > 0)
8483 fld_bit_len =
8484 TYPE_FIELD_BITSIZE (rtype, f) = TYPE_FIELD_BITSIZE (type, f);
8485 else
8486 {
8487 struct type *field_type = TYPE_FIELD_TYPE (type, f);
8488
8489 /* We need to be careful of typedefs when computing
8490 the length of our field. If this is a typedef,
8491 get the length of the target type, not the length
8492 of the typedef. */
8493 if (TYPE_CODE (field_type) == TYPE_CODE_TYPEDEF)
8494 field_type = ada_typedef_target_type (field_type);
8495
8496 fld_bit_len =
8497 TYPE_LENGTH (ada_check_typedef (field_type)) * TARGET_CHAR_BIT;
8498 }
8499 }
8500 if (off + fld_bit_len > bit_len)
8501 bit_len = off + fld_bit_len;
8502 off += fld_bit_len;
8503 TYPE_LENGTH (rtype) =
8504 align_value (bit_len, TARGET_CHAR_BIT) / TARGET_CHAR_BIT;
8505 }
8506
8507 /* We handle the variant part, if any, at the end because of certain
8508 odd cases in which it is re-ordered so as NOT to be the last field of
8509 the record. This can happen in the presence of representation
8510 clauses. */
8511 if (variant_field >= 0)
8512 {
8513 struct type *branch_type;
8514
8515 off = TYPE_FIELD_BITPOS (rtype, variant_field);
8516
8517 if (dval0 == NULL)
8518 {
8519 /* Using plain value_from_contents_and_address here causes
8520 problems because we will end up trying to resolve a type
8521 that is currently being constructed. */
8522 dval = value_from_contents_and_address_unresolved (rtype, valaddr,
8523 address);
8524 rtype = value_type (dval);
8525 }
8526 else
8527 dval = dval0;
8528
8529 branch_type =
8530 to_fixed_variant_branch_type
8531 (TYPE_FIELD_TYPE (type, variant_field),
8532 cond_offset_host (valaddr, off / TARGET_CHAR_BIT),
8533 cond_offset_target (address, off / TARGET_CHAR_BIT), dval);
8534 if (branch_type == NULL)
8535 {
8536 for (f = variant_field + 1; f < TYPE_NFIELDS (rtype); f += 1)
8537 TYPE_FIELDS (rtype)[f - 1] = TYPE_FIELDS (rtype)[f];
8538 TYPE_NFIELDS (rtype) -= 1;
8539 }
8540 else
8541 {
8542 TYPE_FIELD_TYPE (rtype, variant_field) = branch_type;
8543 TYPE_FIELD_NAME (rtype, variant_field) = "S";
8544 fld_bit_len =
8545 TYPE_LENGTH (TYPE_FIELD_TYPE (rtype, variant_field)) *
8546 TARGET_CHAR_BIT;
8547 if (off + fld_bit_len > bit_len)
8548 bit_len = off + fld_bit_len;
8549 TYPE_LENGTH (rtype) =
8550 align_value (bit_len, TARGET_CHAR_BIT) / TARGET_CHAR_BIT;
8551 }
8552 }
8553
8554 /* According to exp_dbug.ads, the size of TYPE for variable-size records
8555 should contain the alignment of that record, which should be a strictly
8556 positive value. If null or negative, then something is wrong, most
8557 probably in the debug info. In that case, we don't round up the size
8558 of the resulting type. If this record is not part of another structure,
8559 the current RTYPE length might be good enough for our purposes. */
8560 if (TYPE_LENGTH (type) <= 0)
8561 {
8562 if (TYPE_NAME (rtype))
8563 warning (_("Invalid type size for `%s' detected: %d."),
8564 TYPE_NAME (rtype), TYPE_LENGTH (type));
8565 else
8566 warning (_("Invalid type size for <unnamed> detected: %d."),
8567 TYPE_LENGTH (type));
8568 }
8569 else
8570 {
8571 TYPE_LENGTH (rtype) = align_value (TYPE_LENGTH (rtype),
8572 TYPE_LENGTH (type));
8573 }
8574
8575 value_free_to_mark (mark);
8576 if (TYPE_LENGTH (rtype) > varsize_limit)
8577 error (_("record type with dynamic size is larger than varsize-limit"));
8578 return rtype;
8579 }
8580
8581 /* As for ada_template_to_fixed_record_type_1 with KEEP_DYNAMIC_FIELDS
8582 of 1. */
8583
8584 static struct type *
8585 template_to_fixed_record_type (struct type *type, const gdb_byte *valaddr,
8586 CORE_ADDR address, struct value *dval0)
8587 {
8588 return ada_template_to_fixed_record_type_1 (type, valaddr,
8589 address, dval0, 1);
8590 }
8591
8592 /* An ordinary record type in which ___XVL-convention fields and
8593 ___XVU- and ___XVN-convention field types in TYPE0 are replaced with
8594 static approximations, containing all possible fields. Uses
8595 no runtime values. Useless for use in values, but that's OK,
8596 since the results are used only for type determinations. Works on both
8597 structs and unions. Representation note: to save space, we memorize
8598 the result of this function in the TYPE_TARGET_TYPE of the
8599 template type. */
8600
8601 static struct type *
8602 template_to_static_fixed_type (struct type *type0)
8603 {
8604 struct type *type;
8605 int nfields;
8606 int f;
8607
8608 /* No need no do anything if the input type is already fixed. */
8609 if (TYPE_FIXED_INSTANCE (type0))
8610 return type0;
8611
8612 /* Likewise if we already have computed the static approximation. */
8613 if (TYPE_TARGET_TYPE (type0) != NULL)
8614 return TYPE_TARGET_TYPE (type0);
8615
8616 /* Don't clone TYPE0 until we are sure we are going to need a copy. */
8617 type = type0;
8618 nfields = TYPE_NFIELDS (type0);
8619
8620 /* Whether or not we cloned TYPE0, cache the result so that we don't do
8621 recompute all over next time. */
8622 TYPE_TARGET_TYPE (type0) = type;
8623
8624 for (f = 0; f < nfields; f += 1)
8625 {
8626 struct type *field_type = TYPE_FIELD_TYPE (type0, f);
8627 struct type *new_type;
8628
8629 if (is_dynamic_field (type0, f))
8630 {
8631 field_type = ada_check_typedef (field_type);
8632 new_type = to_static_fixed_type (TYPE_TARGET_TYPE (field_type));
8633 }
8634 else
8635 new_type = static_unwrap_type (field_type);
8636
8637 if (new_type != field_type)
8638 {
8639 /* Clone TYPE0 only the first time we get a new field type. */
8640 if (type == type0)
8641 {
8642 TYPE_TARGET_TYPE (type0) = type = alloc_type_copy (type0);
8643 TYPE_CODE (type) = TYPE_CODE (type0);
8644 INIT_CPLUS_SPECIFIC (type);
8645 TYPE_NFIELDS (type) = nfields;
8646 TYPE_FIELDS (type) = (struct field *)
8647 TYPE_ALLOC (type, nfields * sizeof (struct field));
8648 memcpy (TYPE_FIELDS (type), TYPE_FIELDS (type0),
8649 sizeof (struct field) * nfields);
8650 TYPE_NAME (type) = ada_type_name (type0);
8651 TYPE_TAG_NAME (type) = NULL;
8652 TYPE_FIXED_INSTANCE (type) = 1;
8653 TYPE_LENGTH (type) = 0;
8654 }
8655 TYPE_FIELD_TYPE (type, f) = new_type;
8656 TYPE_FIELD_NAME (type, f) = TYPE_FIELD_NAME (type0, f);
8657 }
8658 }
8659
8660 return type;
8661 }
8662
8663 /* Given an object of type TYPE whose contents are at VALADDR and
8664 whose address in memory is ADDRESS, returns a revision of TYPE,
8665 which should be a non-dynamic-sized record, in which the variant
8666 part, if any, is replaced with the appropriate branch. Looks
8667 for discriminant values in DVAL0, which can be NULL if the record
8668 contains the necessary discriminant values. */
8669
8670 static struct type *
8671 to_record_with_fixed_variant_part (struct type *type, const gdb_byte *valaddr,
8672 CORE_ADDR address, struct value *dval0)
8673 {
8674 struct value *mark = value_mark ();
8675 struct value *dval;
8676 struct type *rtype;
8677 struct type *branch_type;
8678 int nfields = TYPE_NFIELDS (type);
8679 int variant_field = variant_field_index (type);
8680
8681 if (variant_field == -1)
8682 return type;
8683
8684 if (dval0 == NULL)
8685 {
8686 dval = value_from_contents_and_address (type, valaddr, address);
8687 type = value_type (dval);
8688 }
8689 else
8690 dval = dval0;
8691
8692 rtype = alloc_type_copy (type);
8693 TYPE_CODE (rtype) = TYPE_CODE_STRUCT;
8694 INIT_CPLUS_SPECIFIC (rtype);
8695 TYPE_NFIELDS (rtype) = nfields;
8696 TYPE_FIELDS (rtype) =
8697 (struct field *) TYPE_ALLOC (rtype, nfields * sizeof (struct field));
8698 memcpy (TYPE_FIELDS (rtype), TYPE_FIELDS (type),
8699 sizeof (struct field) * nfields);
8700 TYPE_NAME (rtype) = ada_type_name (type);
8701 TYPE_TAG_NAME (rtype) = NULL;
8702 TYPE_FIXED_INSTANCE (rtype) = 1;
8703 TYPE_LENGTH (rtype) = TYPE_LENGTH (type);
8704
8705 branch_type = to_fixed_variant_branch_type
8706 (TYPE_FIELD_TYPE (type, variant_field),
8707 cond_offset_host (valaddr,
8708 TYPE_FIELD_BITPOS (type, variant_field)
8709 / TARGET_CHAR_BIT),
8710 cond_offset_target (address,
8711 TYPE_FIELD_BITPOS (type, variant_field)
8712 / TARGET_CHAR_BIT), dval);
8713 if (branch_type == NULL)
8714 {
8715 int f;
8716
8717 for (f = variant_field + 1; f < nfields; f += 1)
8718 TYPE_FIELDS (rtype)[f - 1] = TYPE_FIELDS (rtype)[f];
8719 TYPE_NFIELDS (rtype) -= 1;
8720 }
8721 else
8722 {
8723 TYPE_FIELD_TYPE (rtype, variant_field) = branch_type;
8724 TYPE_FIELD_NAME (rtype, variant_field) = "S";
8725 TYPE_FIELD_BITSIZE (rtype, variant_field) = 0;
8726 TYPE_LENGTH (rtype) += TYPE_LENGTH (branch_type);
8727 }
8728 TYPE_LENGTH (rtype) -= TYPE_LENGTH (TYPE_FIELD_TYPE (type, variant_field));
8729
8730 value_free_to_mark (mark);
8731 return rtype;
8732 }
8733
8734 /* An ordinary record type (with fixed-length fields) that describes
8735 the value at (TYPE0, VALADDR, ADDRESS) [see explanation at
8736 beginning of this section]. Any necessary discriminants' values
8737 should be in DVAL, a record value; it may be NULL if the object
8738 at ADDR itself contains any necessary discriminant values.
8739 Additionally, VALADDR and ADDRESS may also be NULL if no discriminant
8740 values from the record are needed. Except in the case that DVAL,
8741 VALADDR, and ADDRESS are all 0 or NULL, a variant field (unless
8742 unchecked) is replaced by a particular branch of the variant.
8743
8744 NOTE: the case in which DVAL and VALADDR are NULL and ADDRESS is 0
8745 is questionable and may be removed. It can arise during the
8746 processing of an unconstrained-array-of-record type where all the
8747 variant branches have exactly the same size. This is because in
8748 such cases, the compiler does not bother to use the XVS convention
8749 when encoding the record. I am currently dubious of this
8750 shortcut and suspect the compiler should be altered. FIXME. */
8751
8752 static struct type *
8753 to_fixed_record_type (struct type *type0, const gdb_byte *valaddr,
8754 CORE_ADDR address, struct value *dval)
8755 {
8756 struct type *templ_type;
8757
8758 if (TYPE_FIXED_INSTANCE (type0))
8759 return type0;
8760
8761 templ_type = dynamic_template_type (type0);
8762
8763 if (templ_type != NULL)
8764 return template_to_fixed_record_type (templ_type, valaddr, address, dval);
8765 else if (variant_field_index (type0) >= 0)
8766 {
8767 if (dval == NULL && valaddr == NULL && address == 0)
8768 return type0;
8769 return to_record_with_fixed_variant_part (type0, valaddr, address,
8770 dval);
8771 }
8772 else
8773 {
8774 TYPE_FIXED_INSTANCE (type0) = 1;
8775 return type0;
8776 }
8777
8778 }
8779
8780 /* An ordinary record type (with fixed-length fields) that describes
8781 the value at (VAR_TYPE0, VALADDR, ADDRESS), where VAR_TYPE0 is a
8782 union type. Any necessary discriminants' values should be in DVAL,
8783 a record value. That is, this routine selects the appropriate
8784 branch of the union at ADDR according to the discriminant value
8785 indicated in the union's type name. Returns VAR_TYPE0 itself if
8786 it represents a variant subject to a pragma Unchecked_Union. */
8787
8788 static struct type *
8789 to_fixed_variant_branch_type (struct type *var_type0, const gdb_byte *valaddr,
8790 CORE_ADDR address, struct value *dval)
8791 {
8792 int which;
8793 struct type *templ_type;
8794 struct type *var_type;
8795
8796 if (TYPE_CODE (var_type0) == TYPE_CODE_PTR)
8797 var_type = TYPE_TARGET_TYPE (var_type0);
8798 else
8799 var_type = var_type0;
8800
8801 templ_type = ada_find_parallel_type (var_type, "___XVU");
8802
8803 if (templ_type != NULL)
8804 var_type = templ_type;
8805
8806 if (is_unchecked_variant (var_type, value_type (dval)))
8807 return var_type0;
8808 which =
8809 ada_which_variant_applies (var_type,
8810 value_type (dval), value_contents (dval));
8811
8812 if (which < 0)
8813 return empty_record (var_type);
8814 else if (is_dynamic_field (var_type, which))
8815 return to_fixed_record_type
8816 (TYPE_TARGET_TYPE (TYPE_FIELD_TYPE (var_type, which)),
8817 valaddr, address, dval);
8818 else if (variant_field_index (TYPE_FIELD_TYPE (var_type, which)) >= 0)
8819 return
8820 to_fixed_record_type
8821 (TYPE_FIELD_TYPE (var_type, which), valaddr, address, dval);
8822 else
8823 return TYPE_FIELD_TYPE (var_type, which);
8824 }
8825
8826 /* Assuming RANGE_TYPE is a TYPE_CODE_RANGE, return nonzero if
8827 ENCODING_TYPE, a type following the GNAT conventions for discrete
8828 type encodings, only carries redundant information. */
8829
8830 static int
8831 ada_is_redundant_range_encoding (struct type *range_type,
8832 struct type *encoding_type)
8833 {
8834 const char *bounds_str;
8835 int n;
8836 LONGEST lo, hi;
8837
8838 gdb_assert (TYPE_CODE (range_type) == TYPE_CODE_RANGE);
8839
8840 if (TYPE_CODE (get_base_type (range_type))
8841 != TYPE_CODE (get_base_type (encoding_type)))
8842 {
8843 /* The compiler probably used a simple base type to describe
8844 the range type instead of the range's actual base type,
8845 expecting us to get the real base type from the encoding
8846 anyway. In this situation, the encoding cannot be ignored
8847 as redundant. */
8848 return 0;
8849 }
8850
8851 if (is_dynamic_type (range_type))
8852 return 0;
8853
8854 if (TYPE_NAME (encoding_type) == NULL)
8855 return 0;
8856
8857 bounds_str = strstr (TYPE_NAME (encoding_type), "___XDLU_");
8858 if (bounds_str == NULL)
8859 return 0;
8860
8861 n = 8; /* Skip "___XDLU_". */
8862 if (!ada_scan_number (bounds_str, n, &lo, &n))
8863 return 0;
8864 if (TYPE_LOW_BOUND (range_type) != lo)
8865 return 0;
8866
8867 n += 2; /* Skip the "__" separator between the two bounds. */
8868 if (!ada_scan_number (bounds_str, n, &hi, &n))
8869 return 0;
8870 if (TYPE_HIGH_BOUND (range_type) != hi)
8871 return 0;
8872
8873 return 1;
8874 }
8875
8876 /* Given the array type ARRAY_TYPE, return nonzero if DESC_TYPE,
8877 a type following the GNAT encoding for describing array type
8878 indices, only carries redundant information. */
8879
8880 static int
8881 ada_is_redundant_index_type_desc (struct type *array_type,
8882 struct type *desc_type)
8883 {
8884 struct type *this_layer = check_typedef (array_type);
8885 int i;
8886
8887 for (i = 0; i < TYPE_NFIELDS (desc_type); i++)
8888 {
8889 if (!ada_is_redundant_range_encoding (TYPE_INDEX_TYPE (this_layer),
8890 TYPE_FIELD_TYPE (desc_type, i)))
8891 return 0;
8892 this_layer = check_typedef (TYPE_TARGET_TYPE (this_layer));
8893 }
8894
8895 return 1;
8896 }
8897
8898 /* Assuming that TYPE0 is an array type describing the type of a value
8899 at ADDR, and that DVAL describes a record containing any
8900 discriminants used in TYPE0, returns a type for the value that
8901 contains no dynamic components (that is, no components whose sizes
8902 are determined by run-time quantities). Unless IGNORE_TOO_BIG is
8903 true, gives an error message if the resulting type's size is over
8904 varsize_limit. */
8905
8906 static struct type *
8907 to_fixed_array_type (struct type *type0, struct value *dval,
8908 int ignore_too_big)
8909 {
8910 struct type *index_type_desc;
8911 struct type *result;
8912 int constrained_packed_array_p;
8913 static const char *xa_suffix = "___XA";
8914
8915 type0 = ada_check_typedef (type0);
8916 if (TYPE_FIXED_INSTANCE (type0))
8917 return type0;
8918
8919 constrained_packed_array_p = ada_is_constrained_packed_array_type (type0);
8920 if (constrained_packed_array_p)
8921 type0 = decode_constrained_packed_array_type (type0);
8922
8923 index_type_desc = ada_find_parallel_type (type0, xa_suffix);
8924
8925 /* As mentioned in exp_dbug.ads, for non bit-packed arrays an
8926 encoding suffixed with 'P' may still be generated. If so,
8927 it should be used to find the XA type. */
8928
8929 if (index_type_desc == NULL)
8930 {
8931 const char *type_name = ada_type_name (type0);
8932
8933 if (type_name != NULL)
8934 {
8935 const int len = strlen (type_name);
8936 char *name = (char *) alloca (len + strlen (xa_suffix));
8937
8938 if (type_name[len - 1] == 'P')
8939 {
8940 strcpy (name, type_name);
8941 strcpy (name + len - 1, xa_suffix);
8942 index_type_desc = ada_find_parallel_type_with_name (type0, name);
8943 }
8944 }
8945 }
8946
8947 ada_fixup_array_indexes_type (index_type_desc);
8948 if (index_type_desc != NULL
8949 && ada_is_redundant_index_type_desc (type0, index_type_desc))
8950 {
8951 /* Ignore this ___XA parallel type, as it does not bring any
8952 useful information. This allows us to avoid creating fixed
8953 versions of the array's index types, which would be identical
8954 to the original ones. This, in turn, can also help avoid
8955 the creation of fixed versions of the array itself. */
8956 index_type_desc = NULL;
8957 }
8958
8959 if (index_type_desc == NULL)
8960 {
8961 struct type *elt_type0 = ada_check_typedef (TYPE_TARGET_TYPE (type0));
8962
8963 /* NOTE: elt_type---the fixed version of elt_type0---should never
8964 depend on the contents of the array in properly constructed
8965 debugging data. */
8966 /* Create a fixed version of the array element type.
8967 We're not providing the address of an element here,
8968 and thus the actual object value cannot be inspected to do
8969 the conversion. This should not be a problem, since arrays of
8970 unconstrained objects are not allowed. In particular, all
8971 the elements of an array of a tagged type should all be of
8972 the same type specified in the debugging info. No need to
8973 consult the object tag. */
8974 struct type *elt_type = ada_to_fixed_type (elt_type0, 0, 0, dval, 1);
8975
8976 /* Make sure we always create a new array type when dealing with
8977 packed array types, since we're going to fix-up the array
8978 type length and element bitsize a little further down. */
8979 if (elt_type0 == elt_type && !constrained_packed_array_p)
8980 result = type0;
8981 else
8982 result = create_array_type (alloc_type_copy (type0),
8983 elt_type, TYPE_INDEX_TYPE (type0));
8984 }
8985 else
8986 {
8987 int i;
8988 struct type *elt_type0;
8989
8990 elt_type0 = type0;
8991 for (i = TYPE_NFIELDS (index_type_desc); i > 0; i -= 1)
8992 elt_type0 = TYPE_TARGET_TYPE (elt_type0);
8993
8994 /* NOTE: result---the fixed version of elt_type0---should never
8995 depend on the contents of the array in properly constructed
8996 debugging data. */
8997 /* Create a fixed version of the array element type.
8998 We're not providing the address of an element here,
8999 and thus the actual object value cannot be inspected to do
9000 the conversion. This should not be a problem, since arrays of
9001 unconstrained objects are not allowed. In particular, all
9002 the elements of an array of a tagged type should all be of
9003 the same type specified in the debugging info. No need to
9004 consult the object tag. */
9005 result =
9006 ada_to_fixed_type (ada_check_typedef (elt_type0), 0, 0, dval, 1);
9007
9008 elt_type0 = type0;
9009 for (i = TYPE_NFIELDS (index_type_desc) - 1; i >= 0; i -= 1)
9010 {
9011 struct type *range_type =
9012 to_fixed_range_type (TYPE_FIELD_TYPE (index_type_desc, i), dval);
9013
9014 result = create_array_type (alloc_type_copy (elt_type0),
9015 result, range_type);
9016 elt_type0 = TYPE_TARGET_TYPE (elt_type0);
9017 }
9018 if (!ignore_too_big && TYPE_LENGTH (result) > varsize_limit)
9019 error (_("array type with dynamic size is larger than varsize-limit"));
9020 }
9021
9022 /* We want to preserve the type name. This can be useful when
9023 trying to get the type name of a value that has already been
9024 printed (for instance, if the user did "print VAR; whatis $". */
9025 TYPE_NAME (result) = TYPE_NAME (type0);
9026
9027 if (constrained_packed_array_p)
9028 {
9029 /* So far, the resulting type has been created as if the original
9030 type was a regular (non-packed) array type. As a result, the
9031 bitsize of the array elements needs to be set again, and the array
9032 length needs to be recomputed based on that bitsize. */
9033 int len = TYPE_LENGTH (result) / TYPE_LENGTH (TYPE_TARGET_TYPE (result));
9034 int elt_bitsize = TYPE_FIELD_BITSIZE (type0, 0);
9035
9036 TYPE_FIELD_BITSIZE (result, 0) = TYPE_FIELD_BITSIZE (type0, 0);
9037 TYPE_LENGTH (result) = len * elt_bitsize / HOST_CHAR_BIT;
9038 if (TYPE_LENGTH (result) * HOST_CHAR_BIT < len * elt_bitsize)
9039 TYPE_LENGTH (result)++;
9040 }
9041
9042 TYPE_FIXED_INSTANCE (result) = 1;
9043 return result;
9044 }
9045
9046
9047 /* A standard type (containing no dynamically sized components)
9048 corresponding to TYPE for the value (TYPE, VALADDR, ADDRESS)
9049 DVAL describes a record containing any discriminants used in TYPE0,
9050 and may be NULL if there are none, or if the object of type TYPE at
9051 ADDRESS or in VALADDR contains these discriminants.
9052
9053 If CHECK_TAG is not null, in the case of tagged types, this function
9054 attempts to locate the object's tag and use it to compute the actual
9055 type. However, when ADDRESS is null, we cannot use it to determine the
9056 location of the tag, and therefore compute the tagged type's actual type.
9057 So we return the tagged type without consulting the tag. */
9058
9059 static struct type *
9060 ada_to_fixed_type_1 (struct type *type, const gdb_byte *valaddr,
9061 CORE_ADDR address, struct value *dval, int check_tag)
9062 {
9063 type = ada_check_typedef (type);
9064 switch (TYPE_CODE (type))
9065 {
9066 default:
9067 return type;
9068 case TYPE_CODE_STRUCT:
9069 {
9070 struct type *static_type = to_static_fixed_type (type);
9071 struct type *fixed_record_type =
9072 to_fixed_record_type (type, valaddr, address, NULL);
9073
9074 /* If STATIC_TYPE is a tagged type and we know the object's address,
9075 then we can determine its tag, and compute the object's actual
9076 type from there. Note that we have to use the fixed record
9077 type (the parent part of the record may have dynamic fields
9078 and the way the location of _tag is expressed may depend on
9079 them). */
9080
9081 if (check_tag && address != 0 && ada_is_tagged_type (static_type, 0))
9082 {
9083 struct value *tag =
9084 value_tag_from_contents_and_address
9085 (fixed_record_type,
9086 valaddr,
9087 address);
9088 struct type *real_type = type_from_tag (tag);
9089 struct value *obj =
9090 value_from_contents_and_address (fixed_record_type,
9091 valaddr,
9092 address);
9093 fixed_record_type = value_type (obj);
9094 if (real_type != NULL)
9095 return to_fixed_record_type
9096 (real_type, NULL,
9097 value_address (ada_tag_value_at_base_address (obj)), NULL);
9098 }
9099
9100 /* Check to see if there is a parallel ___XVZ variable.
9101 If there is, then it provides the actual size of our type. */
9102 else if (ada_type_name (fixed_record_type) != NULL)
9103 {
9104 const char *name = ada_type_name (fixed_record_type);
9105 char *xvz_name
9106 = (char *) alloca (strlen (name) + 7 /* "___XVZ\0" */);
9107 bool xvz_found = false;
9108 LONGEST size;
9109
9110 xsnprintf (xvz_name, strlen (name) + 7, "%s___XVZ", name);
9111 TRY
9112 {
9113 xvz_found = get_int_var_value (xvz_name, size);
9114 }
9115 CATCH (except, RETURN_MASK_ERROR)
9116 {
9117 /* We found the variable, but somehow failed to read
9118 its value. Rethrow the same error, but with a little
9119 bit more information, to help the user understand
9120 what went wrong (Eg: the variable might have been
9121 optimized out). */
9122 throw_error (except.error,
9123 _("unable to read value of %s (%s)"),
9124 xvz_name, except.message);
9125 }
9126 END_CATCH
9127
9128 if (xvz_found && TYPE_LENGTH (fixed_record_type) != size)
9129 {
9130 fixed_record_type = copy_type (fixed_record_type);
9131 TYPE_LENGTH (fixed_record_type) = size;
9132
9133 /* The FIXED_RECORD_TYPE may have be a stub. We have
9134 observed this when the debugging info is STABS, and
9135 apparently it is something that is hard to fix.
9136
9137 In practice, we don't need the actual type definition
9138 at all, because the presence of the XVZ variable allows us
9139 to assume that there must be a XVS type as well, which we
9140 should be able to use later, when we need the actual type
9141 definition.
9142
9143 In the meantime, pretend that the "fixed" type we are
9144 returning is NOT a stub, because this can cause trouble
9145 when using this type to create new types targeting it.
9146 Indeed, the associated creation routines often check
9147 whether the target type is a stub and will try to replace
9148 it, thus using a type with the wrong size. This, in turn,
9149 might cause the new type to have the wrong size too.
9150 Consider the case of an array, for instance, where the size
9151 of the array is computed from the number of elements in
9152 our array multiplied by the size of its element. */
9153 TYPE_STUB (fixed_record_type) = 0;
9154 }
9155 }
9156 return fixed_record_type;
9157 }
9158 case TYPE_CODE_ARRAY:
9159 return to_fixed_array_type (type, dval, 1);
9160 case TYPE_CODE_UNION:
9161 if (dval == NULL)
9162 return type;
9163 else
9164 return to_fixed_variant_branch_type (type, valaddr, address, dval);
9165 }
9166 }
9167
9168 /* The same as ada_to_fixed_type_1, except that it preserves the type
9169 if it is a TYPE_CODE_TYPEDEF of a type that is already fixed.
9170
9171 The typedef layer needs be preserved in order to differentiate between
9172 arrays and array pointers when both types are implemented using the same
9173 fat pointer. In the array pointer case, the pointer is encoded as
9174 a typedef of the pointer type. For instance, considering:
9175
9176 type String_Access is access String;
9177 S1 : String_Access := null;
9178
9179 To the debugger, S1 is defined as a typedef of type String. But
9180 to the user, it is a pointer. So if the user tries to print S1,
9181 we should not dereference the array, but print the array address
9182 instead.
9183
9184 If we didn't preserve the typedef layer, we would lose the fact that
9185 the type is to be presented as a pointer (needs de-reference before
9186 being printed). And we would also use the source-level type name. */
9187
9188 struct type *
9189 ada_to_fixed_type (struct type *type, const gdb_byte *valaddr,
9190 CORE_ADDR address, struct value *dval, int check_tag)
9191
9192 {
9193 struct type *fixed_type =
9194 ada_to_fixed_type_1 (type, valaddr, address, dval, check_tag);
9195
9196 /* If TYPE is a typedef and its target type is the same as the FIXED_TYPE,
9197 then preserve the typedef layer.
9198
9199 Implementation note: We can only check the main-type portion of
9200 the TYPE and FIXED_TYPE, because eliminating the typedef layer
9201 from TYPE now returns a type that has the same instance flags
9202 as TYPE. For instance, if TYPE is a "typedef const", and its
9203 target type is a "struct", then the typedef elimination will return
9204 a "const" version of the target type. See check_typedef for more
9205 details about how the typedef layer elimination is done.
9206
9207 brobecker/2010-11-19: It seems to me that the only case where it is
9208 useful to preserve the typedef layer is when dealing with fat pointers.
9209 Perhaps, we could add a check for that and preserve the typedef layer
9210 only in that situation. But this seems unecessary so far, probably
9211 because we call check_typedef/ada_check_typedef pretty much everywhere.
9212 */
9213 if (TYPE_CODE (type) == TYPE_CODE_TYPEDEF
9214 && (TYPE_MAIN_TYPE (ada_typedef_target_type (type))
9215 == TYPE_MAIN_TYPE (fixed_type)))
9216 return type;
9217
9218 return fixed_type;
9219 }
9220
9221 /* A standard (static-sized) type corresponding as well as possible to
9222 TYPE0, but based on no runtime data. */
9223
9224 static struct type *
9225 to_static_fixed_type (struct type *type0)
9226 {
9227 struct type *type;
9228
9229 if (type0 == NULL)
9230 return NULL;
9231
9232 if (TYPE_FIXED_INSTANCE (type0))
9233 return type0;
9234
9235 type0 = ada_check_typedef (type0);
9236
9237 switch (TYPE_CODE (type0))
9238 {
9239 default:
9240 return type0;
9241 case TYPE_CODE_STRUCT:
9242 type = dynamic_template_type (type0);
9243 if (type != NULL)
9244 return template_to_static_fixed_type (type);
9245 else
9246 return template_to_static_fixed_type (type0);
9247 case TYPE_CODE_UNION:
9248 type = ada_find_parallel_type (type0, "___XVU");
9249 if (type != NULL)
9250 return template_to_static_fixed_type (type);
9251 else
9252 return template_to_static_fixed_type (type0);
9253 }
9254 }
9255
9256 /* A static approximation of TYPE with all type wrappers removed. */
9257
9258 static struct type *
9259 static_unwrap_type (struct type *type)
9260 {
9261 if (ada_is_aligner_type (type))
9262 {
9263 struct type *type1 = TYPE_FIELD_TYPE (ada_check_typedef (type), 0);
9264 if (ada_type_name (type1) == NULL)
9265 TYPE_NAME (type1) = ada_type_name (type);
9266
9267 return static_unwrap_type (type1);
9268 }
9269 else
9270 {
9271 struct type *raw_real_type = ada_get_base_type (type);
9272
9273 if (raw_real_type == type)
9274 return type;
9275 else
9276 return to_static_fixed_type (raw_real_type);
9277 }
9278 }
9279
9280 /* In some cases, incomplete and private types require
9281 cross-references that are not resolved as records (for example,
9282 type Foo;
9283 type FooP is access Foo;
9284 V: FooP;
9285 type Foo is array ...;
9286 ). In these cases, since there is no mechanism for producing
9287 cross-references to such types, we instead substitute for FooP a
9288 stub enumeration type that is nowhere resolved, and whose tag is
9289 the name of the actual type. Call these types "non-record stubs". */
9290
9291 /* A type equivalent to TYPE that is not a non-record stub, if one
9292 exists, otherwise TYPE. */
9293
9294 struct type *
9295 ada_check_typedef (struct type *type)
9296 {
9297 if (type == NULL)
9298 return NULL;
9299
9300 /* If our type is a typedef type of a fat pointer, then we're done.
9301 We don't want to strip the TYPE_CODE_TYPDEF layer, because this is
9302 what allows us to distinguish between fat pointers that represent
9303 array types, and fat pointers that represent array access types
9304 (in both cases, the compiler implements them as fat pointers). */
9305 if (TYPE_CODE (type) == TYPE_CODE_TYPEDEF
9306 && is_thick_pntr (ada_typedef_target_type (type)))
9307 return type;
9308
9309 type = check_typedef (type);
9310 if (type == NULL || TYPE_CODE (type) != TYPE_CODE_ENUM
9311 || !TYPE_STUB (type)
9312 || TYPE_TAG_NAME (type) == NULL)
9313 return type;
9314 else
9315 {
9316 const char *name = TYPE_TAG_NAME (type);
9317 struct type *type1 = ada_find_any_type (name);
9318
9319 if (type1 == NULL)
9320 return type;
9321
9322 /* TYPE1 might itself be a TYPE_CODE_TYPEDEF (this can happen with
9323 stubs pointing to arrays, as we don't create symbols for array
9324 types, only for the typedef-to-array types). If that's the case,
9325 strip the typedef layer. */
9326 if (TYPE_CODE (type1) == TYPE_CODE_TYPEDEF)
9327 type1 = ada_check_typedef (type1);
9328
9329 return type1;
9330 }
9331 }
9332
9333 /* A value representing the data at VALADDR/ADDRESS as described by
9334 type TYPE0, but with a standard (static-sized) type that correctly
9335 describes it. If VAL0 is not NULL and TYPE0 already is a standard
9336 type, then return VAL0 [this feature is simply to avoid redundant
9337 creation of struct values]. */
9338
9339 static struct value *
9340 ada_to_fixed_value_create (struct type *type0, CORE_ADDR address,
9341 struct value *val0)
9342 {
9343 struct type *type = ada_to_fixed_type (type0, 0, address, NULL, 1);
9344
9345 if (type == type0 && val0 != NULL)
9346 return val0;
9347
9348 if (VALUE_LVAL (val0) != lval_memory)
9349 {
9350 /* Our value does not live in memory; it could be a convenience
9351 variable, for instance. Create a not_lval value using val0's
9352 contents. */
9353 return value_from_contents (type, value_contents (val0));
9354 }
9355
9356 return value_from_contents_and_address (type, 0, address);
9357 }
9358
9359 /* A value representing VAL, but with a standard (static-sized) type
9360 that correctly describes it. Does not necessarily create a new
9361 value. */
9362
9363 struct value *
9364 ada_to_fixed_value (struct value *val)
9365 {
9366 val = unwrap_value (val);
9367 val = ada_to_fixed_value_create (value_type (val),
9368 value_address (val),
9369 val);
9370 return val;
9371 }
9372 \f
9373
9374 /* Attributes */
9375
9376 /* Table mapping attribute numbers to names.
9377 NOTE: Keep up to date with enum ada_attribute definition in ada-lang.h. */
9378
9379 static const char *attribute_names[] = {
9380 "<?>",
9381
9382 "first",
9383 "last",
9384 "length",
9385 "image",
9386 "max",
9387 "min",
9388 "modulus",
9389 "pos",
9390 "size",
9391 "tag",
9392 "val",
9393 0
9394 };
9395
9396 const char *
9397 ada_attribute_name (enum exp_opcode n)
9398 {
9399 if (n >= OP_ATR_FIRST && n <= (int) OP_ATR_VAL)
9400 return attribute_names[n - OP_ATR_FIRST + 1];
9401 else
9402 return attribute_names[0];
9403 }
9404
9405 /* Evaluate the 'POS attribute applied to ARG. */
9406
9407 static LONGEST
9408 pos_atr (struct value *arg)
9409 {
9410 struct value *val = coerce_ref (arg);
9411 struct type *type = value_type (val);
9412 LONGEST result;
9413
9414 if (!discrete_type_p (type))
9415 error (_("'POS only defined on discrete types"));
9416
9417 if (!discrete_position (type, value_as_long (val), &result))
9418 error (_("enumeration value is invalid: can't find 'POS"));
9419
9420 return result;
9421 }
9422
9423 static struct value *
9424 value_pos_atr (struct type *type, struct value *arg)
9425 {
9426 return value_from_longest (type, pos_atr (arg));
9427 }
9428
9429 /* Evaluate the TYPE'VAL attribute applied to ARG. */
9430
9431 static struct value *
9432 value_val_atr (struct type *type, struct value *arg)
9433 {
9434 if (!discrete_type_p (type))
9435 error (_("'VAL only defined on discrete types"));
9436 if (!integer_type_p (value_type (arg)))
9437 error (_("'VAL requires integral argument"));
9438
9439 if (TYPE_CODE (type) == TYPE_CODE_ENUM)
9440 {
9441 long pos = value_as_long (arg);
9442
9443 if (pos < 0 || pos >= TYPE_NFIELDS (type))
9444 error (_("argument to 'VAL out of range"));
9445 return value_from_longest (type, TYPE_FIELD_ENUMVAL (type, pos));
9446 }
9447 else
9448 return value_from_longest (type, value_as_long (arg));
9449 }
9450 \f
9451
9452 /* Evaluation */
9453
9454 /* True if TYPE appears to be an Ada character type.
9455 [At the moment, this is true only for Character and Wide_Character;
9456 It is a heuristic test that could stand improvement]. */
9457
9458 int
9459 ada_is_character_type (struct type *type)
9460 {
9461 const char *name;
9462
9463 /* If the type code says it's a character, then assume it really is,
9464 and don't check any further. */
9465 if (TYPE_CODE (type) == TYPE_CODE_CHAR)
9466 return 1;
9467
9468 /* Otherwise, assume it's a character type iff it is a discrete type
9469 with a known character type name. */
9470 name = ada_type_name (type);
9471 return (name != NULL
9472 && (TYPE_CODE (type) == TYPE_CODE_INT
9473 || TYPE_CODE (type) == TYPE_CODE_RANGE)
9474 && (strcmp (name, "character") == 0
9475 || strcmp (name, "wide_character") == 0
9476 || strcmp (name, "wide_wide_character") == 0
9477 || strcmp (name, "unsigned char") == 0));
9478 }
9479
9480 /* True if TYPE appears to be an Ada string type. */
9481
9482 int
9483 ada_is_string_type (struct type *type)
9484 {
9485 type = ada_check_typedef (type);
9486 if (type != NULL
9487 && TYPE_CODE (type) != TYPE_CODE_PTR
9488 && (ada_is_simple_array_type (type)
9489 || ada_is_array_descriptor_type (type))
9490 && ada_array_arity (type) == 1)
9491 {
9492 struct type *elttype = ada_array_element_type (type, 1);
9493
9494 return ada_is_character_type (elttype);
9495 }
9496 else
9497 return 0;
9498 }
9499
9500 /* The compiler sometimes provides a parallel XVS type for a given
9501 PAD type. Normally, it is safe to follow the PAD type directly,
9502 but older versions of the compiler have a bug that causes the offset
9503 of its "F" field to be wrong. Following that field in that case
9504 would lead to incorrect results, but this can be worked around
9505 by ignoring the PAD type and using the associated XVS type instead.
9506
9507 Set to True if the debugger should trust the contents of PAD types.
9508 Otherwise, ignore the PAD type if there is a parallel XVS type. */
9509 static int trust_pad_over_xvs = 1;
9510
9511 /* True if TYPE is a struct type introduced by the compiler to force the
9512 alignment of a value. Such types have a single field with a
9513 distinctive name. */
9514
9515 int
9516 ada_is_aligner_type (struct type *type)
9517 {
9518 type = ada_check_typedef (type);
9519
9520 if (!trust_pad_over_xvs && ada_find_parallel_type (type, "___XVS") != NULL)
9521 return 0;
9522
9523 return (TYPE_CODE (type) == TYPE_CODE_STRUCT
9524 && TYPE_NFIELDS (type) == 1
9525 && strcmp (TYPE_FIELD_NAME (type, 0), "F") == 0);
9526 }
9527
9528 /* If there is an ___XVS-convention type parallel to SUBTYPE, return
9529 the parallel type. */
9530
9531 struct type *
9532 ada_get_base_type (struct type *raw_type)
9533 {
9534 struct type *real_type_namer;
9535 struct type *raw_real_type;
9536
9537 if (raw_type == NULL || TYPE_CODE (raw_type) != TYPE_CODE_STRUCT)
9538 return raw_type;
9539
9540 if (ada_is_aligner_type (raw_type))
9541 /* The encoding specifies that we should always use the aligner type.
9542 So, even if this aligner type has an associated XVS type, we should
9543 simply ignore it.
9544
9545 According to the compiler gurus, an XVS type parallel to an aligner
9546 type may exist because of a stabs limitation. In stabs, aligner
9547 types are empty because the field has a variable-sized type, and
9548 thus cannot actually be used as an aligner type. As a result,
9549 we need the associated parallel XVS type to decode the type.
9550 Since the policy in the compiler is to not change the internal
9551 representation based on the debugging info format, we sometimes
9552 end up having a redundant XVS type parallel to the aligner type. */
9553 return raw_type;
9554
9555 real_type_namer = ada_find_parallel_type (raw_type, "___XVS");
9556 if (real_type_namer == NULL
9557 || TYPE_CODE (real_type_namer) != TYPE_CODE_STRUCT
9558 || TYPE_NFIELDS (real_type_namer) != 1)
9559 return raw_type;
9560
9561 if (TYPE_CODE (TYPE_FIELD_TYPE (real_type_namer, 0)) != TYPE_CODE_REF)
9562 {
9563 /* This is an older encoding form where the base type needs to be
9564 looked up by name. We prefer the newer enconding because it is
9565 more efficient. */
9566 raw_real_type = ada_find_any_type (TYPE_FIELD_NAME (real_type_namer, 0));
9567 if (raw_real_type == NULL)
9568 return raw_type;
9569 else
9570 return raw_real_type;
9571 }
9572
9573 /* The field in our XVS type is a reference to the base type. */
9574 return TYPE_TARGET_TYPE (TYPE_FIELD_TYPE (real_type_namer, 0));
9575 }
9576
9577 /* The type of value designated by TYPE, with all aligners removed. */
9578
9579 struct type *
9580 ada_aligned_type (struct type *type)
9581 {
9582 if (ada_is_aligner_type (type))
9583 return ada_aligned_type (TYPE_FIELD_TYPE (type, 0));
9584 else
9585 return ada_get_base_type (type);
9586 }
9587
9588
9589 /* The address of the aligned value in an object at address VALADDR
9590 having type TYPE. Assumes ada_is_aligner_type (TYPE). */
9591
9592 const gdb_byte *
9593 ada_aligned_value_addr (struct type *type, const gdb_byte *valaddr)
9594 {
9595 if (ada_is_aligner_type (type))
9596 return ada_aligned_value_addr (TYPE_FIELD_TYPE (type, 0),
9597 valaddr +
9598 TYPE_FIELD_BITPOS (type,
9599 0) / TARGET_CHAR_BIT);
9600 else
9601 return valaddr;
9602 }
9603
9604
9605
9606 /* The printed representation of an enumeration literal with encoded
9607 name NAME. The value is good to the next call of ada_enum_name. */
9608 const char *
9609 ada_enum_name (const char *name)
9610 {
9611 static char *result;
9612 static size_t result_len = 0;
9613 const char *tmp;
9614
9615 /* First, unqualify the enumeration name:
9616 1. Search for the last '.' character. If we find one, then skip
9617 all the preceding characters, the unqualified name starts
9618 right after that dot.
9619 2. Otherwise, we may be debugging on a target where the compiler
9620 translates dots into "__". Search forward for double underscores,
9621 but stop searching when we hit an overloading suffix, which is
9622 of the form "__" followed by digits. */
9623
9624 tmp = strrchr (name, '.');
9625 if (tmp != NULL)
9626 name = tmp + 1;
9627 else
9628 {
9629 while ((tmp = strstr (name, "__")) != NULL)
9630 {
9631 if (isdigit (tmp[2]))
9632 break;
9633 else
9634 name = tmp + 2;
9635 }
9636 }
9637
9638 if (name[0] == 'Q')
9639 {
9640 int v;
9641
9642 if (name[1] == 'U' || name[1] == 'W')
9643 {
9644 if (sscanf (name + 2, "%x", &v) != 1)
9645 return name;
9646 }
9647 else
9648 return name;
9649
9650 GROW_VECT (result, result_len, 16);
9651 if (isascii (v) && isprint (v))
9652 xsnprintf (result, result_len, "'%c'", v);
9653 else if (name[1] == 'U')
9654 xsnprintf (result, result_len, "[\"%02x\"]", v);
9655 else
9656 xsnprintf (result, result_len, "[\"%04x\"]", v);
9657
9658 return result;
9659 }
9660 else
9661 {
9662 tmp = strstr (name, "__");
9663 if (tmp == NULL)
9664 tmp = strstr (name, "$");
9665 if (tmp != NULL)
9666 {
9667 GROW_VECT (result, result_len, tmp - name + 1);
9668 strncpy (result, name, tmp - name);
9669 result[tmp - name] = '\0';
9670 return result;
9671 }
9672
9673 return name;
9674 }
9675 }
9676
9677 /* Evaluate the subexpression of EXP starting at *POS as for
9678 evaluate_type, updating *POS to point just past the evaluated
9679 expression. */
9680
9681 static struct value *
9682 evaluate_subexp_type (struct expression *exp, int *pos)
9683 {
9684 return evaluate_subexp (NULL_TYPE, exp, pos, EVAL_AVOID_SIDE_EFFECTS);
9685 }
9686
9687 /* If VAL is wrapped in an aligner or subtype wrapper, return the
9688 value it wraps. */
9689
9690 static struct value *
9691 unwrap_value (struct value *val)
9692 {
9693 struct type *type = ada_check_typedef (value_type (val));
9694
9695 if (ada_is_aligner_type (type))
9696 {
9697 struct value *v = ada_value_struct_elt (val, "F", 0);
9698 struct type *val_type = ada_check_typedef (value_type (v));
9699
9700 if (ada_type_name (val_type) == NULL)
9701 TYPE_NAME (val_type) = ada_type_name (type);
9702
9703 return unwrap_value (v);
9704 }
9705 else
9706 {
9707 struct type *raw_real_type =
9708 ada_check_typedef (ada_get_base_type (type));
9709
9710 /* If there is no parallel XVS or XVE type, then the value is
9711 already unwrapped. Return it without further modification. */
9712 if ((type == raw_real_type)
9713 && ada_find_parallel_type (type, "___XVE") == NULL)
9714 return val;
9715
9716 return
9717 coerce_unspec_val_to_type
9718 (val, ada_to_fixed_type (raw_real_type, 0,
9719 value_address (val),
9720 NULL, 1));
9721 }
9722 }
9723
9724 static struct value *
9725 cast_from_fixed (struct type *type, struct value *arg)
9726 {
9727 struct value *scale = ada_scaling_factor (value_type (arg));
9728 arg = value_cast (value_type (scale), arg);
9729
9730 arg = value_binop (arg, scale, BINOP_MUL);
9731 return value_cast (type, arg);
9732 }
9733
9734 static struct value *
9735 cast_to_fixed (struct type *type, struct value *arg)
9736 {
9737 if (type == value_type (arg))
9738 return arg;
9739
9740 struct value *scale = ada_scaling_factor (type);
9741 if (ada_is_fixed_point_type (value_type (arg)))
9742 arg = cast_from_fixed (value_type (scale), arg);
9743 else
9744 arg = value_cast (value_type (scale), arg);
9745
9746 arg = value_binop (arg, scale, BINOP_DIV);
9747 return value_cast (type, arg);
9748 }
9749
9750 /* Given two array types T1 and T2, return nonzero iff both arrays
9751 contain the same number of elements. */
9752
9753 static int
9754 ada_same_array_size_p (struct type *t1, struct type *t2)
9755 {
9756 LONGEST lo1, hi1, lo2, hi2;
9757
9758 /* Get the array bounds in order to verify that the size of
9759 the two arrays match. */
9760 if (!get_array_bounds (t1, &lo1, &hi1)
9761 || !get_array_bounds (t2, &lo2, &hi2))
9762 error (_("unable to determine array bounds"));
9763
9764 /* To make things easier for size comparison, normalize a bit
9765 the case of empty arrays by making sure that the difference
9766 between upper bound and lower bound is always -1. */
9767 if (lo1 > hi1)
9768 hi1 = lo1 - 1;
9769 if (lo2 > hi2)
9770 hi2 = lo2 - 1;
9771
9772 return (hi1 - lo1 == hi2 - lo2);
9773 }
9774
9775 /* Assuming that VAL is an array of integrals, and TYPE represents
9776 an array with the same number of elements, but with wider integral
9777 elements, return an array "casted" to TYPE. In practice, this
9778 means that the returned array is built by casting each element
9779 of the original array into TYPE's (wider) element type. */
9780
9781 static struct value *
9782 ada_promote_array_of_integrals (struct type *type, struct value *val)
9783 {
9784 struct type *elt_type = TYPE_TARGET_TYPE (type);
9785 LONGEST lo, hi;
9786 struct value *res;
9787 LONGEST i;
9788
9789 /* Verify that both val and type are arrays of scalars, and
9790 that the size of val's elements is smaller than the size
9791 of type's element. */
9792 gdb_assert (TYPE_CODE (type) == TYPE_CODE_ARRAY);
9793 gdb_assert (is_integral_type (TYPE_TARGET_TYPE (type)));
9794 gdb_assert (TYPE_CODE (value_type (val)) == TYPE_CODE_ARRAY);
9795 gdb_assert (is_integral_type (TYPE_TARGET_TYPE (value_type (val))));
9796 gdb_assert (TYPE_LENGTH (TYPE_TARGET_TYPE (type))
9797 > TYPE_LENGTH (TYPE_TARGET_TYPE (value_type (val))));
9798
9799 if (!get_array_bounds (type, &lo, &hi))
9800 error (_("unable to determine array bounds"));
9801
9802 res = allocate_value (type);
9803
9804 /* Promote each array element. */
9805 for (i = 0; i < hi - lo + 1; i++)
9806 {
9807 struct value *elt = value_cast (elt_type, value_subscript (val, lo + i));
9808
9809 memcpy (value_contents_writeable (res) + (i * TYPE_LENGTH (elt_type)),
9810 value_contents_all (elt), TYPE_LENGTH (elt_type));
9811 }
9812
9813 return res;
9814 }
9815
9816 /* Coerce VAL as necessary for assignment to an lval of type TYPE, and
9817 return the converted value. */
9818
9819 static struct value *
9820 coerce_for_assign (struct type *type, struct value *val)
9821 {
9822 struct type *type2 = value_type (val);
9823
9824 if (type == type2)
9825 return val;
9826
9827 type2 = ada_check_typedef (type2);
9828 type = ada_check_typedef (type);
9829
9830 if (TYPE_CODE (type2) == TYPE_CODE_PTR
9831 && TYPE_CODE (type) == TYPE_CODE_ARRAY)
9832 {
9833 val = ada_value_ind (val);
9834 type2 = value_type (val);
9835 }
9836
9837 if (TYPE_CODE (type2) == TYPE_CODE_ARRAY
9838 && TYPE_CODE (type) == TYPE_CODE_ARRAY)
9839 {
9840 if (!ada_same_array_size_p (type, type2))
9841 error (_("cannot assign arrays of different length"));
9842
9843 if (is_integral_type (TYPE_TARGET_TYPE (type))
9844 && is_integral_type (TYPE_TARGET_TYPE (type2))
9845 && TYPE_LENGTH (TYPE_TARGET_TYPE (type2))
9846 < TYPE_LENGTH (TYPE_TARGET_TYPE (type)))
9847 {
9848 /* Allow implicit promotion of the array elements to
9849 a wider type. */
9850 return ada_promote_array_of_integrals (type, val);
9851 }
9852
9853 if (TYPE_LENGTH (TYPE_TARGET_TYPE (type2))
9854 != TYPE_LENGTH (TYPE_TARGET_TYPE (type)))
9855 error (_("Incompatible types in assignment"));
9856 deprecated_set_value_type (val, type);
9857 }
9858 return val;
9859 }
9860
9861 static struct value *
9862 ada_value_binop (struct value *arg1, struct value *arg2, enum exp_opcode op)
9863 {
9864 struct value *val;
9865 struct type *type1, *type2;
9866 LONGEST v, v1, v2;
9867
9868 arg1 = coerce_ref (arg1);
9869 arg2 = coerce_ref (arg2);
9870 type1 = get_base_type (ada_check_typedef (value_type (arg1)));
9871 type2 = get_base_type (ada_check_typedef (value_type (arg2)));
9872
9873 if (TYPE_CODE (type1) != TYPE_CODE_INT
9874 || TYPE_CODE (type2) != TYPE_CODE_INT)
9875 return value_binop (arg1, arg2, op);
9876
9877 switch (op)
9878 {
9879 case BINOP_MOD:
9880 case BINOP_DIV:
9881 case BINOP_REM:
9882 break;
9883 default:
9884 return value_binop (arg1, arg2, op);
9885 }
9886
9887 v2 = value_as_long (arg2);
9888 if (v2 == 0)
9889 error (_("second operand of %s must not be zero."), op_string (op));
9890
9891 if (TYPE_UNSIGNED (type1) || op == BINOP_MOD)
9892 return value_binop (arg1, arg2, op);
9893
9894 v1 = value_as_long (arg1);
9895 switch (op)
9896 {
9897 case BINOP_DIV:
9898 v = v1 / v2;
9899 if (!TRUNCATION_TOWARDS_ZERO && v1 * (v1 % v2) < 0)
9900 v += v > 0 ? -1 : 1;
9901 break;
9902 case BINOP_REM:
9903 v = v1 % v2;
9904 if (v * v1 < 0)
9905 v -= v2;
9906 break;
9907 default:
9908 /* Should not reach this point. */
9909 v = 0;
9910 }
9911
9912 val = allocate_value (type1);
9913 store_unsigned_integer (value_contents_raw (val),
9914 TYPE_LENGTH (value_type (val)),
9915 gdbarch_byte_order (get_type_arch (type1)), v);
9916 return val;
9917 }
9918
9919 static int
9920 ada_value_equal (struct value *arg1, struct value *arg2)
9921 {
9922 if (ada_is_direct_array_type (value_type (arg1))
9923 || ada_is_direct_array_type (value_type (arg2)))
9924 {
9925 struct type *arg1_type, *arg2_type;
9926
9927 /* Automatically dereference any array reference before
9928 we attempt to perform the comparison. */
9929 arg1 = ada_coerce_ref (arg1);
9930 arg2 = ada_coerce_ref (arg2);
9931
9932 arg1 = ada_coerce_to_simple_array (arg1);
9933 arg2 = ada_coerce_to_simple_array (arg2);
9934
9935 arg1_type = ada_check_typedef (value_type (arg1));
9936 arg2_type = ada_check_typedef (value_type (arg2));
9937
9938 if (TYPE_CODE (arg1_type) != TYPE_CODE_ARRAY
9939 || TYPE_CODE (arg2_type) != TYPE_CODE_ARRAY)
9940 error (_("Attempt to compare array with non-array"));
9941 /* FIXME: The following works only for types whose
9942 representations use all bits (no padding or undefined bits)
9943 and do not have user-defined equality. */
9944 return (TYPE_LENGTH (arg1_type) == TYPE_LENGTH (arg2_type)
9945 && memcmp (value_contents (arg1), value_contents (arg2),
9946 TYPE_LENGTH (arg1_type)) == 0);
9947 }
9948 return value_equal (arg1, arg2);
9949 }
9950
9951 /* Total number of component associations in the aggregate starting at
9952 index PC in EXP. Assumes that index PC is the start of an
9953 OP_AGGREGATE. */
9954
9955 static int
9956 num_component_specs (struct expression *exp, int pc)
9957 {
9958 int n, m, i;
9959
9960 m = exp->elts[pc + 1].longconst;
9961 pc += 3;
9962 n = 0;
9963 for (i = 0; i < m; i += 1)
9964 {
9965 switch (exp->elts[pc].opcode)
9966 {
9967 default:
9968 n += 1;
9969 break;
9970 case OP_CHOICES:
9971 n += exp->elts[pc + 1].longconst;
9972 break;
9973 }
9974 ada_evaluate_subexp (NULL, exp, &pc, EVAL_SKIP);
9975 }
9976 return n;
9977 }
9978
9979 /* Assign the result of evaluating EXP starting at *POS to the INDEXth
9980 component of LHS (a simple array or a record), updating *POS past
9981 the expression, assuming that LHS is contained in CONTAINER. Does
9982 not modify the inferior's memory, nor does it modify LHS (unless
9983 LHS == CONTAINER). */
9984
9985 static void
9986 assign_component (struct value *container, struct value *lhs, LONGEST index,
9987 struct expression *exp, int *pos)
9988 {
9989 struct value *mark = value_mark ();
9990 struct value *elt;
9991 struct type *lhs_type = check_typedef (value_type (lhs));
9992
9993 if (TYPE_CODE (lhs_type) == TYPE_CODE_ARRAY)
9994 {
9995 struct type *index_type = builtin_type (exp->gdbarch)->builtin_int;
9996 struct value *index_val = value_from_longest (index_type, index);
9997
9998 elt = unwrap_value (ada_value_subscript (lhs, 1, &index_val));
9999 }
10000 else
10001 {
10002 elt = ada_index_struct_field (index, lhs, 0, value_type (lhs));
10003 elt = ada_to_fixed_value (elt);
10004 }
10005
10006 if (exp->elts[*pos].opcode == OP_AGGREGATE)
10007 assign_aggregate (container, elt, exp, pos, EVAL_NORMAL);
10008 else
10009 value_assign_to_component (container, elt,
10010 ada_evaluate_subexp (NULL, exp, pos,
10011 EVAL_NORMAL));
10012
10013 value_free_to_mark (mark);
10014 }
10015
10016 /* Assuming that LHS represents an lvalue having a record or array
10017 type, and EXP->ELTS[*POS] is an OP_AGGREGATE, evaluate an assignment
10018 of that aggregate's value to LHS, advancing *POS past the
10019 aggregate. NOSIDE is as for evaluate_subexp. CONTAINER is an
10020 lvalue containing LHS (possibly LHS itself). Does not modify
10021 the inferior's memory, nor does it modify the contents of
10022 LHS (unless == CONTAINER). Returns the modified CONTAINER. */
10023
10024 static struct value *
10025 assign_aggregate (struct value *container,
10026 struct value *lhs, struct expression *exp,
10027 int *pos, enum noside noside)
10028 {
10029 struct type *lhs_type;
10030 int n = exp->elts[*pos+1].longconst;
10031 LONGEST low_index, high_index;
10032 int num_specs;
10033 LONGEST *indices;
10034 int max_indices, num_indices;
10035 int i;
10036
10037 *pos += 3;
10038 if (noside != EVAL_NORMAL)
10039 {
10040 for (i = 0; i < n; i += 1)
10041 ada_evaluate_subexp (NULL, exp, pos, noside);
10042 return container;
10043 }
10044
10045 container = ada_coerce_ref (container);
10046 if (ada_is_direct_array_type (value_type (container)))
10047 container = ada_coerce_to_simple_array (container);
10048 lhs = ada_coerce_ref (lhs);
10049 if (!deprecated_value_modifiable (lhs))
10050 error (_("Left operand of assignment is not a modifiable lvalue."));
10051
10052 lhs_type = check_typedef (value_type (lhs));
10053 if (ada_is_direct_array_type (lhs_type))
10054 {
10055 lhs = ada_coerce_to_simple_array (lhs);
10056 lhs_type = check_typedef (value_type (lhs));
10057 low_index = TYPE_ARRAY_LOWER_BOUND_VALUE (lhs_type);
10058 high_index = TYPE_ARRAY_UPPER_BOUND_VALUE (lhs_type);
10059 }
10060 else if (TYPE_CODE (lhs_type) == TYPE_CODE_STRUCT)
10061 {
10062 low_index = 0;
10063 high_index = num_visible_fields (lhs_type) - 1;
10064 }
10065 else
10066 error (_("Left-hand side must be array or record."));
10067
10068 num_specs = num_component_specs (exp, *pos - 3);
10069 max_indices = 4 * num_specs + 4;
10070 indices = XALLOCAVEC (LONGEST, max_indices);
10071 indices[0] = indices[1] = low_index - 1;
10072 indices[2] = indices[3] = high_index + 1;
10073 num_indices = 4;
10074
10075 for (i = 0; i < n; i += 1)
10076 {
10077 switch (exp->elts[*pos].opcode)
10078 {
10079 case OP_CHOICES:
10080 aggregate_assign_from_choices (container, lhs, exp, pos, indices,
10081 &num_indices, max_indices,
10082 low_index, high_index);
10083 break;
10084 case OP_POSITIONAL:
10085 aggregate_assign_positional (container, lhs, exp, pos, indices,
10086 &num_indices, max_indices,
10087 low_index, high_index);
10088 break;
10089 case OP_OTHERS:
10090 if (i != n-1)
10091 error (_("Misplaced 'others' clause"));
10092 aggregate_assign_others (container, lhs, exp, pos, indices,
10093 num_indices, low_index, high_index);
10094 break;
10095 default:
10096 error (_("Internal error: bad aggregate clause"));
10097 }
10098 }
10099
10100 return container;
10101 }
10102
10103 /* Assign into the component of LHS indexed by the OP_POSITIONAL
10104 construct at *POS, updating *POS past the construct, given that
10105 the positions are relative to lower bound LOW, where HIGH is the
10106 upper bound. Record the position in INDICES[0 .. MAX_INDICES-1]
10107 updating *NUM_INDICES as needed. CONTAINER is as for
10108 assign_aggregate. */
10109 static void
10110 aggregate_assign_positional (struct value *container,
10111 struct value *lhs, struct expression *exp,
10112 int *pos, LONGEST *indices, int *num_indices,
10113 int max_indices, LONGEST low, LONGEST high)
10114 {
10115 LONGEST ind = longest_to_int (exp->elts[*pos + 1].longconst) + low;
10116
10117 if (ind - 1 == high)
10118 warning (_("Extra components in aggregate ignored."));
10119 if (ind <= high)
10120 {
10121 add_component_interval (ind, ind, indices, num_indices, max_indices);
10122 *pos += 3;
10123 assign_component (container, lhs, ind, exp, pos);
10124 }
10125 else
10126 ada_evaluate_subexp (NULL, exp, pos, EVAL_SKIP);
10127 }
10128
10129 /* Assign into the components of LHS indexed by the OP_CHOICES
10130 construct at *POS, updating *POS past the construct, given that
10131 the allowable indices are LOW..HIGH. Record the indices assigned
10132 to in INDICES[0 .. MAX_INDICES-1], updating *NUM_INDICES as
10133 needed. CONTAINER is as for assign_aggregate. */
10134 static void
10135 aggregate_assign_from_choices (struct value *container,
10136 struct value *lhs, struct expression *exp,
10137 int *pos, LONGEST *indices, int *num_indices,
10138 int max_indices, LONGEST low, LONGEST high)
10139 {
10140 int j;
10141 int n_choices = longest_to_int (exp->elts[*pos+1].longconst);
10142 int choice_pos, expr_pc;
10143 int is_array = ada_is_direct_array_type (value_type (lhs));
10144
10145 choice_pos = *pos += 3;
10146
10147 for (j = 0; j < n_choices; j += 1)
10148 ada_evaluate_subexp (NULL, exp, pos, EVAL_SKIP);
10149 expr_pc = *pos;
10150 ada_evaluate_subexp (NULL, exp, pos, EVAL_SKIP);
10151
10152 for (j = 0; j < n_choices; j += 1)
10153 {
10154 LONGEST lower, upper;
10155 enum exp_opcode op = exp->elts[choice_pos].opcode;
10156
10157 if (op == OP_DISCRETE_RANGE)
10158 {
10159 choice_pos += 1;
10160 lower = value_as_long (ada_evaluate_subexp (NULL, exp, pos,
10161 EVAL_NORMAL));
10162 upper = value_as_long (ada_evaluate_subexp (NULL, exp, pos,
10163 EVAL_NORMAL));
10164 }
10165 else if (is_array)
10166 {
10167 lower = value_as_long (ada_evaluate_subexp (NULL, exp, &choice_pos,
10168 EVAL_NORMAL));
10169 upper = lower;
10170 }
10171 else
10172 {
10173 int ind;
10174 const char *name;
10175
10176 switch (op)
10177 {
10178 case OP_NAME:
10179 name = &exp->elts[choice_pos + 2].string;
10180 break;
10181 case OP_VAR_VALUE:
10182 name = SYMBOL_NATURAL_NAME (exp->elts[choice_pos + 2].symbol);
10183 break;
10184 default:
10185 error (_("Invalid record component association."));
10186 }
10187 ada_evaluate_subexp (NULL, exp, &choice_pos, EVAL_SKIP);
10188 ind = 0;
10189 if (! find_struct_field (name, value_type (lhs), 0,
10190 NULL, NULL, NULL, NULL, &ind))
10191 error (_("Unknown component name: %s."), name);
10192 lower = upper = ind;
10193 }
10194
10195 if (lower <= upper && (lower < low || upper > high))
10196 error (_("Index in component association out of bounds."));
10197
10198 add_component_interval (lower, upper, indices, num_indices,
10199 max_indices);
10200 while (lower <= upper)
10201 {
10202 int pos1;
10203
10204 pos1 = expr_pc;
10205 assign_component (container, lhs, lower, exp, &pos1);
10206 lower += 1;
10207 }
10208 }
10209 }
10210
10211 /* Assign the value of the expression in the OP_OTHERS construct in
10212 EXP at *POS into the components of LHS indexed from LOW .. HIGH that
10213 have not been previously assigned. The index intervals already assigned
10214 are in INDICES[0 .. NUM_INDICES-1]. Updates *POS to after the
10215 OP_OTHERS clause. CONTAINER is as for assign_aggregate. */
10216 static void
10217 aggregate_assign_others (struct value *container,
10218 struct value *lhs, struct expression *exp,
10219 int *pos, LONGEST *indices, int num_indices,
10220 LONGEST low, LONGEST high)
10221 {
10222 int i;
10223 int expr_pc = *pos + 1;
10224
10225 for (i = 0; i < num_indices - 2; i += 2)
10226 {
10227 LONGEST ind;
10228
10229 for (ind = indices[i + 1] + 1; ind < indices[i + 2]; ind += 1)
10230 {
10231 int localpos;
10232
10233 localpos = expr_pc;
10234 assign_component (container, lhs, ind, exp, &localpos);
10235 }
10236 }
10237 ada_evaluate_subexp (NULL, exp, pos, EVAL_SKIP);
10238 }
10239
10240 /* Add the interval [LOW .. HIGH] to the sorted set of intervals
10241 [ INDICES[0] .. INDICES[1] ],..., [ INDICES[*SIZE-2] .. INDICES[*SIZE-1] ],
10242 modifying *SIZE as needed. It is an error if *SIZE exceeds
10243 MAX_SIZE. The resulting intervals do not overlap. */
10244 static void
10245 add_component_interval (LONGEST low, LONGEST high,
10246 LONGEST* indices, int *size, int max_size)
10247 {
10248 int i, j;
10249
10250 for (i = 0; i < *size; i += 2) {
10251 if (high >= indices[i] && low <= indices[i + 1])
10252 {
10253 int kh;
10254
10255 for (kh = i + 2; kh < *size; kh += 2)
10256 if (high < indices[kh])
10257 break;
10258 if (low < indices[i])
10259 indices[i] = low;
10260 indices[i + 1] = indices[kh - 1];
10261 if (high > indices[i + 1])
10262 indices[i + 1] = high;
10263 memcpy (indices + i + 2, indices + kh, *size - kh);
10264 *size -= kh - i - 2;
10265 return;
10266 }
10267 else if (high < indices[i])
10268 break;
10269 }
10270
10271 if (*size == max_size)
10272 error (_("Internal error: miscounted aggregate components."));
10273 *size += 2;
10274 for (j = *size-1; j >= i+2; j -= 1)
10275 indices[j] = indices[j - 2];
10276 indices[i] = low;
10277 indices[i + 1] = high;
10278 }
10279
10280 /* Perform and Ada cast of ARG2 to type TYPE if the type of ARG2
10281 is different. */
10282
10283 static struct value *
10284 ada_value_cast (struct type *type, struct value *arg2)
10285 {
10286 if (type == ada_check_typedef (value_type (arg2)))
10287 return arg2;
10288
10289 if (ada_is_fixed_point_type (type))
10290 return (cast_to_fixed (type, arg2));
10291
10292 if (ada_is_fixed_point_type (value_type (arg2)))
10293 return cast_from_fixed (type, arg2);
10294
10295 return value_cast (type, arg2);
10296 }
10297
10298 /* Evaluating Ada expressions, and printing their result.
10299 ------------------------------------------------------
10300
10301 1. Introduction:
10302 ----------------
10303
10304 We usually evaluate an Ada expression in order to print its value.
10305 We also evaluate an expression in order to print its type, which
10306 happens during the EVAL_AVOID_SIDE_EFFECTS phase of the evaluation,
10307 but we'll focus mostly on the EVAL_NORMAL phase. In practice, the
10308 EVAL_AVOID_SIDE_EFFECTS phase allows us to simplify certain aspects of
10309 the evaluation compared to the EVAL_NORMAL, but is otherwise very
10310 similar.
10311
10312 Evaluating expressions is a little more complicated for Ada entities
10313 than it is for entities in languages such as C. The main reason for
10314 this is that Ada provides types whose definition might be dynamic.
10315 One example of such types is variant records. Or another example
10316 would be an array whose bounds can only be known at run time.
10317
10318 The following description is a general guide as to what should be
10319 done (and what should NOT be done) in order to evaluate an expression
10320 involving such types, and when. This does not cover how the semantic
10321 information is encoded by GNAT as this is covered separatly. For the
10322 document used as the reference for the GNAT encoding, see exp_dbug.ads
10323 in the GNAT sources.
10324
10325 Ideally, we should embed each part of this description next to its
10326 associated code. Unfortunately, the amount of code is so vast right
10327 now that it's hard to see whether the code handling a particular
10328 situation might be duplicated or not. One day, when the code is
10329 cleaned up, this guide might become redundant with the comments
10330 inserted in the code, and we might want to remove it.
10331
10332 2. ``Fixing'' an Entity, the Simple Case:
10333 -----------------------------------------
10334
10335 When evaluating Ada expressions, the tricky issue is that they may
10336 reference entities whose type contents and size are not statically
10337 known. Consider for instance a variant record:
10338
10339 type Rec (Empty : Boolean := True) is record
10340 case Empty is
10341 when True => null;
10342 when False => Value : Integer;
10343 end case;
10344 end record;
10345 Yes : Rec := (Empty => False, Value => 1);
10346 No : Rec := (empty => True);
10347
10348 The size and contents of that record depends on the value of the
10349 descriminant (Rec.Empty). At this point, neither the debugging
10350 information nor the associated type structure in GDB are able to
10351 express such dynamic types. So what the debugger does is to create
10352 "fixed" versions of the type that applies to the specific object.
10353 We also informally refer to this opperation as "fixing" an object,
10354 which means creating its associated fixed type.
10355
10356 Example: when printing the value of variable "Yes" above, its fixed
10357 type would look like this:
10358
10359 type Rec is record
10360 Empty : Boolean;
10361 Value : Integer;
10362 end record;
10363
10364 On the other hand, if we printed the value of "No", its fixed type
10365 would become:
10366
10367 type Rec is record
10368 Empty : Boolean;
10369 end record;
10370
10371 Things become a little more complicated when trying to fix an entity
10372 with a dynamic type that directly contains another dynamic type,
10373 such as an array of variant records, for instance. There are
10374 two possible cases: Arrays, and records.
10375
10376 3. ``Fixing'' Arrays:
10377 ---------------------
10378
10379 The type structure in GDB describes an array in terms of its bounds,
10380 and the type of its elements. By design, all elements in the array
10381 have the same type and we cannot represent an array of variant elements
10382 using the current type structure in GDB. When fixing an array,
10383 we cannot fix the array element, as we would potentially need one
10384 fixed type per element of the array. As a result, the best we can do
10385 when fixing an array is to produce an array whose bounds and size
10386 are correct (allowing us to read it from memory), but without having
10387 touched its element type. Fixing each element will be done later,
10388 when (if) necessary.
10389
10390 Arrays are a little simpler to handle than records, because the same
10391 amount of memory is allocated for each element of the array, even if
10392 the amount of space actually used by each element differs from element
10393 to element. Consider for instance the following array of type Rec:
10394
10395 type Rec_Array is array (1 .. 2) of Rec;
10396
10397 The actual amount of memory occupied by each element might be different
10398 from element to element, depending on the value of their discriminant.
10399 But the amount of space reserved for each element in the array remains
10400 fixed regardless. So we simply need to compute that size using
10401 the debugging information available, from which we can then determine
10402 the array size (we multiply the number of elements of the array by
10403 the size of each element).
10404
10405 The simplest case is when we have an array of a constrained element
10406 type. For instance, consider the following type declarations:
10407
10408 type Bounded_String (Max_Size : Integer) is
10409 Length : Integer;
10410 Buffer : String (1 .. Max_Size);
10411 end record;
10412 type Bounded_String_Array is array (1 ..2) of Bounded_String (80);
10413
10414 In this case, the compiler describes the array as an array of
10415 variable-size elements (identified by its XVS suffix) for which
10416 the size can be read in the parallel XVZ variable.
10417
10418 In the case of an array of an unconstrained element type, the compiler
10419 wraps the array element inside a private PAD type. This type should not
10420 be shown to the user, and must be "unwrap"'ed before printing. Note
10421 that we also use the adjective "aligner" in our code to designate
10422 these wrapper types.
10423
10424 In some cases, the size allocated for each element is statically
10425 known. In that case, the PAD type already has the correct size,
10426 and the array element should remain unfixed.
10427
10428 But there are cases when this size is not statically known.
10429 For instance, assuming that "Five" is an integer variable:
10430
10431 type Dynamic is array (1 .. Five) of Integer;
10432 type Wrapper (Has_Length : Boolean := False) is record
10433 Data : Dynamic;
10434 case Has_Length is
10435 when True => Length : Integer;
10436 when False => null;
10437 end case;
10438 end record;
10439 type Wrapper_Array is array (1 .. 2) of Wrapper;
10440
10441 Hello : Wrapper_Array := (others => (Has_Length => True,
10442 Data => (others => 17),
10443 Length => 1));
10444
10445
10446 The debugging info would describe variable Hello as being an
10447 array of a PAD type. The size of that PAD type is not statically
10448 known, but can be determined using a parallel XVZ variable.
10449 In that case, a copy of the PAD type with the correct size should
10450 be used for the fixed array.
10451
10452 3. ``Fixing'' record type objects:
10453 ----------------------------------
10454
10455 Things are slightly different from arrays in the case of dynamic
10456 record types. In this case, in order to compute the associated
10457 fixed type, we need to determine the size and offset of each of
10458 its components. This, in turn, requires us to compute the fixed
10459 type of each of these components.
10460
10461 Consider for instance the example:
10462
10463 type Bounded_String (Max_Size : Natural) is record
10464 Str : String (1 .. Max_Size);
10465 Length : Natural;
10466 end record;
10467 My_String : Bounded_String (Max_Size => 10);
10468
10469 In that case, the position of field "Length" depends on the size
10470 of field Str, which itself depends on the value of the Max_Size
10471 discriminant. In order to fix the type of variable My_String,
10472 we need to fix the type of field Str. Therefore, fixing a variant
10473 record requires us to fix each of its components.
10474
10475 However, if a component does not have a dynamic size, the component
10476 should not be fixed. In particular, fields that use a PAD type
10477 should not fixed. Here is an example where this might happen
10478 (assuming type Rec above):
10479
10480 type Container (Big : Boolean) is record
10481 First : Rec;
10482 After : Integer;
10483 case Big is
10484 when True => Another : Integer;
10485 when False => null;
10486 end case;
10487 end record;
10488 My_Container : Container := (Big => False,
10489 First => (Empty => True),
10490 After => 42);
10491
10492 In that example, the compiler creates a PAD type for component First,
10493 whose size is constant, and then positions the component After just
10494 right after it. The offset of component After is therefore constant
10495 in this case.
10496
10497 The debugger computes the position of each field based on an algorithm
10498 that uses, among other things, the actual position and size of the field
10499 preceding it. Let's now imagine that the user is trying to print
10500 the value of My_Container. If the type fixing was recursive, we would
10501 end up computing the offset of field After based on the size of the
10502 fixed version of field First. And since in our example First has
10503 only one actual field, the size of the fixed type is actually smaller
10504 than the amount of space allocated to that field, and thus we would
10505 compute the wrong offset of field After.
10506
10507 To make things more complicated, we need to watch out for dynamic
10508 components of variant records (identified by the ___XVL suffix in
10509 the component name). Even if the target type is a PAD type, the size
10510 of that type might not be statically known. So the PAD type needs
10511 to be unwrapped and the resulting type needs to be fixed. Otherwise,
10512 we might end up with the wrong size for our component. This can be
10513 observed with the following type declarations:
10514
10515 type Octal is new Integer range 0 .. 7;
10516 type Octal_Array is array (Positive range <>) of Octal;
10517 pragma Pack (Octal_Array);
10518
10519 type Octal_Buffer (Size : Positive) is record
10520 Buffer : Octal_Array (1 .. Size);
10521 Length : Integer;
10522 end record;
10523
10524 In that case, Buffer is a PAD type whose size is unset and needs
10525 to be computed by fixing the unwrapped type.
10526
10527 4. When to ``Fix'' un-``Fixed'' sub-elements of an entity:
10528 ----------------------------------------------------------
10529
10530 Lastly, when should the sub-elements of an entity that remained unfixed
10531 thus far, be actually fixed?
10532
10533 The answer is: Only when referencing that element. For instance
10534 when selecting one component of a record, this specific component
10535 should be fixed at that point in time. Or when printing the value
10536 of a record, each component should be fixed before its value gets
10537 printed. Similarly for arrays, the element of the array should be
10538 fixed when printing each element of the array, or when extracting
10539 one element out of that array. On the other hand, fixing should
10540 not be performed on the elements when taking a slice of an array!
10541
10542 Note that one of the side effects of miscomputing the offset and
10543 size of each field is that we end up also miscomputing the size
10544 of the containing type. This can have adverse results when computing
10545 the value of an entity. GDB fetches the value of an entity based
10546 on the size of its type, and thus a wrong size causes GDB to fetch
10547 the wrong amount of memory. In the case where the computed size is
10548 too small, GDB fetches too little data to print the value of our
10549 entity. Results in this case are unpredictable, as we usually read
10550 past the buffer containing the data =:-o. */
10551
10552 /* Evaluate a subexpression of EXP, at index *POS, and return a value
10553 for that subexpression cast to TO_TYPE. Advance *POS over the
10554 subexpression. */
10555
10556 static value *
10557 ada_evaluate_subexp_for_cast (expression *exp, int *pos,
10558 enum noside noside, struct type *to_type)
10559 {
10560 int pc = *pos;
10561
10562 if (exp->elts[pc].opcode == OP_VAR_MSYM_VALUE
10563 || exp->elts[pc].opcode == OP_VAR_VALUE)
10564 {
10565 (*pos) += 4;
10566
10567 value *val;
10568 if (exp->elts[pc].opcode == OP_VAR_MSYM_VALUE)
10569 {
10570 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10571 return value_zero (to_type, not_lval);
10572
10573 val = evaluate_var_msym_value (noside,
10574 exp->elts[pc + 1].objfile,
10575 exp->elts[pc + 2].msymbol);
10576 }
10577 else
10578 val = evaluate_var_value (noside,
10579 exp->elts[pc + 1].block,
10580 exp->elts[pc + 2].symbol);
10581
10582 if (noside == EVAL_SKIP)
10583 return eval_skip_value (exp);
10584
10585 val = ada_value_cast (to_type, val);
10586
10587 /* Follow the Ada language semantics that do not allow taking
10588 an address of the result of a cast (view conversion in Ada). */
10589 if (VALUE_LVAL (val) == lval_memory)
10590 {
10591 if (value_lazy (val))
10592 value_fetch_lazy (val);
10593 VALUE_LVAL (val) = not_lval;
10594 }
10595 return val;
10596 }
10597
10598 value *val = evaluate_subexp (to_type, exp, pos, noside);
10599 if (noside == EVAL_SKIP)
10600 return eval_skip_value (exp);
10601 return ada_value_cast (to_type, val);
10602 }
10603
10604 /* Implement the evaluate_exp routine in the exp_descriptor structure
10605 for the Ada language. */
10606
10607 static struct value *
10608 ada_evaluate_subexp (struct type *expect_type, struct expression *exp,
10609 int *pos, enum noside noside)
10610 {
10611 enum exp_opcode op;
10612 int tem;
10613 int pc;
10614 int preeval_pos;
10615 struct value *arg1 = NULL, *arg2 = NULL, *arg3;
10616 struct type *type;
10617 int nargs, oplen;
10618 struct value **argvec;
10619
10620 pc = *pos;
10621 *pos += 1;
10622 op = exp->elts[pc].opcode;
10623
10624 switch (op)
10625 {
10626 default:
10627 *pos -= 1;
10628 arg1 = evaluate_subexp_standard (expect_type, exp, pos, noside);
10629
10630 if (noside == EVAL_NORMAL)
10631 arg1 = unwrap_value (arg1);
10632
10633 /* If evaluating an OP_FLOAT and an EXPECT_TYPE was provided,
10634 then we need to perform the conversion manually, because
10635 evaluate_subexp_standard doesn't do it. This conversion is
10636 necessary in Ada because the different kinds of float/fixed
10637 types in Ada have different representations.
10638
10639 Similarly, we need to perform the conversion from OP_LONG
10640 ourselves. */
10641 if ((op == OP_FLOAT || op == OP_LONG) && expect_type != NULL)
10642 arg1 = ada_value_cast (expect_type, arg1);
10643
10644 return arg1;
10645
10646 case OP_STRING:
10647 {
10648 struct value *result;
10649
10650 *pos -= 1;
10651 result = evaluate_subexp_standard (expect_type, exp, pos, noside);
10652 /* The result type will have code OP_STRING, bashed there from
10653 OP_ARRAY. Bash it back. */
10654 if (TYPE_CODE (value_type (result)) == TYPE_CODE_STRING)
10655 TYPE_CODE (value_type (result)) = TYPE_CODE_ARRAY;
10656 return result;
10657 }
10658
10659 case UNOP_CAST:
10660 (*pos) += 2;
10661 type = exp->elts[pc + 1].type;
10662 return ada_evaluate_subexp_for_cast (exp, pos, noside, type);
10663
10664 case UNOP_QUAL:
10665 (*pos) += 2;
10666 type = exp->elts[pc + 1].type;
10667 return ada_evaluate_subexp (type, exp, pos, noside);
10668
10669 case BINOP_ASSIGN:
10670 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10671 if (exp->elts[*pos].opcode == OP_AGGREGATE)
10672 {
10673 arg1 = assign_aggregate (arg1, arg1, exp, pos, noside);
10674 if (noside == EVAL_SKIP || noside == EVAL_AVOID_SIDE_EFFECTS)
10675 return arg1;
10676 return ada_value_assign (arg1, arg1);
10677 }
10678 /* Force the evaluation of the rhs ARG2 to the type of the lhs ARG1,
10679 except if the lhs of our assignment is a convenience variable.
10680 In the case of assigning to a convenience variable, the lhs
10681 should be exactly the result of the evaluation of the rhs. */
10682 type = value_type (arg1);
10683 if (VALUE_LVAL (arg1) == lval_internalvar)
10684 type = NULL;
10685 arg2 = evaluate_subexp (type, exp, pos, noside);
10686 if (noside == EVAL_SKIP || noside == EVAL_AVOID_SIDE_EFFECTS)
10687 return arg1;
10688 if (ada_is_fixed_point_type (value_type (arg1)))
10689 arg2 = cast_to_fixed (value_type (arg1), arg2);
10690 else if (ada_is_fixed_point_type (value_type (arg2)))
10691 error
10692 (_("Fixed-point values must be assigned to fixed-point variables"));
10693 else
10694 arg2 = coerce_for_assign (value_type (arg1), arg2);
10695 return ada_value_assign (arg1, arg2);
10696
10697 case BINOP_ADD:
10698 arg1 = evaluate_subexp_with_coercion (exp, pos, noside);
10699 arg2 = evaluate_subexp_with_coercion (exp, pos, noside);
10700 if (noside == EVAL_SKIP)
10701 goto nosideret;
10702 if (TYPE_CODE (value_type (arg1)) == TYPE_CODE_PTR)
10703 return (value_from_longest
10704 (value_type (arg1),
10705 value_as_long (arg1) + value_as_long (arg2)));
10706 if (TYPE_CODE (value_type (arg2)) == TYPE_CODE_PTR)
10707 return (value_from_longest
10708 (value_type (arg2),
10709 value_as_long (arg1) + value_as_long (arg2)));
10710 if ((ada_is_fixed_point_type (value_type (arg1))
10711 || ada_is_fixed_point_type (value_type (arg2)))
10712 && value_type (arg1) != value_type (arg2))
10713 error (_("Operands of fixed-point addition must have the same type"));
10714 /* Do the addition, and cast the result to the type of the first
10715 argument. We cannot cast the result to a reference type, so if
10716 ARG1 is a reference type, find its underlying type. */
10717 type = value_type (arg1);
10718 while (TYPE_CODE (type) == TYPE_CODE_REF)
10719 type = TYPE_TARGET_TYPE (type);
10720 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10721 return value_cast (type, value_binop (arg1, arg2, BINOP_ADD));
10722
10723 case BINOP_SUB:
10724 arg1 = evaluate_subexp_with_coercion (exp, pos, noside);
10725 arg2 = evaluate_subexp_with_coercion (exp, pos, noside);
10726 if (noside == EVAL_SKIP)
10727 goto nosideret;
10728 if (TYPE_CODE (value_type (arg1)) == TYPE_CODE_PTR)
10729 return (value_from_longest
10730 (value_type (arg1),
10731 value_as_long (arg1) - value_as_long (arg2)));
10732 if (TYPE_CODE (value_type (arg2)) == TYPE_CODE_PTR)
10733 return (value_from_longest
10734 (value_type (arg2),
10735 value_as_long (arg1) - value_as_long (arg2)));
10736 if ((ada_is_fixed_point_type (value_type (arg1))
10737 || ada_is_fixed_point_type (value_type (arg2)))
10738 && value_type (arg1) != value_type (arg2))
10739 error (_("Operands of fixed-point subtraction "
10740 "must have the same type"));
10741 /* Do the substraction, and cast the result to the type of the first
10742 argument. We cannot cast the result to a reference type, so if
10743 ARG1 is a reference type, find its underlying type. */
10744 type = value_type (arg1);
10745 while (TYPE_CODE (type) == TYPE_CODE_REF)
10746 type = TYPE_TARGET_TYPE (type);
10747 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10748 return value_cast (type, value_binop (arg1, arg2, BINOP_SUB));
10749
10750 case BINOP_MUL:
10751 case BINOP_DIV:
10752 case BINOP_REM:
10753 case BINOP_MOD:
10754 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10755 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10756 if (noside == EVAL_SKIP)
10757 goto nosideret;
10758 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
10759 {
10760 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10761 return value_zero (value_type (arg1), not_lval);
10762 }
10763 else
10764 {
10765 type = builtin_type (exp->gdbarch)->builtin_double;
10766 if (ada_is_fixed_point_type (value_type (arg1)))
10767 arg1 = cast_from_fixed (type, arg1);
10768 if (ada_is_fixed_point_type (value_type (arg2)))
10769 arg2 = cast_from_fixed (type, arg2);
10770 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10771 return ada_value_binop (arg1, arg2, op);
10772 }
10773
10774 case BINOP_EQUAL:
10775 case BINOP_NOTEQUAL:
10776 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10777 arg2 = evaluate_subexp (value_type (arg1), exp, pos, noside);
10778 if (noside == EVAL_SKIP)
10779 goto nosideret;
10780 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10781 tem = 0;
10782 else
10783 {
10784 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10785 tem = ada_value_equal (arg1, arg2);
10786 }
10787 if (op == BINOP_NOTEQUAL)
10788 tem = !tem;
10789 type = language_bool_type (exp->language_defn, exp->gdbarch);
10790 return value_from_longest (type, (LONGEST) tem);
10791
10792 case UNOP_NEG:
10793 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10794 if (noside == EVAL_SKIP)
10795 goto nosideret;
10796 else if (ada_is_fixed_point_type (value_type (arg1)))
10797 return value_cast (value_type (arg1), value_neg (arg1));
10798 else
10799 {
10800 unop_promote (exp->language_defn, exp->gdbarch, &arg1);
10801 return value_neg (arg1);
10802 }
10803
10804 case BINOP_LOGICAL_AND:
10805 case BINOP_LOGICAL_OR:
10806 case UNOP_LOGICAL_NOT:
10807 {
10808 struct value *val;
10809
10810 *pos -= 1;
10811 val = evaluate_subexp_standard (expect_type, exp, pos, noside);
10812 type = language_bool_type (exp->language_defn, exp->gdbarch);
10813 return value_cast (type, val);
10814 }
10815
10816 case BINOP_BITWISE_AND:
10817 case BINOP_BITWISE_IOR:
10818 case BINOP_BITWISE_XOR:
10819 {
10820 struct value *val;
10821
10822 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, EVAL_AVOID_SIDE_EFFECTS);
10823 *pos = pc;
10824 val = evaluate_subexp_standard (expect_type, exp, pos, noside);
10825
10826 return value_cast (value_type (arg1), val);
10827 }
10828
10829 case OP_VAR_VALUE:
10830 *pos -= 1;
10831
10832 if (noside == EVAL_SKIP)
10833 {
10834 *pos += 4;
10835 goto nosideret;
10836 }
10837
10838 if (SYMBOL_DOMAIN (exp->elts[pc + 2].symbol) == UNDEF_DOMAIN)
10839 /* Only encountered when an unresolved symbol occurs in a
10840 context other than a function call, in which case, it is
10841 invalid. */
10842 error (_("Unexpected unresolved symbol, %s, during evaluation"),
10843 SYMBOL_PRINT_NAME (exp->elts[pc + 2].symbol));
10844
10845 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10846 {
10847 type = static_unwrap_type (SYMBOL_TYPE (exp->elts[pc + 2].symbol));
10848 /* Check to see if this is a tagged type. We also need to handle
10849 the case where the type is a reference to a tagged type, but
10850 we have to be careful to exclude pointers to tagged types.
10851 The latter should be shown as usual (as a pointer), whereas
10852 a reference should mostly be transparent to the user. */
10853 if (ada_is_tagged_type (type, 0)
10854 || (TYPE_CODE (type) == TYPE_CODE_REF
10855 && ada_is_tagged_type (TYPE_TARGET_TYPE (type), 0)))
10856 {
10857 /* Tagged types are a little special in the fact that the real
10858 type is dynamic and can only be determined by inspecting the
10859 object's tag. This means that we need to get the object's
10860 value first (EVAL_NORMAL) and then extract the actual object
10861 type from its tag.
10862
10863 Note that we cannot skip the final step where we extract
10864 the object type from its tag, because the EVAL_NORMAL phase
10865 results in dynamic components being resolved into fixed ones.
10866 This can cause problems when trying to print the type
10867 description of tagged types whose parent has a dynamic size:
10868 We use the type name of the "_parent" component in order
10869 to print the name of the ancestor type in the type description.
10870 If that component had a dynamic size, the resolution into
10871 a fixed type would result in the loss of that type name,
10872 thus preventing us from printing the name of the ancestor
10873 type in the type description. */
10874 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, EVAL_NORMAL);
10875
10876 if (TYPE_CODE (type) != TYPE_CODE_REF)
10877 {
10878 struct type *actual_type;
10879
10880 actual_type = type_from_tag (ada_value_tag (arg1));
10881 if (actual_type == NULL)
10882 /* If, for some reason, we were unable to determine
10883 the actual type from the tag, then use the static
10884 approximation that we just computed as a fallback.
10885 This can happen if the debugging information is
10886 incomplete, for instance. */
10887 actual_type = type;
10888 return value_zero (actual_type, not_lval);
10889 }
10890 else
10891 {
10892 /* In the case of a ref, ada_coerce_ref takes care
10893 of determining the actual type. But the evaluation
10894 should return a ref as it should be valid to ask
10895 for its address; so rebuild a ref after coerce. */
10896 arg1 = ada_coerce_ref (arg1);
10897 return value_ref (arg1, TYPE_CODE_REF);
10898 }
10899 }
10900
10901 /* Records and unions for which GNAT encodings have been
10902 generated need to be statically fixed as well.
10903 Otherwise, non-static fixing produces a type where
10904 all dynamic properties are removed, which prevents "ptype"
10905 from being able to completely describe the type.
10906 For instance, a case statement in a variant record would be
10907 replaced by the relevant components based on the actual
10908 value of the discriminants. */
10909 if ((TYPE_CODE (type) == TYPE_CODE_STRUCT
10910 && dynamic_template_type (type) != NULL)
10911 || (TYPE_CODE (type) == TYPE_CODE_UNION
10912 && ada_find_parallel_type (type, "___XVU") != NULL))
10913 {
10914 *pos += 4;
10915 return value_zero (to_static_fixed_type (type), not_lval);
10916 }
10917 }
10918
10919 arg1 = evaluate_subexp_standard (expect_type, exp, pos, noside);
10920 return ada_to_fixed_value (arg1);
10921
10922 case OP_FUNCALL:
10923 (*pos) += 2;
10924
10925 /* Allocate arg vector, including space for the function to be
10926 called in argvec[0] and a terminating NULL. */
10927 nargs = longest_to_int (exp->elts[pc + 1].longconst);
10928 argvec = XALLOCAVEC (struct value *, nargs + 2);
10929
10930 if (exp->elts[*pos].opcode == OP_VAR_VALUE
10931 && SYMBOL_DOMAIN (exp->elts[pc + 5].symbol) == UNDEF_DOMAIN)
10932 error (_("Unexpected unresolved symbol, %s, during evaluation"),
10933 SYMBOL_PRINT_NAME (exp->elts[pc + 5].symbol));
10934 else
10935 {
10936 for (tem = 0; tem <= nargs; tem += 1)
10937 argvec[tem] = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10938 argvec[tem] = 0;
10939
10940 if (noside == EVAL_SKIP)
10941 goto nosideret;
10942 }
10943
10944 if (ada_is_constrained_packed_array_type
10945 (desc_base_type (value_type (argvec[0]))))
10946 argvec[0] = ada_coerce_to_simple_array (argvec[0]);
10947 else if (TYPE_CODE (value_type (argvec[0])) == TYPE_CODE_ARRAY
10948 && TYPE_FIELD_BITSIZE (value_type (argvec[0]), 0) != 0)
10949 /* This is a packed array that has already been fixed, and
10950 therefore already coerced to a simple array. Nothing further
10951 to do. */
10952 ;
10953 else if (TYPE_CODE (value_type (argvec[0])) == TYPE_CODE_REF)
10954 {
10955 /* Make sure we dereference references so that all the code below
10956 feels like it's really handling the referenced value. Wrapping
10957 types (for alignment) may be there, so make sure we strip them as
10958 well. */
10959 argvec[0] = ada_to_fixed_value (coerce_ref (argvec[0]));
10960 }
10961 else if (TYPE_CODE (value_type (argvec[0])) == TYPE_CODE_ARRAY
10962 && VALUE_LVAL (argvec[0]) == lval_memory)
10963 argvec[0] = value_addr (argvec[0]);
10964
10965 type = ada_check_typedef (value_type (argvec[0]));
10966
10967 /* Ada allows us to implicitly dereference arrays when subscripting
10968 them. So, if this is an array typedef (encoding use for array
10969 access types encoded as fat pointers), strip it now. */
10970 if (TYPE_CODE (type) == TYPE_CODE_TYPEDEF)
10971 type = ada_typedef_target_type (type);
10972
10973 if (TYPE_CODE (type) == TYPE_CODE_PTR)
10974 {
10975 switch (TYPE_CODE (ada_check_typedef (TYPE_TARGET_TYPE (type))))
10976 {
10977 case TYPE_CODE_FUNC:
10978 type = ada_check_typedef (TYPE_TARGET_TYPE (type));
10979 break;
10980 case TYPE_CODE_ARRAY:
10981 break;
10982 case TYPE_CODE_STRUCT:
10983 if (noside != EVAL_AVOID_SIDE_EFFECTS)
10984 argvec[0] = ada_value_ind (argvec[0]);
10985 type = ada_check_typedef (TYPE_TARGET_TYPE (type));
10986 break;
10987 default:
10988 error (_("cannot subscript or call something of type `%s'"),
10989 ada_type_name (value_type (argvec[0])));
10990 break;
10991 }
10992 }
10993
10994 switch (TYPE_CODE (type))
10995 {
10996 case TYPE_CODE_FUNC:
10997 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10998 {
10999 if (TYPE_TARGET_TYPE (type) == NULL)
11000 error_call_unknown_return_type (NULL);
11001 return allocate_value (TYPE_TARGET_TYPE (type));
11002 }
11003 return call_function_by_hand (argvec[0], NULL, nargs, argvec + 1);
11004 case TYPE_CODE_INTERNAL_FUNCTION:
11005 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11006 /* We don't know anything about what the internal
11007 function might return, but we have to return
11008 something. */
11009 return value_zero (builtin_type (exp->gdbarch)->builtin_int,
11010 not_lval);
11011 else
11012 return call_internal_function (exp->gdbarch, exp->language_defn,
11013 argvec[0], nargs, argvec + 1);
11014
11015 case TYPE_CODE_STRUCT:
11016 {
11017 int arity;
11018
11019 arity = ada_array_arity (type);
11020 type = ada_array_element_type (type, nargs);
11021 if (type == NULL)
11022 error (_("cannot subscript or call a record"));
11023 if (arity != nargs)
11024 error (_("wrong number of subscripts; expecting %d"), arity);
11025 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11026 return value_zero (ada_aligned_type (type), lval_memory);
11027 return
11028 unwrap_value (ada_value_subscript
11029 (argvec[0], nargs, argvec + 1));
11030 }
11031 case TYPE_CODE_ARRAY:
11032 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11033 {
11034 type = ada_array_element_type (type, nargs);
11035 if (type == NULL)
11036 error (_("element type of array unknown"));
11037 else
11038 return value_zero (ada_aligned_type (type), lval_memory);
11039 }
11040 return
11041 unwrap_value (ada_value_subscript
11042 (ada_coerce_to_simple_array (argvec[0]),
11043 nargs, argvec + 1));
11044 case TYPE_CODE_PTR: /* Pointer to array */
11045 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11046 {
11047 type = to_fixed_array_type (TYPE_TARGET_TYPE (type), NULL, 1);
11048 type = ada_array_element_type (type, nargs);
11049 if (type == NULL)
11050 error (_("element type of array unknown"));
11051 else
11052 return value_zero (ada_aligned_type (type), lval_memory);
11053 }
11054 return
11055 unwrap_value (ada_value_ptr_subscript (argvec[0],
11056 nargs, argvec + 1));
11057
11058 default:
11059 error (_("Attempt to index or call something other than an "
11060 "array or function"));
11061 }
11062
11063 case TERNOP_SLICE:
11064 {
11065 struct value *array = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11066 struct value *low_bound_val =
11067 evaluate_subexp (NULL_TYPE, exp, pos, noside);
11068 struct value *high_bound_val =
11069 evaluate_subexp (NULL_TYPE, exp, pos, noside);
11070 LONGEST low_bound;
11071 LONGEST high_bound;
11072
11073 low_bound_val = coerce_ref (low_bound_val);
11074 high_bound_val = coerce_ref (high_bound_val);
11075 low_bound = value_as_long (low_bound_val);
11076 high_bound = value_as_long (high_bound_val);
11077
11078 if (noside == EVAL_SKIP)
11079 goto nosideret;
11080
11081 /* If this is a reference to an aligner type, then remove all
11082 the aligners. */
11083 if (TYPE_CODE (value_type (array)) == TYPE_CODE_REF
11084 && ada_is_aligner_type (TYPE_TARGET_TYPE (value_type (array))))
11085 TYPE_TARGET_TYPE (value_type (array)) =
11086 ada_aligned_type (TYPE_TARGET_TYPE (value_type (array)));
11087
11088 if (ada_is_constrained_packed_array_type (value_type (array)))
11089 error (_("cannot slice a packed array"));
11090
11091 /* If this is a reference to an array or an array lvalue,
11092 convert to a pointer. */
11093 if (TYPE_CODE (value_type (array)) == TYPE_CODE_REF
11094 || (TYPE_CODE (value_type (array)) == TYPE_CODE_ARRAY
11095 && VALUE_LVAL (array) == lval_memory))
11096 array = value_addr (array);
11097
11098 if (noside == EVAL_AVOID_SIDE_EFFECTS
11099 && ada_is_array_descriptor_type (ada_check_typedef
11100 (value_type (array))))
11101 return empty_array (ada_type_of_array (array, 0), low_bound);
11102
11103 array = ada_coerce_to_simple_array_ptr (array);
11104
11105 /* If we have more than one level of pointer indirection,
11106 dereference the value until we get only one level. */
11107 while (TYPE_CODE (value_type (array)) == TYPE_CODE_PTR
11108 && (TYPE_CODE (TYPE_TARGET_TYPE (value_type (array)))
11109 == TYPE_CODE_PTR))
11110 array = value_ind (array);
11111
11112 /* Make sure we really do have an array type before going further,
11113 to avoid a SEGV when trying to get the index type or the target
11114 type later down the road if the debug info generated by
11115 the compiler is incorrect or incomplete. */
11116 if (!ada_is_simple_array_type (value_type (array)))
11117 error (_("cannot take slice of non-array"));
11118
11119 if (TYPE_CODE (ada_check_typedef (value_type (array)))
11120 == TYPE_CODE_PTR)
11121 {
11122 struct type *type0 = ada_check_typedef (value_type (array));
11123
11124 if (high_bound < low_bound || noside == EVAL_AVOID_SIDE_EFFECTS)
11125 return empty_array (TYPE_TARGET_TYPE (type0), low_bound);
11126 else
11127 {
11128 struct type *arr_type0 =
11129 to_fixed_array_type (TYPE_TARGET_TYPE (type0), NULL, 1);
11130
11131 return ada_value_slice_from_ptr (array, arr_type0,
11132 longest_to_int (low_bound),
11133 longest_to_int (high_bound));
11134 }
11135 }
11136 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11137 return array;
11138 else if (high_bound < low_bound)
11139 return empty_array (value_type (array), low_bound);
11140 else
11141 return ada_value_slice (array, longest_to_int (low_bound),
11142 longest_to_int (high_bound));
11143 }
11144
11145 case UNOP_IN_RANGE:
11146 (*pos) += 2;
11147 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11148 type = check_typedef (exp->elts[pc + 1].type);
11149
11150 if (noside == EVAL_SKIP)
11151 goto nosideret;
11152
11153 switch (TYPE_CODE (type))
11154 {
11155 default:
11156 lim_warning (_("Membership test incompletely implemented; "
11157 "always returns true"));
11158 type = language_bool_type (exp->language_defn, exp->gdbarch);
11159 return value_from_longest (type, (LONGEST) 1);
11160
11161 case TYPE_CODE_RANGE:
11162 arg2 = value_from_longest (type, TYPE_LOW_BOUND (type));
11163 arg3 = value_from_longest (type, TYPE_HIGH_BOUND (type));
11164 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
11165 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg3);
11166 type = language_bool_type (exp->language_defn, exp->gdbarch);
11167 return
11168 value_from_longest (type,
11169 (value_less (arg1, arg3)
11170 || value_equal (arg1, arg3))
11171 && (value_less (arg2, arg1)
11172 || value_equal (arg2, arg1)));
11173 }
11174
11175 case BINOP_IN_BOUNDS:
11176 (*pos) += 2;
11177 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11178 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11179
11180 if (noside == EVAL_SKIP)
11181 goto nosideret;
11182
11183 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11184 {
11185 type = language_bool_type (exp->language_defn, exp->gdbarch);
11186 return value_zero (type, not_lval);
11187 }
11188
11189 tem = longest_to_int (exp->elts[pc + 1].longconst);
11190
11191 type = ada_index_type (value_type (arg2), tem, "range");
11192 if (!type)
11193 type = value_type (arg1);
11194
11195 arg3 = value_from_longest (type, ada_array_bound (arg2, tem, 1));
11196 arg2 = value_from_longest (type, ada_array_bound (arg2, tem, 0));
11197
11198 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
11199 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg3);
11200 type = language_bool_type (exp->language_defn, exp->gdbarch);
11201 return
11202 value_from_longest (type,
11203 (value_less (arg1, arg3)
11204 || value_equal (arg1, arg3))
11205 && (value_less (arg2, arg1)
11206 || value_equal (arg2, arg1)));
11207
11208 case TERNOP_IN_RANGE:
11209 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11210 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11211 arg3 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11212
11213 if (noside == EVAL_SKIP)
11214 goto nosideret;
11215
11216 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
11217 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg3);
11218 type = language_bool_type (exp->language_defn, exp->gdbarch);
11219 return
11220 value_from_longest (type,
11221 (value_less (arg1, arg3)
11222 || value_equal (arg1, arg3))
11223 && (value_less (arg2, arg1)
11224 || value_equal (arg2, arg1)));
11225
11226 case OP_ATR_FIRST:
11227 case OP_ATR_LAST:
11228 case OP_ATR_LENGTH:
11229 {
11230 struct type *type_arg;
11231
11232 if (exp->elts[*pos].opcode == OP_TYPE)
11233 {
11234 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11235 arg1 = NULL;
11236 type_arg = check_typedef (exp->elts[pc + 2].type);
11237 }
11238 else
11239 {
11240 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11241 type_arg = NULL;
11242 }
11243
11244 if (exp->elts[*pos].opcode != OP_LONG)
11245 error (_("Invalid operand to '%s"), ada_attribute_name (op));
11246 tem = longest_to_int (exp->elts[*pos + 2].longconst);
11247 *pos += 4;
11248
11249 if (noside == EVAL_SKIP)
11250 goto nosideret;
11251
11252 if (type_arg == NULL)
11253 {
11254 arg1 = ada_coerce_ref (arg1);
11255
11256 if (ada_is_constrained_packed_array_type (value_type (arg1)))
11257 arg1 = ada_coerce_to_simple_array (arg1);
11258
11259 if (op == OP_ATR_LENGTH)
11260 type = builtin_type (exp->gdbarch)->builtin_int;
11261 else
11262 {
11263 type = ada_index_type (value_type (arg1), tem,
11264 ada_attribute_name (op));
11265 if (type == NULL)
11266 type = builtin_type (exp->gdbarch)->builtin_int;
11267 }
11268
11269 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11270 return allocate_value (type);
11271
11272 switch (op)
11273 {
11274 default: /* Should never happen. */
11275 error (_("unexpected attribute encountered"));
11276 case OP_ATR_FIRST:
11277 return value_from_longest
11278 (type, ada_array_bound (arg1, tem, 0));
11279 case OP_ATR_LAST:
11280 return value_from_longest
11281 (type, ada_array_bound (arg1, tem, 1));
11282 case OP_ATR_LENGTH:
11283 return value_from_longest
11284 (type, ada_array_length (arg1, tem));
11285 }
11286 }
11287 else if (discrete_type_p (type_arg))
11288 {
11289 struct type *range_type;
11290 const char *name = ada_type_name (type_arg);
11291
11292 range_type = NULL;
11293 if (name != NULL && TYPE_CODE (type_arg) != TYPE_CODE_ENUM)
11294 range_type = to_fixed_range_type (type_arg, NULL);
11295 if (range_type == NULL)
11296 range_type = type_arg;
11297 switch (op)
11298 {
11299 default:
11300 error (_("unexpected attribute encountered"));
11301 case OP_ATR_FIRST:
11302 return value_from_longest
11303 (range_type, ada_discrete_type_low_bound (range_type));
11304 case OP_ATR_LAST:
11305 return value_from_longest
11306 (range_type, ada_discrete_type_high_bound (range_type));
11307 case OP_ATR_LENGTH:
11308 error (_("the 'length attribute applies only to array types"));
11309 }
11310 }
11311 else if (TYPE_CODE (type_arg) == TYPE_CODE_FLT)
11312 error (_("unimplemented type attribute"));
11313 else
11314 {
11315 LONGEST low, high;
11316
11317 if (ada_is_constrained_packed_array_type (type_arg))
11318 type_arg = decode_constrained_packed_array_type (type_arg);
11319
11320 if (op == OP_ATR_LENGTH)
11321 type = builtin_type (exp->gdbarch)->builtin_int;
11322 else
11323 {
11324 type = ada_index_type (type_arg, tem, ada_attribute_name (op));
11325 if (type == NULL)
11326 type = builtin_type (exp->gdbarch)->builtin_int;
11327 }
11328
11329 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11330 return allocate_value (type);
11331
11332 switch (op)
11333 {
11334 default:
11335 error (_("unexpected attribute encountered"));
11336 case OP_ATR_FIRST:
11337 low = ada_array_bound_from_type (type_arg, tem, 0);
11338 return value_from_longest (type, low);
11339 case OP_ATR_LAST:
11340 high = ada_array_bound_from_type (type_arg, tem, 1);
11341 return value_from_longest (type, high);
11342 case OP_ATR_LENGTH:
11343 low = ada_array_bound_from_type (type_arg, tem, 0);
11344 high = ada_array_bound_from_type (type_arg, tem, 1);
11345 return value_from_longest (type, high - low + 1);
11346 }
11347 }
11348 }
11349
11350 case OP_ATR_TAG:
11351 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11352 if (noside == EVAL_SKIP)
11353 goto nosideret;
11354
11355 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11356 return value_zero (ada_tag_type (arg1), not_lval);
11357
11358 return ada_value_tag (arg1);
11359
11360 case OP_ATR_MIN:
11361 case OP_ATR_MAX:
11362 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11363 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11364 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11365 if (noside == EVAL_SKIP)
11366 goto nosideret;
11367 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11368 return value_zero (value_type (arg1), not_lval);
11369 else
11370 {
11371 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
11372 return value_binop (arg1, arg2,
11373 op == OP_ATR_MIN ? BINOP_MIN : BINOP_MAX);
11374 }
11375
11376 case OP_ATR_MODULUS:
11377 {
11378 struct type *type_arg = check_typedef (exp->elts[pc + 2].type);
11379
11380 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11381 if (noside == EVAL_SKIP)
11382 goto nosideret;
11383
11384 if (!ada_is_modular_type (type_arg))
11385 error (_("'modulus must be applied to modular type"));
11386
11387 return value_from_longest (TYPE_TARGET_TYPE (type_arg),
11388 ada_modulus (type_arg));
11389 }
11390
11391
11392 case OP_ATR_POS:
11393 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11394 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11395 if (noside == EVAL_SKIP)
11396 goto nosideret;
11397 type = builtin_type (exp->gdbarch)->builtin_int;
11398 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11399 return value_zero (type, not_lval);
11400 else
11401 return value_pos_atr (type, arg1);
11402
11403 case OP_ATR_SIZE:
11404 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11405 type = value_type (arg1);
11406
11407 /* If the argument is a reference, then dereference its type, since
11408 the user is really asking for the size of the actual object,
11409 not the size of the pointer. */
11410 if (TYPE_CODE (type) == TYPE_CODE_REF)
11411 type = TYPE_TARGET_TYPE (type);
11412
11413 if (noside == EVAL_SKIP)
11414 goto nosideret;
11415 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11416 return value_zero (builtin_type (exp->gdbarch)->builtin_int, not_lval);
11417 else
11418 return value_from_longest (builtin_type (exp->gdbarch)->builtin_int,
11419 TARGET_CHAR_BIT * TYPE_LENGTH (type));
11420
11421 case OP_ATR_VAL:
11422 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11423 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11424 type = exp->elts[pc + 2].type;
11425 if (noside == EVAL_SKIP)
11426 goto nosideret;
11427 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11428 return value_zero (type, not_lval);
11429 else
11430 return value_val_atr (type, arg1);
11431
11432 case BINOP_EXP:
11433 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11434 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11435 if (noside == EVAL_SKIP)
11436 goto nosideret;
11437 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11438 return value_zero (value_type (arg1), not_lval);
11439 else
11440 {
11441 /* For integer exponentiation operations,
11442 only promote the first argument. */
11443 if (is_integral_type (value_type (arg2)))
11444 unop_promote (exp->language_defn, exp->gdbarch, &arg1);
11445 else
11446 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
11447
11448 return value_binop (arg1, arg2, op);
11449 }
11450
11451 case UNOP_PLUS:
11452 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11453 if (noside == EVAL_SKIP)
11454 goto nosideret;
11455 else
11456 return arg1;
11457
11458 case UNOP_ABS:
11459 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11460 if (noside == EVAL_SKIP)
11461 goto nosideret;
11462 unop_promote (exp->language_defn, exp->gdbarch, &arg1);
11463 if (value_less (arg1, value_zero (value_type (arg1), not_lval)))
11464 return value_neg (arg1);
11465 else
11466 return arg1;
11467
11468 case UNOP_IND:
11469 preeval_pos = *pos;
11470 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11471 if (noside == EVAL_SKIP)
11472 goto nosideret;
11473 type = ada_check_typedef (value_type (arg1));
11474 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11475 {
11476 if (ada_is_array_descriptor_type (type))
11477 /* GDB allows dereferencing GNAT array descriptors. */
11478 {
11479 struct type *arrType = ada_type_of_array (arg1, 0);
11480
11481 if (arrType == NULL)
11482 error (_("Attempt to dereference null array pointer."));
11483 return value_at_lazy (arrType, 0);
11484 }
11485 else if (TYPE_CODE (type) == TYPE_CODE_PTR
11486 || TYPE_CODE (type) == TYPE_CODE_REF
11487 /* In C you can dereference an array to get the 1st elt. */
11488 || TYPE_CODE (type) == TYPE_CODE_ARRAY)
11489 {
11490 /* As mentioned in the OP_VAR_VALUE case, tagged types can
11491 only be determined by inspecting the object's tag.
11492 This means that we need to evaluate completely the
11493 expression in order to get its type. */
11494
11495 if ((TYPE_CODE (type) == TYPE_CODE_REF
11496 || TYPE_CODE (type) == TYPE_CODE_PTR)
11497 && ada_is_tagged_type (TYPE_TARGET_TYPE (type), 0))
11498 {
11499 arg1 = evaluate_subexp (NULL_TYPE, exp, &preeval_pos,
11500 EVAL_NORMAL);
11501 type = value_type (ada_value_ind (arg1));
11502 }
11503 else
11504 {
11505 type = to_static_fixed_type
11506 (ada_aligned_type
11507 (ada_check_typedef (TYPE_TARGET_TYPE (type))));
11508 }
11509 ada_ensure_varsize_limit (type);
11510 return value_zero (type, lval_memory);
11511 }
11512 else if (TYPE_CODE (type) == TYPE_CODE_INT)
11513 {
11514 /* GDB allows dereferencing an int. */
11515 if (expect_type == NULL)
11516 return value_zero (builtin_type (exp->gdbarch)->builtin_int,
11517 lval_memory);
11518 else
11519 {
11520 expect_type =
11521 to_static_fixed_type (ada_aligned_type (expect_type));
11522 return value_zero (expect_type, lval_memory);
11523 }
11524 }
11525 else
11526 error (_("Attempt to take contents of a non-pointer value."));
11527 }
11528 arg1 = ada_coerce_ref (arg1); /* FIXME: What is this for?? */
11529 type = ada_check_typedef (value_type (arg1));
11530
11531 if (TYPE_CODE (type) == TYPE_CODE_INT)
11532 /* GDB allows dereferencing an int. If we were given
11533 the expect_type, then use that as the target type.
11534 Otherwise, assume that the target type is an int. */
11535 {
11536 if (expect_type != NULL)
11537 return ada_value_ind (value_cast (lookup_pointer_type (expect_type),
11538 arg1));
11539 else
11540 return value_at_lazy (builtin_type (exp->gdbarch)->builtin_int,
11541 (CORE_ADDR) value_as_address (arg1));
11542 }
11543
11544 if (ada_is_array_descriptor_type (type))
11545 /* GDB allows dereferencing GNAT array descriptors. */
11546 return ada_coerce_to_simple_array (arg1);
11547 else
11548 return ada_value_ind (arg1);
11549
11550 case STRUCTOP_STRUCT:
11551 tem = longest_to_int (exp->elts[pc + 1].longconst);
11552 (*pos) += 3 + BYTES_TO_EXP_ELEM (tem + 1);
11553 preeval_pos = *pos;
11554 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11555 if (noside == EVAL_SKIP)
11556 goto nosideret;
11557 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11558 {
11559 struct type *type1 = value_type (arg1);
11560
11561 if (ada_is_tagged_type (type1, 1))
11562 {
11563 type = ada_lookup_struct_elt_type (type1,
11564 &exp->elts[pc + 2].string,
11565 1, 1);
11566
11567 /* If the field is not found, check if it exists in the
11568 extension of this object's type. This means that we
11569 need to evaluate completely the expression. */
11570
11571 if (type == NULL)
11572 {
11573 arg1 = evaluate_subexp (NULL_TYPE, exp, &preeval_pos,
11574 EVAL_NORMAL);
11575 arg1 = ada_value_struct_elt (arg1,
11576 &exp->elts[pc + 2].string,
11577 0);
11578 arg1 = unwrap_value (arg1);
11579 type = value_type (ada_to_fixed_value (arg1));
11580 }
11581 }
11582 else
11583 type =
11584 ada_lookup_struct_elt_type (type1, &exp->elts[pc + 2].string, 1,
11585 0);
11586
11587 return value_zero (ada_aligned_type (type), lval_memory);
11588 }
11589 else
11590 {
11591 arg1 = ada_value_struct_elt (arg1, &exp->elts[pc + 2].string, 0);
11592 arg1 = unwrap_value (arg1);
11593 return ada_to_fixed_value (arg1);
11594 }
11595
11596 case OP_TYPE:
11597 /* The value is not supposed to be used. This is here to make it
11598 easier to accommodate expressions that contain types. */
11599 (*pos) += 2;
11600 if (noside == EVAL_SKIP)
11601 goto nosideret;
11602 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11603 return allocate_value (exp->elts[pc + 1].type);
11604 else
11605 error (_("Attempt to use a type name as an expression"));
11606
11607 case OP_AGGREGATE:
11608 case OP_CHOICES:
11609 case OP_OTHERS:
11610 case OP_DISCRETE_RANGE:
11611 case OP_POSITIONAL:
11612 case OP_NAME:
11613 if (noside == EVAL_NORMAL)
11614 switch (op)
11615 {
11616 case OP_NAME:
11617 error (_("Undefined name, ambiguous name, or renaming used in "
11618 "component association: %s."), &exp->elts[pc+2].string);
11619 case OP_AGGREGATE:
11620 error (_("Aggregates only allowed on the right of an assignment"));
11621 default:
11622 internal_error (__FILE__, __LINE__,
11623 _("aggregate apparently mangled"));
11624 }
11625
11626 ada_forward_operator_length (exp, pc, &oplen, &nargs);
11627 *pos += oplen - 1;
11628 for (tem = 0; tem < nargs; tem += 1)
11629 ada_evaluate_subexp (NULL, exp, pos, noside);
11630 goto nosideret;
11631 }
11632
11633 nosideret:
11634 return eval_skip_value (exp);
11635 }
11636 \f
11637
11638 /* Fixed point */
11639
11640 /* If TYPE encodes an Ada fixed-point type, return the suffix of the
11641 type name that encodes the 'small and 'delta information.
11642 Otherwise, return NULL. */
11643
11644 static const char *
11645 fixed_type_info (struct type *type)
11646 {
11647 const char *name = ada_type_name (type);
11648 enum type_code code = (type == NULL) ? TYPE_CODE_UNDEF : TYPE_CODE (type);
11649
11650 if ((code == TYPE_CODE_INT || code == TYPE_CODE_RANGE) && name != NULL)
11651 {
11652 const char *tail = strstr (name, "___XF_");
11653
11654 if (tail == NULL)
11655 return NULL;
11656 else
11657 return tail + 5;
11658 }
11659 else if (code == TYPE_CODE_RANGE && TYPE_TARGET_TYPE (type) != type)
11660 return fixed_type_info (TYPE_TARGET_TYPE (type));
11661 else
11662 return NULL;
11663 }
11664
11665 /* Returns non-zero iff TYPE represents an Ada fixed-point type. */
11666
11667 int
11668 ada_is_fixed_point_type (struct type *type)
11669 {
11670 return fixed_type_info (type) != NULL;
11671 }
11672
11673 /* Return non-zero iff TYPE represents a System.Address type. */
11674
11675 int
11676 ada_is_system_address_type (struct type *type)
11677 {
11678 return (TYPE_NAME (type)
11679 && strcmp (TYPE_NAME (type), "system__address") == 0);
11680 }
11681
11682 /* Assuming that TYPE is the representation of an Ada fixed-point
11683 type, return the target floating-point type to be used to represent
11684 of this type during internal computation. */
11685
11686 static struct type *
11687 ada_scaling_type (struct type *type)
11688 {
11689 return builtin_type (get_type_arch (type))->builtin_long_double;
11690 }
11691
11692 /* Assuming that TYPE is the representation of an Ada fixed-point
11693 type, return its delta, or NULL if the type is malformed and the
11694 delta cannot be determined. */
11695
11696 struct value *
11697 ada_delta (struct type *type)
11698 {
11699 const char *encoding = fixed_type_info (type);
11700 struct type *scale_type = ada_scaling_type (type);
11701
11702 long long num, den;
11703
11704 if (sscanf (encoding, "_%lld_%lld", &num, &den) < 2)
11705 return nullptr;
11706 else
11707 return value_binop (value_from_longest (scale_type, num),
11708 value_from_longest (scale_type, den), BINOP_DIV);
11709 }
11710
11711 /* Assuming that ada_is_fixed_point_type (TYPE), return the scaling
11712 factor ('SMALL value) associated with the type. */
11713
11714 struct value *
11715 ada_scaling_factor (struct type *type)
11716 {
11717 const char *encoding = fixed_type_info (type);
11718 struct type *scale_type = ada_scaling_type (type);
11719
11720 long long num0, den0, num1, den1;
11721 int n;
11722
11723 n = sscanf (encoding, "_%lld_%lld_%lld_%lld",
11724 &num0, &den0, &num1, &den1);
11725
11726 if (n < 2)
11727 return value_from_longest (scale_type, 1);
11728 else if (n == 4)
11729 return value_binop (value_from_longest (scale_type, num1),
11730 value_from_longest (scale_type, den1), BINOP_DIV);
11731 else
11732 return value_binop (value_from_longest (scale_type, num0),
11733 value_from_longest (scale_type, den0), BINOP_DIV);
11734 }
11735
11736 \f
11737
11738 /* Range types */
11739
11740 /* Scan STR beginning at position K for a discriminant name, and
11741 return the value of that discriminant field of DVAL in *PX. If
11742 PNEW_K is not null, put the position of the character beyond the
11743 name scanned in *PNEW_K. Return 1 if successful; return 0 and do
11744 not alter *PX and *PNEW_K if unsuccessful. */
11745
11746 static int
11747 scan_discrim_bound (const char *str, int k, struct value *dval, LONGEST * px,
11748 int *pnew_k)
11749 {
11750 static char *bound_buffer = NULL;
11751 static size_t bound_buffer_len = 0;
11752 const char *pstart, *pend, *bound;
11753 struct value *bound_val;
11754
11755 if (dval == NULL || str == NULL || str[k] == '\0')
11756 return 0;
11757
11758 pstart = str + k;
11759 pend = strstr (pstart, "__");
11760 if (pend == NULL)
11761 {
11762 bound = pstart;
11763 k += strlen (bound);
11764 }
11765 else
11766 {
11767 int len = pend - pstart;
11768
11769 /* Strip __ and beyond. */
11770 GROW_VECT (bound_buffer, bound_buffer_len, len + 1);
11771 strncpy (bound_buffer, pstart, len);
11772 bound_buffer[len] = '\0';
11773
11774 bound = bound_buffer;
11775 k = pend - str;
11776 }
11777
11778 bound_val = ada_search_struct_field (bound, dval, 0, value_type (dval));
11779 if (bound_val == NULL)
11780 return 0;
11781
11782 *px = value_as_long (bound_val);
11783 if (pnew_k != NULL)
11784 *pnew_k = k;
11785 return 1;
11786 }
11787
11788 /* Value of variable named NAME in the current environment. If
11789 no such variable found, then if ERR_MSG is null, returns 0, and
11790 otherwise causes an error with message ERR_MSG. */
11791
11792 static struct value *
11793 get_var_value (const char *name, const char *err_msg)
11794 {
11795 lookup_name_info lookup_name (name, symbol_name_match_type::FULL);
11796
11797 struct block_symbol *syms;
11798 int nsyms = ada_lookup_symbol_list_worker (lookup_name,
11799 get_selected_block (0),
11800 VAR_DOMAIN, &syms, 1);
11801 struct cleanup *old_chain = make_cleanup (xfree, syms);
11802
11803 if (nsyms != 1)
11804 {
11805 do_cleanups (old_chain);
11806 if (err_msg == NULL)
11807 return 0;
11808 else
11809 error (("%s"), err_msg);
11810 }
11811
11812 struct value *result = value_of_variable (syms[0].symbol, syms[0].block);
11813 do_cleanups (old_chain);
11814 return result;
11815 }
11816
11817 /* Value of integer variable named NAME in the current environment.
11818 If no such variable is found, returns false. Otherwise, sets VALUE
11819 to the variable's value and returns true. */
11820
11821 bool
11822 get_int_var_value (const char *name, LONGEST &value)
11823 {
11824 struct value *var_val = get_var_value (name, 0);
11825
11826 if (var_val == 0)
11827 return false;
11828
11829 value = value_as_long (var_val);
11830 return true;
11831 }
11832
11833
11834 /* Return a range type whose base type is that of the range type named
11835 NAME in the current environment, and whose bounds are calculated
11836 from NAME according to the GNAT range encoding conventions.
11837 Extract discriminant values, if needed, from DVAL. ORIG_TYPE is the
11838 corresponding range type from debug information; fall back to using it
11839 if symbol lookup fails. If a new type must be created, allocate it
11840 like ORIG_TYPE was. The bounds information, in general, is encoded
11841 in NAME, the base type given in the named range type. */
11842
11843 static struct type *
11844 to_fixed_range_type (struct type *raw_type, struct value *dval)
11845 {
11846 const char *name;
11847 struct type *base_type;
11848 const char *subtype_info;
11849
11850 gdb_assert (raw_type != NULL);
11851 gdb_assert (TYPE_NAME (raw_type) != NULL);
11852
11853 if (TYPE_CODE (raw_type) == TYPE_CODE_RANGE)
11854 base_type = TYPE_TARGET_TYPE (raw_type);
11855 else
11856 base_type = raw_type;
11857
11858 name = TYPE_NAME (raw_type);
11859 subtype_info = strstr (name, "___XD");
11860 if (subtype_info == NULL)
11861 {
11862 LONGEST L = ada_discrete_type_low_bound (raw_type);
11863 LONGEST U = ada_discrete_type_high_bound (raw_type);
11864
11865 if (L < INT_MIN || U > INT_MAX)
11866 return raw_type;
11867 else
11868 return create_static_range_type (alloc_type_copy (raw_type), raw_type,
11869 L, U);
11870 }
11871 else
11872 {
11873 static char *name_buf = NULL;
11874 static size_t name_len = 0;
11875 int prefix_len = subtype_info - name;
11876 LONGEST L, U;
11877 struct type *type;
11878 const char *bounds_str;
11879 int n;
11880
11881 GROW_VECT (name_buf, name_len, prefix_len + 5);
11882 strncpy (name_buf, name, prefix_len);
11883 name_buf[prefix_len] = '\0';
11884
11885 subtype_info += 5;
11886 bounds_str = strchr (subtype_info, '_');
11887 n = 1;
11888
11889 if (*subtype_info == 'L')
11890 {
11891 if (!ada_scan_number (bounds_str, n, &L, &n)
11892 && !scan_discrim_bound (bounds_str, n, dval, &L, &n))
11893 return raw_type;
11894 if (bounds_str[n] == '_')
11895 n += 2;
11896 else if (bounds_str[n] == '.') /* FIXME? SGI Workshop kludge. */
11897 n += 1;
11898 subtype_info += 1;
11899 }
11900 else
11901 {
11902 strcpy (name_buf + prefix_len, "___L");
11903 if (!get_int_var_value (name_buf, L))
11904 {
11905 lim_warning (_("Unknown lower bound, using 1."));
11906 L = 1;
11907 }
11908 }
11909
11910 if (*subtype_info == 'U')
11911 {
11912 if (!ada_scan_number (bounds_str, n, &U, &n)
11913 && !scan_discrim_bound (bounds_str, n, dval, &U, &n))
11914 return raw_type;
11915 }
11916 else
11917 {
11918 strcpy (name_buf + prefix_len, "___U");
11919 if (!get_int_var_value (name_buf, U))
11920 {
11921 lim_warning (_("Unknown upper bound, using %ld."), (long) L);
11922 U = L;
11923 }
11924 }
11925
11926 type = create_static_range_type (alloc_type_copy (raw_type),
11927 base_type, L, U);
11928 /* create_static_range_type alters the resulting type's length
11929 to match the size of the base_type, which is not what we want.
11930 Set it back to the original range type's length. */
11931 TYPE_LENGTH (type) = TYPE_LENGTH (raw_type);
11932 TYPE_NAME (type) = name;
11933 return type;
11934 }
11935 }
11936
11937 /* True iff NAME is the name of a range type. */
11938
11939 int
11940 ada_is_range_type_name (const char *name)
11941 {
11942 return (name != NULL && strstr (name, "___XD"));
11943 }
11944 \f
11945
11946 /* Modular types */
11947
11948 /* True iff TYPE is an Ada modular type. */
11949
11950 int
11951 ada_is_modular_type (struct type *type)
11952 {
11953 struct type *subranged_type = get_base_type (type);
11954
11955 return (subranged_type != NULL && TYPE_CODE (type) == TYPE_CODE_RANGE
11956 && TYPE_CODE (subranged_type) == TYPE_CODE_INT
11957 && TYPE_UNSIGNED (subranged_type));
11958 }
11959
11960 /* Assuming ada_is_modular_type (TYPE), the modulus of TYPE. */
11961
11962 ULONGEST
11963 ada_modulus (struct type *type)
11964 {
11965 return (ULONGEST) TYPE_HIGH_BOUND (type) + 1;
11966 }
11967 \f
11968
11969 /* Ada exception catchpoint support:
11970 ---------------------------------
11971
11972 We support 3 kinds of exception catchpoints:
11973 . catchpoints on Ada exceptions
11974 . catchpoints on unhandled Ada exceptions
11975 . catchpoints on failed assertions
11976
11977 Exceptions raised during failed assertions, or unhandled exceptions
11978 could perfectly be caught with the general catchpoint on Ada exceptions.
11979 However, we can easily differentiate these two special cases, and having
11980 the option to distinguish these two cases from the rest can be useful
11981 to zero-in on certain situations.
11982
11983 Exception catchpoints are a specialized form of breakpoint,
11984 since they rely on inserting breakpoints inside known routines
11985 of the GNAT runtime. The implementation therefore uses a standard
11986 breakpoint structure of the BP_BREAKPOINT type, but with its own set
11987 of breakpoint_ops.
11988
11989 Support in the runtime for exception catchpoints have been changed
11990 a few times already, and these changes affect the implementation
11991 of these catchpoints. In order to be able to support several
11992 variants of the runtime, we use a sniffer that will determine
11993 the runtime variant used by the program being debugged. */
11994
11995 /* Ada's standard exceptions.
11996
11997 The Ada 83 standard also defined Numeric_Error. But there so many
11998 situations where it was unclear from the Ada 83 Reference Manual
11999 (RM) whether Constraint_Error or Numeric_Error should be raised,
12000 that the ARG (Ada Rapporteur Group) eventually issued a Binding
12001 Interpretation saying that anytime the RM says that Numeric_Error
12002 should be raised, the implementation may raise Constraint_Error.
12003 Ada 95 went one step further and pretty much removed Numeric_Error
12004 from the list of standard exceptions (it made it a renaming of
12005 Constraint_Error, to help preserve compatibility when compiling
12006 an Ada83 compiler). As such, we do not include Numeric_Error from
12007 this list of standard exceptions. */
12008
12009 static const char *standard_exc[] = {
12010 "constraint_error",
12011 "program_error",
12012 "storage_error",
12013 "tasking_error"
12014 };
12015
12016 typedef CORE_ADDR (ada_unhandled_exception_name_addr_ftype) (void);
12017
12018 /* A structure that describes how to support exception catchpoints
12019 for a given executable. */
12020
12021 struct exception_support_info
12022 {
12023 /* The name of the symbol to break on in order to insert
12024 a catchpoint on exceptions. */
12025 const char *catch_exception_sym;
12026
12027 /* The name of the symbol to break on in order to insert
12028 a catchpoint on unhandled exceptions. */
12029 const char *catch_exception_unhandled_sym;
12030
12031 /* The name of the symbol to break on in order to insert
12032 a catchpoint on failed assertions. */
12033 const char *catch_assert_sym;
12034
12035 /* The name of the symbol to break on in order to insert
12036 a catchpoint on exception handling. */
12037 const char *catch_handlers_sym;
12038
12039 /* Assuming that the inferior just triggered an unhandled exception
12040 catchpoint, this function is responsible for returning the address
12041 in inferior memory where the name of that exception is stored.
12042 Return zero if the address could not be computed. */
12043 ada_unhandled_exception_name_addr_ftype *unhandled_exception_name_addr;
12044 };
12045
12046 static CORE_ADDR ada_unhandled_exception_name_addr (void);
12047 static CORE_ADDR ada_unhandled_exception_name_addr_from_raise (void);
12048
12049 /* The following exception support info structure describes how to
12050 implement exception catchpoints with the latest version of the
12051 Ada runtime (as of 2007-03-06). */
12052
12053 static const struct exception_support_info default_exception_support_info =
12054 {
12055 "__gnat_debug_raise_exception", /* catch_exception_sym */
12056 "__gnat_unhandled_exception", /* catch_exception_unhandled_sym */
12057 "__gnat_debug_raise_assert_failure", /* catch_assert_sym */
12058 "__gnat_begin_handler", /* catch_handlers_sym */
12059 ada_unhandled_exception_name_addr
12060 };
12061
12062 /* The following exception support info structure describes how to
12063 implement exception catchpoints with a slightly older version
12064 of the Ada runtime. */
12065
12066 static const struct exception_support_info exception_support_info_fallback =
12067 {
12068 "__gnat_raise_nodefer_with_msg", /* catch_exception_sym */
12069 "__gnat_unhandled_exception", /* catch_exception_unhandled_sym */
12070 "system__assertions__raise_assert_failure", /* catch_assert_sym */
12071 "__gnat_begin_handler", /* catch_handlers_sym */
12072 ada_unhandled_exception_name_addr_from_raise
12073 };
12074
12075 /* Return nonzero if we can detect the exception support routines
12076 described in EINFO.
12077
12078 This function errors out if an abnormal situation is detected
12079 (for instance, if we find the exception support routines, but
12080 that support is found to be incomplete). */
12081
12082 static int
12083 ada_has_this_exception_support (const struct exception_support_info *einfo)
12084 {
12085 struct symbol *sym;
12086
12087 /* The symbol we're looking up is provided by a unit in the GNAT runtime
12088 that should be compiled with debugging information. As a result, we
12089 expect to find that symbol in the symtabs. */
12090
12091 sym = standard_lookup (einfo->catch_exception_sym, NULL, VAR_DOMAIN);
12092 if (sym == NULL)
12093 {
12094 /* Perhaps we did not find our symbol because the Ada runtime was
12095 compiled without debugging info, or simply stripped of it.
12096 It happens on some GNU/Linux distributions for instance, where
12097 users have to install a separate debug package in order to get
12098 the runtime's debugging info. In that situation, let the user
12099 know why we cannot insert an Ada exception catchpoint.
12100
12101 Note: Just for the purpose of inserting our Ada exception
12102 catchpoint, we could rely purely on the associated minimal symbol.
12103 But we would be operating in degraded mode anyway, since we are
12104 still lacking the debugging info needed later on to extract
12105 the name of the exception being raised (this name is printed in
12106 the catchpoint message, and is also used when trying to catch
12107 a specific exception). We do not handle this case for now. */
12108 struct bound_minimal_symbol msym
12109 = lookup_minimal_symbol (einfo->catch_exception_sym, NULL, NULL);
12110
12111 if (msym.minsym && MSYMBOL_TYPE (msym.minsym) != mst_solib_trampoline)
12112 error (_("Your Ada runtime appears to be missing some debugging "
12113 "information.\nCannot insert Ada exception catchpoint "
12114 "in this configuration."));
12115
12116 return 0;
12117 }
12118
12119 /* Make sure that the symbol we found corresponds to a function. */
12120
12121 if (SYMBOL_CLASS (sym) != LOC_BLOCK)
12122 error (_("Symbol \"%s\" is not a function (class = %d)"),
12123 SYMBOL_LINKAGE_NAME (sym), SYMBOL_CLASS (sym));
12124
12125 return 1;
12126 }
12127
12128 /* Inspect the Ada runtime and determine which exception info structure
12129 should be used to provide support for exception catchpoints.
12130
12131 This function will always set the per-inferior exception_info,
12132 or raise an error. */
12133
12134 static void
12135 ada_exception_support_info_sniffer (void)
12136 {
12137 struct ada_inferior_data *data = get_ada_inferior_data (current_inferior ());
12138
12139 /* If the exception info is already known, then no need to recompute it. */
12140 if (data->exception_info != NULL)
12141 return;
12142
12143 /* Check the latest (default) exception support info. */
12144 if (ada_has_this_exception_support (&default_exception_support_info))
12145 {
12146 data->exception_info = &default_exception_support_info;
12147 return;
12148 }
12149
12150 /* Try our fallback exception suport info. */
12151 if (ada_has_this_exception_support (&exception_support_info_fallback))
12152 {
12153 data->exception_info = &exception_support_info_fallback;
12154 return;
12155 }
12156
12157 /* Sometimes, it is normal for us to not be able to find the routine
12158 we are looking for. This happens when the program is linked with
12159 the shared version of the GNAT runtime, and the program has not been
12160 started yet. Inform the user of these two possible causes if
12161 applicable. */
12162
12163 if (ada_update_initial_language (language_unknown) != language_ada)
12164 error (_("Unable to insert catchpoint. Is this an Ada main program?"));
12165
12166 /* If the symbol does not exist, then check that the program is
12167 already started, to make sure that shared libraries have been
12168 loaded. If it is not started, this may mean that the symbol is
12169 in a shared library. */
12170
12171 if (ptid_get_pid (inferior_ptid) == 0)
12172 error (_("Unable to insert catchpoint. Try to start the program first."));
12173
12174 /* At this point, we know that we are debugging an Ada program and
12175 that the inferior has been started, but we still are not able to
12176 find the run-time symbols. That can mean that we are in
12177 configurable run time mode, or that a-except as been optimized
12178 out by the linker... In any case, at this point it is not worth
12179 supporting this feature. */
12180
12181 error (_("Cannot insert Ada exception catchpoints in this configuration."));
12182 }
12183
12184 /* True iff FRAME is very likely to be that of a function that is
12185 part of the runtime system. This is all very heuristic, but is
12186 intended to be used as advice as to what frames are uninteresting
12187 to most users. */
12188
12189 static int
12190 is_known_support_routine (struct frame_info *frame)
12191 {
12192 enum language func_lang;
12193 int i;
12194 const char *fullname;
12195
12196 /* If this code does not have any debugging information (no symtab),
12197 This cannot be any user code. */
12198
12199 symtab_and_line sal = find_frame_sal (frame);
12200 if (sal.symtab == NULL)
12201 return 1;
12202
12203 /* If there is a symtab, but the associated source file cannot be
12204 located, then assume this is not user code: Selecting a frame
12205 for which we cannot display the code would not be very helpful
12206 for the user. This should also take care of case such as VxWorks
12207 where the kernel has some debugging info provided for a few units. */
12208
12209 fullname = symtab_to_fullname (sal.symtab);
12210 if (access (fullname, R_OK) != 0)
12211 return 1;
12212
12213 /* Check the unit filename againt the Ada runtime file naming.
12214 We also check the name of the objfile against the name of some
12215 known system libraries that sometimes come with debugging info
12216 too. */
12217
12218 for (i = 0; known_runtime_file_name_patterns[i] != NULL; i += 1)
12219 {
12220 re_comp (known_runtime_file_name_patterns[i]);
12221 if (re_exec (lbasename (sal.symtab->filename)))
12222 return 1;
12223 if (SYMTAB_OBJFILE (sal.symtab) != NULL
12224 && re_exec (objfile_name (SYMTAB_OBJFILE (sal.symtab))))
12225 return 1;
12226 }
12227
12228 /* Check whether the function is a GNAT-generated entity. */
12229
12230 gdb::unique_xmalloc_ptr<char> func_name
12231 = find_frame_funname (frame, &func_lang, NULL);
12232 if (func_name == NULL)
12233 return 1;
12234
12235 for (i = 0; known_auxiliary_function_name_patterns[i] != NULL; i += 1)
12236 {
12237 re_comp (known_auxiliary_function_name_patterns[i]);
12238 if (re_exec (func_name.get ()))
12239 return 1;
12240 }
12241
12242 return 0;
12243 }
12244
12245 /* Find the first frame that contains debugging information and that is not
12246 part of the Ada run-time, starting from FI and moving upward. */
12247
12248 void
12249 ada_find_printable_frame (struct frame_info *fi)
12250 {
12251 for (; fi != NULL; fi = get_prev_frame (fi))
12252 {
12253 if (!is_known_support_routine (fi))
12254 {
12255 select_frame (fi);
12256 break;
12257 }
12258 }
12259
12260 }
12261
12262 /* Assuming that the inferior just triggered an unhandled exception
12263 catchpoint, return the address in inferior memory where the name
12264 of the exception is stored.
12265
12266 Return zero if the address could not be computed. */
12267
12268 static CORE_ADDR
12269 ada_unhandled_exception_name_addr (void)
12270 {
12271 return parse_and_eval_address ("e.full_name");
12272 }
12273
12274 /* Same as ada_unhandled_exception_name_addr, except that this function
12275 should be used when the inferior uses an older version of the runtime,
12276 where the exception name needs to be extracted from a specific frame
12277 several frames up in the callstack. */
12278
12279 static CORE_ADDR
12280 ada_unhandled_exception_name_addr_from_raise (void)
12281 {
12282 int frame_level;
12283 struct frame_info *fi;
12284 struct ada_inferior_data *data = get_ada_inferior_data (current_inferior ());
12285
12286 /* To determine the name of this exception, we need to select
12287 the frame corresponding to RAISE_SYM_NAME. This frame is
12288 at least 3 levels up, so we simply skip the first 3 frames
12289 without checking the name of their associated function. */
12290 fi = get_current_frame ();
12291 for (frame_level = 0; frame_level < 3; frame_level += 1)
12292 if (fi != NULL)
12293 fi = get_prev_frame (fi);
12294
12295 while (fi != NULL)
12296 {
12297 enum language func_lang;
12298
12299 gdb::unique_xmalloc_ptr<char> func_name
12300 = find_frame_funname (fi, &func_lang, NULL);
12301 if (func_name != NULL)
12302 {
12303 if (strcmp (func_name.get (),
12304 data->exception_info->catch_exception_sym) == 0)
12305 break; /* We found the frame we were looking for... */
12306 fi = get_prev_frame (fi);
12307 }
12308 }
12309
12310 if (fi == NULL)
12311 return 0;
12312
12313 select_frame (fi);
12314 return parse_and_eval_address ("id.full_name");
12315 }
12316
12317 /* Assuming the inferior just triggered an Ada exception catchpoint
12318 (of any type), return the address in inferior memory where the name
12319 of the exception is stored, if applicable.
12320
12321 Assumes the selected frame is the current frame.
12322
12323 Return zero if the address could not be computed, or if not relevant. */
12324
12325 static CORE_ADDR
12326 ada_exception_name_addr_1 (enum ada_exception_catchpoint_kind ex,
12327 struct breakpoint *b)
12328 {
12329 struct ada_inferior_data *data = get_ada_inferior_data (current_inferior ());
12330
12331 switch (ex)
12332 {
12333 case ada_catch_exception:
12334 return (parse_and_eval_address ("e.full_name"));
12335 break;
12336
12337 case ada_catch_exception_unhandled:
12338 return data->exception_info->unhandled_exception_name_addr ();
12339 break;
12340
12341 case ada_catch_handlers:
12342 return 0; /* The runtimes does not provide access to the exception
12343 name. */
12344 break;
12345
12346 case ada_catch_assert:
12347 return 0; /* Exception name is not relevant in this case. */
12348 break;
12349
12350 default:
12351 internal_error (__FILE__, __LINE__, _("unexpected catchpoint type"));
12352 break;
12353 }
12354
12355 return 0; /* Should never be reached. */
12356 }
12357
12358 /* Assuming the inferior is stopped at an exception catchpoint,
12359 return the message which was associated to the exception, if
12360 available. Return NULL if the message could not be retrieved.
12361
12362 The caller must xfree the string after use.
12363
12364 Note: The exception message can be associated to an exception
12365 either through the use of the Raise_Exception function, or
12366 more simply (Ada 2005 and later), via:
12367
12368 raise Exception_Name with "exception message";
12369
12370 */
12371
12372 static char *
12373 ada_exception_message_1 (void)
12374 {
12375 struct value *e_msg_val;
12376 char *e_msg = NULL;
12377 int e_msg_len;
12378 struct cleanup *cleanups;
12379
12380 /* For runtimes that support this feature, the exception message
12381 is passed as an unbounded string argument called "message". */
12382 e_msg_val = parse_and_eval ("message");
12383 if (e_msg_val == NULL)
12384 return NULL; /* Exception message not supported. */
12385
12386 e_msg_val = ada_coerce_to_simple_array (e_msg_val);
12387 gdb_assert (e_msg_val != NULL);
12388 e_msg_len = TYPE_LENGTH (value_type (e_msg_val));
12389
12390 /* If the message string is empty, then treat it as if there was
12391 no exception message. */
12392 if (e_msg_len <= 0)
12393 return NULL;
12394
12395 e_msg = (char *) xmalloc (e_msg_len + 1);
12396 cleanups = make_cleanup (xfree, e_msg);
12397 read_memory_string (value_address (e_msg_val), e_msg, e_msg_len + 1);
12398 e_msg[e_msg_len] = '\0';
12399
12400 discard_cleanups (cleanups);
12401 return e_msg;
12402 }
12403
12404 /* Same as ada_exception_message_1, except that all exceptions are
12405 contained here (returning NULL instead). */
12406
12407 static char *
12408 ada_exception_message (void)
12409 {
12410 char *e_msg = NULL; /* Avoid a spurious uninitialized warning. */
12411
12412 TRY
12413 {
12414 e_msg = ada_exception_message_1 ();
12415 }
12416 CATCH (e, RETURN_MASK_ERROR)
12417 {
12418 e_msg = NULL;
12419 }
12420 END_CATCH
12421
12422 return e_msg;
12423 }
12424
12425 /* Same as ada_exception_name_addr_1, except that it intercepts and contains
12426 any error that ada_exception_name_addr_1 might cause to be thrown.
12427 When an error is intercepted, a warning with the error message is printed,
12428 and zero is returned. */
12429
12430 static CORE_ADDR
12431 ada_exception_name_addr (enum ada_exception_catchpoint_kind ex,
12432 struct breakpoint *b)
12433 {
12434 CORE_ADDR result = 0;
12435
12436 TRY
12437 {
12438 result = ada_exception_name_addr_1 (ex, b);
12439 }
12440
12441 CATCH (e, RETURN_MASK_ERROR)
12442 {
12443 warning (_("failed to get exception name: %s"), e.message);
12444 return 0;
12445 }
12446 END_CATCH
12447
12448 return result;
12449 }
12450
12451 static char *ada_exception_catchpoint_cond_string
12452 (const char *excep_string,
12453 enum ada_exception_catchpoint_kind ex);
12454
12455 /* Ada catchpoints.
12456
12457 In the case of catchpoints on Ada exceptions, the catchpoint will
12458 stop the target on every exception the program throws. When a user
12459 specifies the name of a specific exception, we translate this
12460 request into a condition expression (in text form), and then parse
12461 it into an expression stored in each of the catchpoint's locations.
12462 We then use this condition to check whether the exception that was
12463 raised is the one the user is interested in. If not, then the
12464 target is resumed again. We store the name of the requested
12465 exception, in order to be able to re-set the condition expression
12466 when symbols change. */
12467
12468 /* An instance of this type is used to represent an Ada catchpoint
12469 breakpoint location. */
12470
12471 class ada_catchpoint_location : public bp_location
12472 {
12473 public:
12474 ada_catchpoint_location (const bp_location_ops *ops, breakpoint *owner)
12475 : bp_location (ops, owner)
12476 {}
12477
12478 /* The condition that checks whether the exception that was raised
12479 is the specific exception the user specified on catchpoint
12480 creation. */
12481 expression_up excep_cond_expr;
12482 };
12483
12484 /* Implement the DTOR method in the bp_location_ops structure for all
12485 Ada exception catchpoint kinds. */
12486
12487 static void
12488 ada_catchpoint_location_dtor (struct bp_location *bl)
12489 {
12490 struct ada_catchpoint_location *al = (struct ada_catchpoint_location *) bl;
12491
12492 al->excep_cond_expr.reset ();
12493 }
12494
12495 /* The vtable to be used in Ada catchpoint locations. */
12496
12497 static const struct bp_location_ops ada_catchpoint_location_ops =
12498 {
12499 ada_catchpoint_location_dtor
12500 };
12501
12502 /* An instance of this type is used to represent an Ada catchpoint. */
12503
12504 struct ada_catchpoint : public breakpoint
12505 {
12506 ~ada_catchpoint () override;
12507
12508 /* The name of the specific exception the user specified. */
12509 char *excep_string;
12510 };
12511
12512 /* Parse the exception condition string in the context of each of the
12513 catchpoint's locations, and store them for later evaluation. */
12514
12515 static void
12516 create_excep_cond_exprs (struct ada_catchpoint *c,
12517 enum ada_exception_catchpoint_kind ex)
12518 {
12519 struct cleanup *old_chain;
12520 struct bp_location *bl;
12521 char *cond_string;
12522
12523 /* Nothing to do if there's no specific exception to catch. */
12524 if (c->excep_string == NULL)
12525 return;
12526
12527 /* Same if there are no locations... */
12528 if (c->loc == NULL)
12529 return;
12530
12531 /* Compute the condition expression in text form, from the specific
12532 expection we want to catch. */
12533 cond_string = ada_exception_catchpoint_cond_string (c->excep_string, ex);
12534 old_chain = make_cleanup (xfree, cond_string);
12535
12536 /* Iterate over all the catchpoint's locations, and parse an
12537 expression for each. */
12538 for (bl = c->loc; bl != NULL; bl = bl->next)
12539 {
12540 struct ada_catchpoint_location *ada_loc
12541 = (struct ada_catchpoint_location *) bl;
12542 expression_up exp;
12543
12544 if (!bl->shlib_disabled)
12545 {
12546 const char *s;
12547
12548 s = cond_string;
12549 TRY
12550 {
12551 exp = parse_exp_1 (&s, bl->address,
12552 block_for_pc (bl->address),
12553 0);
12554 }
12555 CATCH (e, RETURN_MASK_ERROR)
12556 {
12557 warning (_("failed to reevaluate internal exception condition "
12558 "for catchpoint %d: %s"),
12559 c->number, e.message);
12560 }
12561 END_CATCH
12562 }
12563
12564 ada_loc->excep_cond_expr = std::move (exp);
12565 }
12566
12567 do_cleanups (old_chain);
12568 }
12569
12570 /* ada_catchpoint destructor. */
12571
12572 ada_catchpoint::~ada_catchpoint ()
12573 {
12574 xfree (this->excep_string);
12575 }
12576
12577 /* Implement the ALLOCATE_LOCATION method in the breakpoint_ops
12578 structure for all exception catchpoint kinds. */
12579
12580 static struct bp_location *
12581 allocate_location_exception (enum ada_exception_catchpoint_kind ex,
12582 struct breakpoint *self)
12583 {
12584 return new ada_catchpoint_location (&ada_catchpoint_location_ops, self);
12585 }
12586
12587 /* Implement the RE_SET method in the breakpoint_ops structure for all
12588 exception catchpoint kinds. */
12589
12590 static void
12591 re_set_exception (enum ada_exception_catchpoint_kind ex, struct breakpoint *b)
12592 {
12593 struct ada_catchpoint *c = (struct ada_catchpoint *) b;
12594
12595 /* Call the base class's method. This updates the catchpoint's
12596 locations. */
12597 bkpt_breakpoint_ops.re_set (b);
12598
12599 /* Reparse the exception conditional expressions. One for each
12600 location. */
12601 create_excep_cond_exprs (c, ex);
12602 }
12603
12604 /* Returns true if we should stop for this breakpoint hit. If the
12605 user specified a specific exception, we only want to cause a stop
12606 if the program thrown that exception. */
12607
12608 static int
12609 should_stop_exception (const struct bp_location *bl)
12610 {
12611 struct ada_catchpoint *c = (struct ada_catchpoint *) bl->owner;
12612 const struct ada_catchpoint_location *ada_loc
12613 = (const struct ada_catchpoint_location *) bl;
12614 int stop;
12615
12616 /* With no specific exception, should always stop. */
12617 if (c->excep_string == NULL)
12618 return 1;
12619
12620 if (ada_loc->excep_cond_expr == NULL)
12621 {
12622 /* We will have a NULL expression if back when we were creating
12623 the expressions, this location's had failed to parse. */
12624 return 1;
12625 }
12626
12627 stop = 1;
12628 TRY
12629 {
12630 struct value *mark;
12631
12632 mark = value_mark ();
12633 stop = value_true (evaluate_expression (ada_loc->excep_cond_expr.get ()));
12634 value_free_to_mark (mark);
12635 }
12636 CATCH (ex, RETURN_MASK_ALL)
12637 {
12638 exception_fprintf (gdb_stderr, ex,
12639 _("Error in testing exception condition:\n"));
12640 }
12641 END_CATCH
12642
12643 return stop;
12644 }
12645
12646 /* Implement the CHECK_STATUS method in the breakpoint_ops structure
12647 for all exception catchpoint kinds. */
12648
12649 static void
12650 check_status_exception (enum ada_exception_catchpoint_kind ex, bpstat bs)
12651 {
12652 bs->stop = should_stop_exception (bs->bp_location_at);
12653 }
12654
12655 /* Implement the PRINT_IT method in the breakpoint_ops structure
12656 for all exception catchpoint kinds. */
12657
12658 static enum print_stop_action
12659 print_it_exception (enum ada_exception_catchpoint_kind ex, bpstat bs)
12660 {
12661 struct ui_out *uiout = current_uiout;
12662 struct breakpoint *b = bs->breakpoint_at;
12663 char *exception_message;
12664
12665 annotate_catchpoint (b->number);
12666
12667 if (uiout->is_mi_like_p ())
12668 {
12669 uiout->field_string ("reason",
12670 async_reason_lookup (EXEC_ASYNC_BREAKPOINT_HIT));
12671 uiout->field_string ("disp", bpdisp_text (b->disposition));
12672 }
12673
12674 uiout->text (b->disposition == disp_del
12675 ? "\nTemporary catchpoint " : "\nCatchpoint ");
12676 uiout->field_int ("bkptno", b->number);
12677 uiout->text (", ");
12678
12679 /* ada_exception_name_addr relies on the selected frame being the
12680 current frame. Need to do this here because this function may be
12681 called more than once when printing a stop, and below, we'll
12682 select the first frame past the Ada run-time (see
12683 ada_find_printable_frame). */
12684 select_frame (get_current_frame ());
12685
12686 switch (ex)
12687 {
12688 case ada_catch_exception:
12689 case ada_catch_exception_unhandled:
12690 case ada_catch_handlers:
12691 {
12692 const CORE_ADDR addr = ada_exception_name_addr (ex, b);
12693 char exception_name[256];
12694
12695 if (addr != 0)
12696 {
12697 read_memory (addr, (gdb_byte *) exception_name,
12698 sizeof (exception_name) - 1);
12699 exception_name [sizeof (exception_name) - 1] = '\0';
12700 }
12701 else
12702 {
12703 /* For some reason, we were unable to read the exception
12704 name. This could happen if the Runtime was compiled
12705 without debugging info, for instance. In that case,
12706 just replace the exception name by the generic string
12707 "exception" - it will read as "an exception" in the
12708 notification we are about to print. */
12709 memcpy (exception_name, "exception", sizeof ("exception"));
12710 }
12711 /* In the case of unhandled exception breakpoints, we print
12712 the exception name as "unhandled EXCEPTION_NAME", to make
12713 it clearer to the user which kind of catchpoint just got
12714 hit. We used ui_out_text to make sure that this extra
12715 info does not pollute the exception name in the MI case. */
12716 if (ex == ada_catch_exception_unhandled)
12717 uiout->text ("unhandled ");
12718 uiout->field_string ("exception-name", exception_name);
12719 }
12720 break;
12721 case ada_catch_assert:
12722 /* In this case, the name of the exception is not really
12723 important. Just print "failed assertion" to make it clearer
12724 that his program just hit an assertion-failure catchpoint.
12725 We used ui_out_text because this info does not belong in
12726 the MI output. */
12727 uiout->text ("failed assertion");
12728 break;
12729 }
12730
12731 exception_message = ada_exception_message ();
12732 if (exception_message != NULL)
12733 {
12734 struct cleanup *cleanups = make_cleanup (xfree, exception_message);
12735
12736 uiout->text (" (");
12737 uiout->field_string ("exception-message", exception_message);
12738 uiout->text (")");
12739
12740 do_cleanups (cleanups);
12741 }
12742
12743 uiout->text (" at ");
12744 ada_find_printable_frame (get_current_frame ());
12745
12746 return PRINT_SRC_AND_LOC;
12747 }
12748
12749 /* Implement the PRINT_ONE method in the breakpoint_ops structure
12750 for all exception catchpoint kinds. */
12751
12752 static void
12753 print_one_exception (enum ada_exception_catchpoint_kind ex,
12754 struct breakpoint *b, struct bp_location **last_loc)
12755 {
12756 struct ui_out *uiout = current_uiout;
12757 struct ada_catchpoint *c = (struct ada_catchpoint *) b;
12758 struct value_print_options opts;
12759
12760 get_user_print_options (&opts);
12761 if (opts.addressprint)
12762 {
12763 annotate_field (4);
12764 uiout->field_core_addr ("addr", b->loc->gdbarch, b->loc->address);
12765 }
12766
12767 annotate_field (5);
12768 *last_loc = b->loc;
12769 switch (ex)
12770 {
12771 case ada_catch_exception:
12772 if (c->excep_string != NULL)
12773 {
12774 char *msg = xstrprintf (_("`%s' Ada exception"), c->excep_string);
12775
12776 uiout->field_string ("what", msg);
12777 xfree (msg);
12778 }
12779 else
12780 uiout->field_string ("what", "all Ada exceptions");
12781
12782 break;
12783
12784 case ada_catch_exception_unhandled:
12785 uiout->field_string ("what", "unhandled Ada exceptions");
12786 break;
12787
12788 case ada_catch_handlers:
12789 if (c->excep_string != NULL)
12790 {
12791 uiout->field_fmt ("what",
12792 _("`%s' Ada exception handlers"),
12793 c->excep_string);
12794 }
12795 else
12796 uiout->field_string ("what", "all Ada exceptions handlers");
12797 break;
12798
12799 case ada_catch_assert:
12800 uiout->field_string ("what", "failed Ada assertions");
12801 break;
12802
12803 default:
12804 internal_error (__FILE__, __LINE__, _("unexpected catchpoint type"));
12805 break;
12806 }
12807 }
12808
12809 /* Implement the PRINT_MENTION method in the breakpoint_ops structure
12810 for all exception catchpoint kinds. */
12811
12812 static void
12813 print_mention_exception (enum ada_exception_catchpoint_kind ex,
12814 struct breakpoint *b)
12815 {
12816 struct ada_catchpoint *c = (struct ada_catchpoint *) b;
12817 struct ui_out *uiout = current_uiout;
12818
12819 uiout->text (b->disposition == disp_del ? _("Temporary catchpoint ")
12820 : _("Catchpoint "));
12821 uiout->field_int ("bkptno", b->number);
12822 uiout->text (": ");
12823
12824 switch (ex)
12825 {
12826 case ada_catch_exception:
12827 if (c->excep_string != NULL)
12828 {
12829 char *info = xstrprintf (_("`%s' Ada exception"), c->excep_string);
12830 struct cleanup *old_chain = make_cleanup (xfree, info);
12831
12832 uiout->text (info);
12833 do_cleanups (old_chain);
12834 }
12835 else
12836 uiout->text (_("all Ada exceptions"));
12837 break;
12838
12839 case ada_catch_exception_unhandled:
12840 uiout->text (_("unhandled Ada exceptions"));
12841 break;
12842
12843 case ada_catch_handlers:
12844 if (c->excep_string != NULL)
12845 {
12846 std::string info
12847 = string_printf (_("`%s' Ada exception handlers"),
12848 c->excep_string);
12849 uiout->text (info.c_str ());
12850 }
12851 else
12852 uiout->text (_("all Ada exceptions handlers"));
12853 break;
12854
12855 case ada_catch_assert:
12856 uiout->text (_("failed Ada assertions"));
12857 break;
12858
12859 default:
12860 internal_error (__FILE__, __LINE__, _("unexpected catchpoint type"));
12861 break;
12862 }
12863 }
12864
12865 /* Implement the PRINT_RECREATE method in the breakpoint_ops structure
12866 for all exception catchpoint kinds. */
12867
12868 static void
12869 print_recreate_exception (enum ada_exception_catchpoint_kind ex,
12870 struct breakpoint *b, struct ui_file *fp)
12871 {
12872 struct ada_catchpoint *c = (struct ada_catchpoint *) b;
12873
12874 switch (ex)
12875 {
12876 case ada_catch_exception:
12877 fprintf_filtered (fp, "catch exception");
12878 if (c->excep_string != NULL)
12879 fprintf_filtered (fp, " %s", c->excep_string);
12880 break;
12881
12882 case ada_catch_exception_unhandled:
12883 fprintf_filtered (fp, "catch exception unhandled");
12884 break;
12885
12886 case ada_catch_handlers:
12887 fprintf_filtered (fp, "catch handlers");
12888 break;
12889
12890 case ada_catch_assert:
12891 fprintf_filtered (fp, "catch assert");
12892 break;
12893
12894 default:
12895 internal_error (__FILE__, __LINE__, _("unexpected catchpoint type"));
12896 }
12897 print_recreate_thread (b, fp);
12898 }
12899
12900 /* Virtual table for "catch exception" breakpoints. */
12901
12902 static struct bp_location *
12903 allocate_location_catch_exception (struct breakpoint *self)
12904 {
12905 return allocate_location_exception (ada_catch_exception, self);
12906 }
12907
12908 static void
12909 re_set_catch_exception (struct breakpoint *b)
12910 {
12911 re_set_exception (ada_catch_exception, b);
12912 }
12913
12914 static void
12915 check_status_catch_exception (bpstat bs)
12916 {
12917 check_status_exception (ada_catch_exception, bs);
12918 }
12919
12920 static enum print_stop_action
12921 print_it_catch_exception (bpstat bs)
12922 {
12923 return print_it_exception (ada_catch_exception, bs);
12924 }
12925
12926 static void
12927 print_one_catch_exception (struct breakpoint *b, struct bp_location **last_loc)
12928 {
12929 print_one_exception (ada_catch_exception, b, last_loc);
12930 }
12931
12932 static void
12933 print_mention_catch_exception (struct breakpoint *b)
12934 {
12935 print_mention_exception (ada_catch_exception, b);
12936 }
12937
12938 static void
12939 print_recreate_catch_exception (struct breakpoint *b, struct ui_file *fp)
12940 {
12941 print_recreate_exception (ada_catch_exception, b, fp);
12942 }
12943
12944 static struct breakpoint_ops catch_exception_breakpoint_ops;
12945
12946 /* Virtual table for "catch exception unhandled" breakpoints. */
12947
12948 static struct bp_location *
12949 allocate_location_catch_exception_unhandled (struct breakpoint *self)
12950 {
12951 return allocate_location_exception (ada_catch_exception_unhandled, self);
12952 }
12953
12954 static void
12955 re_set_catch_exception_unhandled (struct breakpoint *b)
12956 {
12957 re_set_exception (ada_catch_exception_unhandled, b);
12958 }
12959
12960 static void
12961 check_status_catch_exception_unhandled (bpstat bs)
12962 {
12963 check_status_exception (ada_catch_exception_unhandled, bs);
12964 }
12965
12966 static enum print_stop_action
12967 print_it_catch_exception_unhandled (bpstat bs)
12968 {
12969 return print_it_exception (ada_catch_exception_unhandled, bs);
12970 }
12971
12972 static void
12973 print_one_catch_exception_unhandled (struct breakpoint *b,
12974 struct bp_location **last_loc)
12975 {
12976 print_one_exception (ada_catch_exception_unhandled, b, last_loc);
12977 }
12978
12979 static void
12980 print_mention_catch_exception_unhandled (struct breakpoint *b)
12981 {
12982 print_mention_exception (ada_catch_exception_unhandled, b);
12983 }
12984
12985 static void
12986 print_recreate_catch_exception_unhandled (struct breakpoint *b,
12987 struct ui_file *fp)
12988 {
12989 print_recreate_exception (ada_catch_exception_unhandled, b, fp);
12990 }
12991
12992 static struct breakpoint_ops catch_exception_unhandled_breakpoint_ops;
12993
12994 /* Virtual table for "catch assert" breakpoints. */
12995
12996 static struct bp_location *
12997 allocate_location_catch_assert (struct breakpoint *self)
12998 {
12999 return allocate_location_exception (ada_catch_assert, self);
13000 }
13001
13002 static void
13003 re_set_catch_assert (struct breakpoint *b)
13004 {
13005 re_set_exception (ada_catch_assert, b);
13006 }
13007
13008 static void
13009 check_status_catch_assert (bpstat bs)
13010 {
13011 check_status_exception (ada_catch_assert, bs);
13012 }
13013
13014 static enum print_stop_action
13015 print_it_catch_assert (bpstat bs)
13016 {
13017 return print_it_exception (ada_catch_assert, bs);
13018 }
13019
13020 static void
13021 print_one_catch_assert (struct breakpoint *b, struct bp_location **last_loc)
13022 {
13023 print_one_exception (ada_catch_assert, b, last_loc);
13024 }
13025
13026 static void
13027 print_mention_catch_assert (struct breakpoint *b)
13028 {
13029 print_mention_exception (ada_catch_assert, b);
13030 }
13031
13032 static void
13033 print_recreate_catch_assert (struct breakpoint *b, struct ui_file *fp)
13034 {
13035 print_recreate_exception (ada_catch_assert, b, fp);
13036 }
13037
13038 static struct breakpoint_ops catch_assert_breakpoint_ops;
13039
13040 /* Virtual table for "catch handlers" breakpoints. */
13041
13042 static struct bp_location *
13043 allocate_location_catch_handlers (struct breakpoint *self)
13044 {
13045 return allocate_location_exception (ada_catch_handlers, self);
13046 }
13047
13048 static void
13049 re_set_catch_handlers (struct breakpoint *b)
13050 {
13051 re_set_exception (ada_catch_handlers, b);
13052 }
13053
13054 static void
13055 check_status_catch_handlers (bpstat bs)
13056 {
13057 check_status_exception (ada_catch_handlers, bs);
13058 }
13059
13060 static enum print_stop_action
13061 print_it_catch_handlers (bpstat bs)
13062 {
13063 return print_it_exception (ada_catch_handlers, bs);
13064 }
13065
13066 static void
13067 print_one_catch_handlers (struct breakpoint *b,
13068 struct bp_location **last_loc)
13069 {
13070 print_one_exception (ada_catch_handlers, b, last_loc);
13071 }
13072
13073 static void
13074 print_mention_catch_handlers (struct breakpoint *b)
13075 {
13076 print_mention_exception (ada_catch_handlers, b);
13077 }
13078
13079 static void
13080 print_recreate_catch_handlers (struct breakpoint *b,
13081 struct ui_file *fp)
13082 {
13083 print_recreate_exception (ada_catch_handlers, b, fp);
13084 }
13085
13086 static struct breakpoint_ops catch_handlers_breakpoint_ops;
13087
13088 /* Return a newly allocated copy of the first space-separated token
13089 in ARGSP, and then adjust ARGSP to point immediately after that
13090 token.
13091
13092 Return NULL if ARGPS does not contain any more tokens. */
13093
13094 static char *
13095 ada_get_next_arg (const char **argsp)
13096 {
13097 const char *args = *argsp;
13098 const char *end;
13099 char *result;
13100
13101 args = skip_spaces (args);
13102 if (args[0] == '\0')
13103 return NULL; /* No more arguments. */
13104
13105 /* Find the end of the current argument. */
13106
13107 end = skip_to_space (args);
13108
13109 /* Adjust ARGSP to point to the start of the next argument. */
13110
13111 *argsp = end;
13112
13113 /* Make a copy of the current argument and return it. */
13114
13115 result = (char *) xmalloc (end - args + 1);
13116 strncpy (result, args, end - args);
13117 result[end - args] = '\0';
13118
13119 return result;
13120 }
13121
13122 /* Split the arguments specified in a "catch exception" command.
13123 Set EX to the appropriate catchpoint type.
13124 Set EXCEP_STRING to the name of the specific exception if
13125 specified by the user.
13126 IS_CATCH_HANDLERS_CMD: True if the arguments are for a
13127 "catch handlers" command. False otherwise.
13128 If a condition is found at the end of the arguments, the condition
13129 expression is stored in COND_STRING (memory must be deallocated
13130 after use). Otherwise COND_STRING is set to NULL. */
13131
13132 static void
13133 catch_ada_exception_command_split (const char *args,
13134 bool is_catch_handlers_cmd,
13135 enum ada_exception_catchpoint_kind *ex,
13136 char **excep_string,
13137 char **cond_string)
13138 {
13139 struct cleanup *old_chain = make_cleanup (null_cleanup, NULL);
13140 char *exception_name;
13141 char *cond = NULL;
13142
13143 exception_name = ada_get_next_arg (&args);
13144 if (exception_name != NULL && strcmp (exception_name, "if") == 0)
13145 {
13146 /* This is not an exception name; this is the start of a condition
13147 expression for a catchpoint on all exceptions. So, "un-get"
13148 this token, and set exception_name to NULL. */
13149 xfree (exception_name);
13150 exception_name = NULL;
13151 args -= 2;
13152 }
13153 make_cleanup (xfree, exception_name);
13154
13155 /* Check to see if we have a condition. */
13156
13157 args = skip_spaces (args);
13158 if (startswith (args, "if")
13159 && (isspace (args[2]) || args[2] == '\0'))
13160 {
13161 args += 2;
13162 args = skip_spaces (args);
13163
13164 if (args[0] == '\0')
13165 error (_("Condition missing after `if' keyword"));
13166 cond = xstrdup (args);
13167 make_cleanup (xfree, cond);
13168
13169 args += strlen (args);
13170 }
13171
13172 /* Check that we do not have any more arguments. Anything else
13173 is unexpected. */
13174
13175 if (args[0] != '\0')
13176 error (_("Junk at end of expression"));
13177
13178 discard_cleanups (old_chain);
13179
13180 if (is_catch_handlers_cmd)
13181 {
13182 /* Catch handling of exceptions. */
13183 *ex = ada_catch_handlers;
13184 *excep_string = exception_name;
13185 }
13186 else if (exception_name == NULL)
13187 {
13188 /* Catch all exceptions. */
13189 *ex = ada_catch_exception;
13190 *excep_string = NULL;
13191 }
13192 else if (strcmp (exception_name, "unhandled") == 0)
13193 {
13194 /* Catch unhandled exceptions. */
13195 *ex = ada_catch_exception_unhandled;
13196 *excep_string = NULL;
13197 }
13198 else
13199 {
13200 /* Catch a specific exception. */
13201 *ex = ada_catch_exception;
13202 *excep_string = exception_name;
13203 }
13204 *cond_string = cond;
13205 }
13206
13207 /* Return the name of the symbol on which we should break in order to
13208 implement a catchpoint of the EX kind. */
13209
13210 static const char *
13211 ada_exception_sym_name (enum ada_exception_catchpoint_kind ex)
13212 {
13213 struct ada_inferior_data *data = get_ada_inferior_data (current_inferior ());
13214
13215 gdb_assert (data->exception_info != NULL);
13216
13217 switch (ex)
13218 {
13219 case ada_catch_exception:
13220 return (data->exception_info->catch_exception_sym);
13221 break;
13222 case ada_catch_exception_unhandled:
13223 return (data->exception_info->catch_exception_unhandled_sym);
13224 break;
13225 case ada_catch_assert:
13226 return (data->exception_info->catch_assert_sym);
13227 break;
13228 case ada_catch_handlers:
13229 return (data->exception_info->catch_handlers_sym);
13230 break;
13231 default:
13232 internal_error (__FILE__, __LINE__,
13233 _("unexpected catchpoint kind (%d)"), ex);
13234 }
13235 }
13236
13237 /* Return the breakpoint ops "virtual table" used for catchpoints
13238 of the EX kind. */
13239
13240 static const struct breakpoint_ops *
13241 ada_exception_breakpoint_ops (enum ada_exception_catchpoint_kind ex)
13242 {
13243 switch (ex)
13244 {
13245 case ada_catch_exception:
13246 return (&catch_exception_breakpoint_ops);
13247 break;
13248 case ada_catch_exception_unhandled:
13249 return (&catch_exception_unhandled_breakpoint_ops);
13250 break;
13251 case ada_catch_assert:
13252 return (&catch_assert_breakpoint_ops);
13253 break;
13254 case ada_catch_handlers:
13255 return (&catch_handlers_breakpoint_ops);
13256 break;
13257 default:
13258 internal_error (__FILE__, __LINE__,
13259 _("unexpected catchpoint kind (%d)"), ex);
13260 }
13261 }
13262
13263 /* Return the condition that will be used to match the current exception
13264 being raised with the exception that the user wants to catch. This
13265 assumes that this condition is used when the inferior just triggered
13266 an exception catchpoint.
13267 EX: the type of catchpoints used for catching Ada exceptions.
13268
13269 The string returned is a newly allocated string that needs to be
13270 deallocated later. */
13271
13272 static char *
13273 ada_exception_catchpoint_cond_string (const char *excep_string,
13274 enum ada_exception_catchpoint_kind ex)
13275 {
13276 int i;
13277 bool is_standard_exc = false;
13278 const char *actual_exc_expr;
13279 char *ref_exc_expr;
13280
13281 if (ex == ada_catch_handlers)
13282 {
13283 /* For exception handlers catchpoints, the condition string does
13284 not use the same parameter as for the other exceptions. */
13285 actual_exc_expr = ("long_integer (GNAT_GCC_exception_Access"
13286 "(gcc_exception).all.occurrence.id)");
13287 }
13288 else
13289 actual_exc_expr = "long_integer (e)";
13290
13291 /* The standard exceptions are a special case. They are defined in
13292 runtime units that have been compiled without debugging info; if
13293 EXCEP_STRING is the not-fully-qualified name of a standard
13294 exception (e.g. "constraint_error") then, during the evaluation
13295 of the condition expression, the symbol lookup on this name would
13296 *not* return this standard exception. The catchpoint condition
13297 may then be set only on user-defined exceptions which have the
13298 same not-fully-qualified name (e.g. my_package.constraint_error).
13299
13300 To avoid this unexcepted behavior, these standard exceptions are
13301 systematically prefixed by "standard". This means that "catch
13302 exception constraint_error" is rewritten into "catch exception
13303 standard.constraint_error".
13304
13305 If an exception named contraint_error is defined in another package of
13306 the inferior program, then the only way to specify this exception as a
13307 breakpoint condition is to use its fully-qualified named:
13308 e.g. my_package.constraint_error. */
13309
13310 for (i = 0; i < sizeof (standard_exc) / sizeof (char *); i++)
13311 {
13312 if (strcmp (standard_exc [i], excep_string) == 0)
13313 {
13314 is_standard_exc = true;
13315 break;
13316 }
13317 }
13318
13319 if (is_standard_exc)
13320 ref_exc_expr = xstrprintf ("long_integer (&standard.%s)", excep_string);
13321 else
13322 ref_exc_expr = xstrprintf ("long_integer (&%s)", excep_string);
13323
13324 char *result = xstrprintf ("%s = %s", actual_exc_expr, ref_exc_expr);
13325 xfree (ref_exc_expr);
13326 return result;
13327 }
13328
13329 /* Return the symtab_and_line that should be used to insert an exception
13330 catchpoint of the TYPE kind.
13331
13332 EXCEP_STRING should contain the name of a specific exception that
13333 the catchpoint should catch, or NULL otherwise.
13334
13335 ADDR_STRING returns the name of the function where the real
13336 breakpoint that implements the catchpoints is set, depending on the
13337 type of catchpoint we need to create. */
13338
13339 static struct symtab_and_line
13340 ada_exception_sal (enum ada_exception_catchpoint_kind ex, char *excep_string,
13341 const char **addr_string, const struct breakpoint_ops **ops)
13342 {
13343 const char *sym_name;
13344 struct symbol *sym;
13345
13346 /* First, find out which exception support info to use. */
13347 ada_exception_support_info_sniffer ();
13348
13349 /* Then lookup the function on which we will break in order to catch
13350 the Ada exceptions requested by the user. */
13351 sym_name = ada_exception_sym_name (ex);
13352 sym = standard_lookup (sym_name, NULL, VAR_DOMAIN);
13353
13354 /* We can assume that SYM is not NULL at this stage. If the symbol
13355 did not exist, ada_exception_support_info_sniffer would have
13356 raised an exception.
13357
13358 Also, ada_exception_support_info_sniffer should have already
13359 verified that SYM is a function symbol. */
13360 gdb_assert (sym != NULL);
13361 gdb_assert (SYMBOL_CLASS (sym) == LOC_BLOCK);
13362
13363 /* Set ADDR_STRING. */
13364 *addr_string = xstrdup (sym_name);
13365
13366 /* Set OPS. */
13367 *ops = ada_exception_breakpoint_ops (ex);
13368
13369 return find_function_start_sal (sym, 1);
13370 }
13371
13372 /* Create an Ada exception catchpoint.
13373
13374 EX_KIND is the kind of exception catchpoint to be created.
13375
13376 If EXCEPT_STRING is NULL, this catchpoint is expected to trigger
13377 for all exceptions. Otherwise, EXCEPT_STRING indicates the name
13378 of the exception to which this catchpoint applies. When not NULL,
13379 the string must be allocated on the heap, and its deallocation
13380 is no longer the responsibility of the caller.
13381
13382 COND_STRING, if not NULL, is the catchpoint condition. This string
13383 must be allocated on the heap, and its deallocation is no longer
13384 the responsibility of the caller.
13385
13386 TEMPFLAG, if nonzero, means that the underlying breakpoint
13387 should be temporary.
13388
13389 FROM_TTY is the usual argument passed to all commands implementations. */
13390
13391 void
13392 create_ada_exception_catchpoint (struct gdbarch *gdbarch,
13393 enum ada_exception_catchpoint_kind ex_kind,
13394 char *excep_string,
13395 char *cond_string,
13396 int tempflag,
13397 int disabled,
13398 int from_tty)
13399 {
13400 const char *addr_string = NULL;
13401 const struct breakpoint_ops *ops = NULL;
13402 struct symtab_and_line sal
13403 = ada_exception_sal (ex_kind, excep_string, &addr_string, &ops);
13404
13405 std::unique_ptr<ada_catchpoint> c (new ada_catchpoint ());
13406 init_ada_exception_breakpoint (c.get (), gdbarch, sal, addr_string,
13407 ops, tempflag, disabled, from_tty);
13408 c->excep_string = excep_string;
13409 create_excep_cond_exprs (c.get (), ex_kind);
13410 if (cond_string != NULL)
13411 set_breakpoint_condition (c.get (), cond_string, from_tty);
13412 install_breakpoint (0, std::move (c), 1);
13413 }
13414
13415 /* Implement the "catch exception" command. */
13416
13417 static void
13418 catch_ada_exception_command (const char *arg_entry, int from_tty,
13419 struct cmd_list_element *command)
13420 {
13421 const char *arg = arg_entry;
13422 struct gdbarch *gdbarch = get_current_arch ();
13423 int tempflag;
13424 enum ada_exception_catchpoint_kind ex_kind;
13425 char *excep_string = NULL;
13426 char *cond_string = NULL;
13427
13428 tempflag = get_cmd_context (command) == CATCH_TEMPORARY;
13429
13430 if (!arg)
13431 arg = "";
13432 catch_ada_exception_command_split (arg, false, &ex_kind, &excep_string,
13433 &cond_string);
13434 create_ada_exception_catchpoint (gdbarch, ex_kind,
13435 excep_string, cond_string,
13436 tempflag, 1 /* enabled */,
13437 from_tty);
13438 }
13439
13440 /* Implement the "catch handlers" command. */
13441
13442 static void
13443 catch_ada_handlers_command (const char *arg_entry, int from_tty,
13444 struct cmd_list_element *command)
13445 {
13446 const char *arg = arg_entry;
13447 struct gdbarch *gdbarch = get_current_arch ();
13448 int tempflag;
13449 enum ada_exception_catchpoint_kind ex_kind;
13450 char *excep_string = NULL;
13451 char *cond_string = NULL;
13452
13453 tempflag = get_cmd_context (command) == CATCH_TEMPORARY;
13454
13455 if (!arg)
13456 arg = "";
13457 catch_ada_exception_command_split (arg, true, &ex_kind, &excep_string,
13458 &cond_string);
13459 create_ada_exception_catchpoint (gdbarch, ex_kind,
13460 excep_string, cond_string,
13461 tempflag, 1 /* enabled */,
13462 from_tty);
13463 }
13464
13465 /* Split the arguments specified in a "catch assert" command.
13466
13467 ARGS contains the command's arguments (or the empty string if
13468 no arguments were passed).
13469
13470 If ARGS contains a condition, set COND_STRING to that condition
13471 (the memory needs to be deallocated after use). */
13472
13473 static void
13474 catch_ada_assert_command_split (const char *args, char **cond_string)
13475 {
13476 args = skip_spaces (args);
13477
13478 /* Check whether a condition was provided. */
13479 if (startswith (args, "if")
13480 && (isspace (args[2]) || args[2] == '\0'))
13481 {
13482 args += 2;
13483 args = skip_spaces (args);
13484 if (args[0] == '\0')
13485 error (_("condition missing after `if' keyword"));
13486 *cond_string = xstrdup (args);
13487 }
13488
13489 /* Otherwise, there should be no other argument at the end of
13490 the command. */
13491 else if (args[0] != '\0')
13492 error (_("Junk at end of arguments."));
13493 }
13494
13495 /* Implement the "catch assert" command. */
13496
13497 static void
13498 catch_assert_command (const char *arg_entry, int from_tty,
13499 struct cmd_list_element *command)
13500 {
13501 const char *arg = arg_entry;
13502 struct gdbarch *gdbarch = get_current_arch ();
13503 int tempflag;
13504 char *cond_string = NULL;
13505
13506 tempflag = get_cmd_context (command) == CATCH_TEMPORARY;
13507
13508 if (!arg)
13509 arg = "";
13510 catch_ada_assert_command_split (arg, &cond_string);
13511 create_ada_exception_catchpoint (gdbarch, ada_catch_assert,
13512 NULL, cond_string,
13513 tempflag, 1 /* enabled */,
13514 from_tty);
13515 }
13516
13517 /* Return non-zero if the symbol SYM is an Ada exception object. */
13518
13519 static int
13520 ada_is_exception_sym (struct symbol *sym)
13521 {
13522 const char *type_name = type_name_no_tag (SYMBOL_TYPE (sym));
13523
13524 return (SYMBOL_CLASS (sym) != LOC_TYPEDEF
13525 && SYMBOL_CLASS (sym) != LOC_BLOCK
13526 && SYMBOL_CLASS (sym) != LOC_CONST
13527 && SYMBOL_CLASS (sym) != LOC_UNRESOLVED
13528 && type_name != NULL && strcmp (type_name, "exception") == 0);
13529 }
13530
13531 /* Given a global symbol SYM, return non-zero iff SYM is a non-standard
13532 Ada exception object. This matches all exceptions except the ones
13533 defined by the Ada language. */
13534
13535 static int
13536 ada_is_non_standard_exception_sym (struct symbol *sym)
13537 {
13538 int i;
13539
13540 if (!ada_is_exception_sym (sym))
13541 return 0;
13542
13543 for (i = 0; i < ARRAY_SIZE (standard_exc); i++)
13544 if (strcmp (SYMBOL_LINKAGE_NAME (sym), standard_exc[i]) == 0)
13545 return 0; /* A standard exception. */
13546
13547 /* Numeric_Error is also a standard exception, so exclude it.
13548 See the STANDARD_EXC description for more details as to why
13549 this exception is not listed in that array. */
13550 if (strcmp (SYMBOL_LINKAGE_NAME (sym), "numeric_error") == 0)
13551 return 0;
13552
13553 return 1;
13554 }
13555
13556 /* A helper function for std::sort, comparing two struct ada_exc_info
13557 objects.
13558
13559 The comparison is determined first by exception name, and then
13560 by exception address. */
13561
13562 bool
13563 ada_exc_info::operator< (const ada_exc_info &other) const
13564 {
13565 int result;
13566
13567 result = strcmp (name, other.name);
13568 if (result < 0)
13569 return true;
13570 if (result == 0 && addr < other.addr)
13571 return true;
13572 return false;
13573 }
13574
13575 bool
13576 ada_exc_info::operator== (const ada_exc_info &other) const
13577 {
13578 return addr == other.addr && strcmp (name, other.name) == 0;
13579 }
13580
13581 /* Sort EXCEPTIONS using compare_ada_exception_info as the comparison
13582 routine, but keeping the first SKIP elements untouched.
13583
13584 All duplicates are also removed. */
13585
13586 static void
13587 sort_remove_dups_ada_exceptions_list (std::vector<ada_exc_info> *exceptions,
13588 int skip)
13589 {
13590 std::sort (exceptions->begin () + skip, exceptions->end ());
13591 exceptions->erase (std::unique (exceptions->begin () + skip, exceptions->end ()),
13592 exceptions->end ());
13593 }
13594
13595 /* Add all exceptions defined by the Ada standard whose name match
13596 a regular expression.
13597
13598 If PREG is not NULL, then this regexp_t object is used to
13599 perform the symbol name matching. Otherwise, no name-based
13600 filtering is performed.
13601
13602 EXCEPTIONS is a vector of exceptions to which matching exceptions
13603 gets pushed. */
13604
13605 static void
13606 ada_add_standard_exceptions (compiled_regex *preg,
13607 std::vector<ada_exc_info> *exceptions)
13608 {
13609 int i;
13610
13611 for (i = 0; i < ARRAY_SIZE (standard_exc); i++)
13612 {
13613 if (preg == NULL
13614 || preg->exec (standard_exc[i], 0, NULL, 0) == 0)
13615 {
13616 struct bound_minimal_symbol msymbol
13617 = ada_lookup_simple_minsym (standard_exc[i]);
13618
13619 if (msymbol.minsym != NULL)
13620 {
13621 struct ada_exc_info info
13622 = {standard_exc[i], BMSYMBOL_VALUE_ADDRESS (msymbol)};
13623
13624 exceptions->push_back (info);
13625 }
13626 }
13627 }
13628 }
13629
13630 /* Add all Ada exceptions defined locally and accessible from the given
13631 FRAME.
13632
13633 If PREG is not NULL, then this regexp_t object is used to
13634 perform the symbol name matching. Otherwise, no name-based
13635 filtering is performed.
13636
13637 EXCEPTIONS is a vector of exceptions to which matching exceptions
13638 gets pushed. */
13639
13640 static void
13641 ada_add_exceptions_from_frame (compiled_regex *preg,
13642 struct frame_info *frame,
13643 std::vector<ada_exc_info> *exceptions)
13644 {
13645 const struct block *block = get_frame_block (frame, 0);
13646
13647 while (block != 0)
13648 {
13649 struct block_iterator iter;
13650 struct symbol *sym;
13651
13652 ALL_BLOCK_SYMBOLS (block, iter, sym)
13653 {
13654 switch (SYMBOL_CLASS (sym))
13655 {
13656 case LOC_TYPEDEF:
13657 case LOC_BLOCK:
13658 case LOC_CONST:
13659 break;
13660 default:
13661 if (ada_is_exception_sym (sym))
13662 {
13663 struct ada_exc_info info = {SYMBOL_PRINT_NAME (sym),
13664 SYMBOL_VALUE_ADDRESS (sym)};
13665
13666 exceptions->push_back (info);
13667 }
13668 }
13669 }
13670 if (BLOCK_FUNCTION (block) != NULL)
13671 break;
13672 block = BLOCK_SUPERBLOCK (block);
13673 }
13674 }
13675
13676 /* Return true if NAME matches PREG or if PREG is NULL. */
13677
13678 static bool
13679 name_matches_regex (const char *name, compiled_regex *preg)
13680 {
13681 return (preg == NULL
13682 || preg->exec (ada_decode (name), 0, NULL, 0) == 0);
13683 }
13684
13685 /* Add all exceptions defined globally whose name name match
13686 a regular expression, excluding standard exceptions.
13687
13688 The reason we exclude standard exceptions is that they need
13689 to be handled separately: Standard exceptions are defined inside
13690 a runtime unit which is normally not compiled with debugging info,
13691 and thus usually do not show up in our symbol search. However,
13692 if the unit was in fact built with debugging info, we need to
13693 exclude them because they would duplicate the entry we found
13694 during the special loop that specifically searches for those
13695 standard exceptions.
13696
13697 If PREG is not NULL, then this regexp_t object is used to
13698 perform the symbol name matching. Otherwise, no name-based
13699 filtering is performed.
13700
13701 EXCEPTIONS is a vector of exceptions to which matching exceptions
13702 gets pushed. */
13703
13704 static void
13705 ada_add_global_exceptions (compiled_regex *preg,
13706 std::vector<ada_exc_info> *exceptions)
13707 {
13708 struct objfile *objfile;
13709 struct compunit_symtab *s;
13710
13711 /* In Ada, the symbol "search name" is a linkage name, whereas the
13712 regular expression used to do the matching refers to the natural
13713 name. So match against the decoded name. */
13714 expand_symtabs_matching (NULL,
13715 lookup_name_info::match_any (),
13716 [&] (const char *search_name)
13717 {
13718 const char *decoded = ada_decode (search_name);
13719 return name_matches_regex (decoded, preg);
13720 },
13721 NULL,
13722 VARIABLES_DOMAIN);
13723
13724 ALL_COMPUNITS (objfile, s)
13725 {
13726 const struct blockvector *bv = COMPUNIT_BLOCKVECTOR (s);
13727 int i;
13728
13729 for (i = GLOBAL_BLOCK; i <= STATIC_BLOCK; i++)
13730 {
13731 struct block *b = BLOCKVECTOR_BLOCK (bv, i);
13732 struct block_iterator iter;
13733 struct symbol *sym;
13734
13735 ALL_BLOCK_SYMBOLS (b, iter, sym)
13736 if (ada_is_non_standard_exception_sym (sym)
13737 && name_matches_regex (SYMBOL_NATURAL_NAME (sym), preg))
13738 {
13739 struct ada_exc_info info
13740 = {SYMBOL_PRINT_NAME (sym), SYMBOL_VALUE_ADDRESS (sym)};
13741
13742 exceptions->push_back (info);
13743 }
13744 }
13745 }
13746 }
13747
13748 /* Implements ada_exceptions_list with the regular expression passed
13749 as a regex_t, rather than a string.
13750
13751 If not NULL, PREG is used to filter out exceptions whose names
13752 do not match. Otherwise, all exceptions are listed. */
13753
13754 static std::vector<ada_exc_info>
13755 ada_exceptions_list_1 (compiled_regex *preg)
13756 {
13757 std::vector<ada_exc_info> result;
13758 int prev_len;
13759
13760 /* First, list the known standard exceptions. These exceptions
13761 need to be handled separately, as they are usually defined in
13762 runtime units that have been compiled without debugging info. */
13763
13764 ada_add_standard_exceptions (preg, &result);
13765
13766 /* Next, find all exceptions whose scope is local and accessible
13767 from the currently selected frame. */
13768
13769 if (has_stack_frames ())
13770 {
13771 prev_len = result.size ();
13772 ada_add_exceptions_from_frame (preg, get_selected_frame (NULL),
13773 &result);
13774 if (result.size () > prev_len)
13775 sort_remove_dups_ada_exceptions_list (&result, prev_len);
13776 }
13777
13778 /* Add all exceptions whose scope is global. */
13779
13780 prev_len = result.size ();
13781 ada_add_global_exceptions (preg, &result);
13782 if (result.size () > prev_len)
13783 sort_remove_dups_ada_exceptions_list (&result, prev_len);
13784
13785 return result;
13786 }
13787
13788 /* Return a vector of ada_exc_info.
13789
13790 If REGEXP is NULL, all exceptions are included in the result.
13791 Otherwise, it should contain a valid regular expression,
13792 and only the exceptions whose names match that regular expression
13793 are included in the result.
13794
13795 The exceptions are sorted in the following order:
13796 - Standard exceptions (defined by the Ada language), in
13797 alphabetical order;
13798 - Exceptions only visible from the current frame, in
13799 alphabetical order;
13800 - Exceptions whose scope is global, in alphabetical order. */
13801
13802 std::vector<ada_exc_info>
13803 ada_exceptions_list (const char *regexp)
13804 {
13805 if (regexp == NULL)
13806 return ada_exceptions_list_1 (NULL);
13807
13808 compiled_regex reg (regexp, REG_NOSUB, _("invalid regular expression"));
13809 return ada_exceptions_list_1 (&reg);
13810 }
13811
13812 /* Implement the "info exceptions" command. */
13813
13814 static void
13815 info_exceptions_command (const char *regexp, int from_tty)
13816 {
13817 struct gdbarch *gdbarch = get_current_arch ();
13818
13819 std::vector<ada_exc_info> exceptions = ada_exceptions_list (regexp);
13820
13821 if (regexp != NULL)
13822 printf_filtered
13823 (_("All Ada exceptions matching regular expression \"%s\":\n"), regexp);
13824 else
13825 printf_filtered (_("All defined Ada exceptions:\n"));
13826
13827 for (const ada_exc_info &info : exceptions)
13828 printf_filtered ("%s: %s\n", info.name, paddress (gdbarch, info.addr));
13829 }
13830
13831 /* Operators */
13832 /* Information about operators given special treatment in functions
13833 below. */
13834 /* Format: OP_DEFN (<operator>, <operator length>, <# args>, <binop>). */
13835
13836 #define ADA_OPERATORS \
13837 OP_DEFN (OP_VAR_VALUE, 4, 0, 0) \
13838 OP_DEFN (BINOP_IN_BOUNDS, 3, 2, 0) \
13839 OP_DEFN (TERNOP_IN_RANGE, 1, 3, 0) \
13840 OP_DEFN (OP_ATR_FIRST, 1, 2, 0) \
13841 OP_DEFN (OP_ATR_LAST, 1, 2, 0) \
13842 OP_DEFN (OP_ATR_LENGTH, 1, 2, 0) \
13843 OP_DEFN (OP_ATR_IMAGE, 1, 2, 0) \
13844 OP_DEFN (OP_ATR_MAX, 1, 3, 0) \
13845 OP_DEFN (OP_ATR_MIN, 1, 3, 0) \
13846 OP_DEFN (OP_ATR_MODULUS, 1, 1, 0) \
13847 OP_DEFN (OP_ATR_POS, 1, 2, 0) \
13848 OP_DEFN (OP_ATR_SIZE, 1, 1, 0) \
13849 OP_DEFN (OP_ATR_TAG, 1, 1, 0) \
13850 OP_DEFN (OP_ATR_VAL, 1, 2, 0) \
13851 OP_DEFN (UNOP_QUAL, 3, 1, 0) \
13852 OP_DEFN (UNOP_IN_RANGE, 3, 1, 0) \
13853 OP_DEFN (OP_OTHERS, 1, 1, 0) \
13854 OP_DEFN (OP_POSITIONAL, 3, 1, 0) \
13855 OP_DEFN (OP_DISCRETE_RANGE, 1, 2, 0)
13856
13857 static void
13858 ada_operator_length (const struct expression *exp, int pc, int *oplenp,
13859 int *argsp)
13860 {
13861 switch (exp->elts[pc - 1].opcode)
13862 {
13863 default:
13864 operator_length_standard (exp, pc, oplenp, argsp);
13865 break;
13866
13867 #define OP_DEFN(op, len, args, binop) \
13868 case op: *oplenp = len; *argsp = args; break;
13869 ADA_OPERATORS;
13870 #undef OP_DEFN
13871
13872 case OP_AGGREGATE:
13873 *oplenp = 3;
13874 *argsp = longest_to_int (exp->elts[pc - 2].longconst);
13875 break;
13876
13877 case OP_CHOICES:
13878 *oplenp = 3;
13879 *argsp = longest_to_int (exp->elts[pc - 2].longconst) + 1;
13880 break;
13881 }
13882 }
13883
13884 /* Implementation of the exp_descriptor method operator_check. */
13885
13886 static int
13887 ada_operator_check (struct expression *exp, int pos,
13888 int (*objfile_func) (struct objfile *objfile, void *data),
13889 void *data)
13890 {
13891 const union exp_element *const elts = exp->elts;
13892 struct type *type = NULL;
13893
13894 switch (elts[pos].opcode)
13895 {
13896 case UNOP_IN_RANGE:
13897 case UNOP_QUAL:
13898 type = elts[pos + 1].type;
13899 break;
13900
13901 default:
13902 return operator_check_standard (exp, pos, objfile_func, data);
13903 }
13904
13905 /* Invoke callbacks for TYPE and OBJFILE if they were set as non-NULL. */
13906
13907 if (type && TYPE_OBJFILE (type)
13908 && (*objfile_func) (TYPE_OBJFILE (type), data))
13909 return 1;
13910
13911 return 0;
13912 }
13913
13914 static const char *
13915 ada_op_name (enum exp_opcode opcode)
13916 {
13917 switch (opcode)
13918 {
13919 default:
13920 return op_name_standard (opcode);
13921
13922 #define OP_DEFN(op, len, args, binop) case op: return #op;
13923 ADA_OPERATORS;
13924 #undef OP_DEFN
13925
13926 case OP_AGGREGATE:
13927 return "OP_AGGREGATE";
13928 case OP_CHOICES:
13929 return "OP_CHOICES";
13930 case OP_NAME:
13931 return "OP_NAME";
13932 }
13933 }
13934
13935 /* As for operator_length, but assumes PC is pointing at the first
13936 element of the operator, and gives meaningful results only for the
13937 Ada-specific operators, returning 0 for *OPLENP and *ARGSP otherwise. */
13938
13939 static void
13940 ada_forward_operator_length (struct expression *exp, int pc,
13941 int *oplenp, int *argsp)
13942 {
13943 switch (exp->elts[pc].opcode)
13944 {
13945 default:
13946 *oplenp = *argsp = 0;
13947 break;
13948
13949 #define OP_DEFN(op, len, args, binop) \
13950 case op: *oplenp = len; *argsp = args; break;
13951 ADA_OPERATORS;
13952 #undef OP_DEFN
13953
13954 case OP_AGGREGATE:
13955 *oplenp = 3;
13956 *argsp = longest_to_int (exp->elts[pc + 1].longconst);
13957 break;
13958
13959 case OP_CHOICES:
13960 *oplenp = 3;
13961 *argsp = longest_to_int (exp->elts[pc + 1].longconst) + 1;
13962 break;
13963
13964 case OP_STRING:
13965 case OP_NAME:
13966 {
13967 int len = longest_to_int (exp->elts[pc + 1].longconst);
13968
13969 *oplenp = 4 + BYTES_TO_EXP_ELEM (len + 1);
13970 *argsp = 0;
13971 break;
13972 }
13973 }
13974 }
13975
13976 static int
13977 ada_dump_subexp_body (struct expression *exp, struct ui_file *stream, int elt)
13978 {
13979 enum exp_opcode op = exp->elts[elt].opcode;
13980 int oplen, nargs;
13981 int pc = elt;
13982 int i;
13983
13984 ada_forward_operator_length (exp, elt, &oplen, &nargs);
13985
13986 switch (op)
13987 {
13988 /* Ada attributes ('Foo). */
13989 case OP_ATR_FIRST:
13990 case OP_ATR_LAST:
13991 case OP_ATR_LENGTH:
13992 case OP_ATR_IMAGE:
13993 case OP_ATR_MAX:
13994 case OP_ATR_MIN:
13995 case OP_ATR_MODULUS:
13996 case OP_ATR_POS:
13997 case OP_ATR_SIZE:
13998 case OP_ATR_TAG:
13999 case OP_ATR_VAL:
14000 break;
14001
14002 case UNOP_IN_RANGE:
14003 case UNOP_QUAL:
14004 /* XXX: gdb_sprint_host_address, type_sprint */
14005 fprintf_filtered (stream, _("Type @"));
14006 gdb_print_host_address (exp->elts[pc + 1].type, stream);
14007 fprintf_filtered (stream, " (");
14008 type_print (exp->elts[pc + 1].type, NULL, stream, 0);
14009 fprintf_filtered (stream, ")");
14010 break;
14011 case BINOP_IN_BOUNDS:
14012 fprintf_filtered (stream, " (%d)",
14013 longest_to_int (exp->elts[pc + 2].longconst));
14014 break;
14015 case TERNOP_IN_RANGE:
14016 break;
14017
14018 case OP_AGGREGATE:
14019 case OP_OTHERS:
14020 case OP_DISCRETE_RANGE:
14021 case OP_POSITIONAL:
14022 case OP_CHOICES:
14023 break;
14024
14025 case OP_NAME:
14026 case OP_STRING:
14027 {
14028 char *name = &exp->elts[elt + 2].string;
14029 int len = longest_to_int (exp->elts[elt + 1].longconst);
14030
14031 fprintf_filtered (stream, "Text: `%.*s'", len, name);
14032 break;
14033 }
14034
14035 default:
14036 return dump_subexp_body_standard (exp, stream, elt);
14037 }
14038
14039 elt += oplen;
14040 for (i = 0; i < nargs; i += 1)
14041 elt = dump_subexp (exp, stream, elt);
14042
14043 return elt;
14044 }
14045
14046 /* The Ada extension of print_subexp (q.v.). */
14047
14048 static void
14049 ada_print_subexp (struct expression *exp, int *pos,
14050 struct ui_file *stream, enum precedence prec)
14051 {
14052 int oplen, nargs, i;
14053 int pc = *pos;
14054 enum exp_opcode op = exp->elts[pc].opcode;
14055
14056 ada_forward_operator_length (exp, pc, &oplen, &nargs);
14057
14058 *pos += oplen;
14059 switch (op)
14060 {
14061 default:
14062 *pos -= oplen;
14063 print_subexp_standard (exp, pos, stream, prec);
14064 return;
14065
14066 case OP_VAR_VALUE:
14067 fputs_filtered (SYMBOL_NATURAL_NAME (exp->elts[pc + 2].symbol), stream);
14068 return;
14069
14070 case BINOP_IN_BOUNDS:
14071 /* XXX: sprint_subexp */
14072 print_subexp (exp, pos, stream, PREC_SUFFIX);
14073 fputs_filtered (" in ", stream);
14074 print_subexp (exp, pos, stream, PREC_SUFFIX);
14075 fputs_filtered ("'range", stream);
14076 if (exp->elts[pc + 1].longconst > 1)
14077 fprintf_filtered (stream, "(%ld)",
14078 (long) exp->elts[pc + 1].longconst);
14079 return;
14080
14081 case TERNOP_IN_RANGE:
14082 if (prec >= PREC_EQUAL)
14083 fputs_filtered ("(", stream);
14084 /* XXX: sprint_subexp */
14085 print_subexp (exp, pos, stream, PREC_SUFFIX);
14086 fputs_filtered (" in ", stream);
14087 print_subexp (exp, pos, stream, PREC_EQUAL);
14088 fputs_filtered (" .. ", stream);
14089 print_subexp (exp, pos, stream, PREC_EQUAL);
14090 if (prec >= PREC_EQUAL)
14091 fputs_filtered (")", stream);
14092 return;
14093
14094 case OP_ATR_FIRST:
14095 case OP_ATR_LAST:
14096 case OP_ATR_LENGTH:
14097 case OP_ATR_IMAGE:
14098 case OP_ATR_MAX:
14099 case OP_ATR_MIN:
14100 case OP_ATR_MODULUS:
14101 case OP_ATR_POS:
14102 case OP_ATR_SIZE:
14103 case OP_ATR_TAG:
14104 case OP_ATR_VAL:
14105 if (exp->elts[*pos].opcode == OP_TYPE)
14106 {
14107 if (TYPE_CODE (exp->elts[*pos + 1].type) != TYPE_CODE_VOID)
14108 LA_PRINT_TYPE (exp->elts[*pos + 1].type, "", stream, 0, 0,
14109 &type_print_raw_options);
14110 *pos += 3;
14111 }
14112 else
14113 print_subexp (exp, pos, stream, PREC_SUFFIX);
14114 fprintf_filtered (stream, "'%s", ada_attribute_name (op));
14115 if (nargs > 1)
14116 {
14117 int tem;
14118
14119 for (tem = 1; tem < nargs; tem += 1)
14120 {
14121 fputs_filtered ((tem == 1) ? " (" : ", ", stream);
14122 print_subexp (exp, pos, stream, PREC_ABOVE_COMMA);
14123 }
14124 fputs_filtered (")", stream);
14125 }
14126 return;
14127
14128 case UNOP_QUAL:
14129 type_print (exp->elts[pc + 1].type, "", stream, 0);
14130 fputs_filtered ("'(", stream);
14131 print_subexp (exp, pos, stream, PREC_PREFIX);
14132 fputs_filtered (")", stream);
14133 return;
14134
14135 case UNOP_IN_RANGE:
14136 /* XXX: sprint_subexp */
14137 print_subexp (exp, pos, stream, PREC_SUFFIX);
14138 fputs_filtered (" in ", stream);
14139 LA_PRINT_TYPE (exp->elts[pc + 1].type, "", stream, 1, 0,
14140 &type_print_raw_options);
14141 return;
14142
14143 case OP_DISCRETE_RANGE:
14144 print_subexp (exp, pos, stream, PREC_SUFFIX);
14145 fputs_filtered ("..", stream);
14146 print_subexp (exp, pos, stream, PREC_SUFFIX);
14147 return;
14148
14149 case OP_OTHERS:
14150 fputs_filtered ("others => ", stream);
14151 print_subexp (exp, pos, stream, PREC_SUFFIX);
14152 return;
14153
14154 case OP_CHOICES:
14155 for (i = 0; i < nargs-1; i += 1)
14156 {
14157 if (i > 0)
14158 fputs_filtered ("|", stream);
14159 print_subexp (exp, pos, stream, PREC_SUFFIX);
14160 }
14161 fputs_filtered (" => ", stream);
14162 print_subexp (exp, pos, stream, PREC_SUFFIX);
14163 return;
14164
14165 case OP_POSITIONAL:
14166 print_subexp (exp, pos, stream, PREC_SUFFIX);
14167 return;
14168
14169 case OP_AGGREGATE:
14170 fputs_filtered ("(", stream);
14171 for (i = 0; i < nargs; i += 1)
14172 {
14173 if (i > 0)
14174 fputs_filtered (", ", stream);
14175 print_subexp (exp, pos, stream, PREC_SUFFIX);
14176 }
14177 fputs_filtered (")", stream);
14178 return;
14179 }
14180 }
14181
14182 /* Table mapping opcodes into strings for printing operators
14183 and precedences of the operators. */
14184
14185 static const struct op_print ada_op_print_tab[] = {
14186 {":=", BINOP_ASSIGN, PREC_ASSIGN, 1},
14187 {"or else", BINOP_LOGICAL_OR, PREC_LOGICAL_OR, 0},
14188 {"and then", BINOP_LOGICAL_AND, PREC_LOGICAL_AND, 0},
14189 {"or", BINOP_BITWISE_IOR, PREC_BITWISE_IOR, 0},
14190 {"xor", BINOP_BITWISE_XOR, PREC_BITWISE_XOR, 0},
14191 {"and", BINOP_BITWISE_AND, PREC_BITWISE_AND, 0},
14192 {"=", BINOP_EQUAL, PREC_EQUAL, 0},
14193 {"/=", BINOP_NOTEQUAL, PREC_EQUAL, 0},
14194 {"<=", BINOP_LEQ, PREC_ORDER, 0},
14195 {">=", BINOP_GEQ, PREC_ORDER, 0},
14196 {">", BINOP_GTR, PREC_ORDER, 0},
14197 {"<", BINOP_LESS, PREC_ORDER, 0},
14198 {">>", BINOP_RSH, PREC_SHIFT, 0},
14199 {"<<", BINOP_LSH, PREC_SHIFT, 0},
14200 {"+", BINOP_ADD, PREC_ADD, 0},
14201 {"-", BINOP_SUB, PREC_ADD, 0},
14202 {"&", BINOP_CONCAT, PREC_ADD, 0},
14203 {"*", BINOP_MUL, PREC_MUL, 0},
14204 {"/", BINOP_DIV, PREC_MUL, 0},
14205 {"rem", BINOP_REM, PREC_MUL, 0},
14206 {"mod", BINOP_MOD, PREC_MUL, 0},
14207 {"**", BINOP_EXP, PREC_REPEAT, 0},
14208 {"@", BINOP_REPEAT, PREC_REPEAT, 0},
14209 {"-", UNOP_NEG, PREC_PREFIX, 0},
14210 {"+", UNOP_PLUS, PREC_PREFIX, 0},
14211 {"not ", UNOP_LOGICAL_NOT, PREC_PREFIX, 0},
14212 {"not ", UNOP_COMPLEMENT, PREC_PREFIX, 0},
14213 {"abs ", UNOP_ABS, PREC_PREFIX, 0},
14214 {".all", UNOP_IND, PREC_SUFFIX, 1},
14215 {"'access", UNOP_ADDR, PREC_SUFFIX, 1},
14216 {"'size", OP_ATR_SIZE, PREC_SUFFIX, 1},
14217 {NULL, OP_NULL, PREC_SUFFIX, 0}
14218 };
14219 \f
14220 enum ada_primitive_types {
14221 ada_primitive_type_int,
14222 ada_primitive_type_long,
14223 ada_primitive_type_short,
14224 ada_primitive_type_char,
14225 ada_primitive_type_float,
14226 ada_primitive_type_double,
14227 ada_primitive_type_void,
14228 ada_primitive_type_long_long,
14229 ada_primitive_type_long_double,
14230 ada_primitive_type_natural,
14231 ada_primitive_type_positive,
14232 ada_primitive_type_system_address,
14233 ada_primitive_type_storage_offset,
14234 nr_ada_primitive_types
14235 };
14236
14237 static void
14238 ada_language_arch_info (struct gdbarch *gdbarch,
14239 struct language_arch_info *lai)
14240 {
14241 const struct builtin_type *builtin = builtin_type (gdbarch);
14242
14243 lai->primitive_type_vector
14244 = GDBARCH_OBSTACK_CALLOC (gdbarch, nr_ada_primitive_types + 1,
14245 struct type *);
14246
14247 lai->primitive_type_vector [ada_primitive_type_int]
14248 = arch_integer_type (gdbarch, gdbarch_int_bit (gdbarch),
14249 0, "integer");
14250 lai->primitive_type_vector [ada_primitive_type_long]
14251 = arch_integer_type (gdbarch, gdbarch_long_bit (gdbarch),
14252 0, "long_integer");
14253 lai->primitive_type_vector [ada_primitive_type_short]
14254 = arch_integer_type (gdbarch, gdbarch_short_bit (gdbarch),
14255 0, "short_integer");
14256 lai->string_char_type
14257 = lai->primitive_type_vector [ada_primitive_type_char]
14258 = arch_character_type (gdbarch, TARGET_CHAR_BIT, 0, "character");
14259 lai->primitive_type_vector [ada_primitive_type_float]
14260 = arch_float_type (gdbarch, gdbarch_float_bit (gdbarch),
14261 "float", gdbarch_float_format (gdbarch));
14262 lai->primitive_type_vector [ada_primitive_type_double]
14263 = arch_float_type (gdbarch, gdbarch_double_bit (gdbarch),
14264 "long_float", gdbarch_double_format (gdbarch));
14265 lai->primitive_type_vector [ada_primitive_type_long_long]
14266 = arch_integer_type (gdbarch, gdbarch_long_long_bit (gdbarch),
14267 0, "long_long_integer");
14268 lai->primitive_type_vector [ada_primitive_type_long_double]
14269 = arch_float_type (gdbarch, gdbarch_long_double_bit (gdbarch),
14270 "long_long_float", gdbarch_long_double_format (gdbarch));
14271 lai->primitive_type_vector [ada_primitive_type_natural]
14272 = arch_integer_type (gdbarch, gdbarch_int_bit (gdbarch),
14273 0, "natural");
14274 lai->primitive_type_vector [ada_primitive_type_positive]
14275 = arch_integer_type (gdbarch, gdbarch_int_bit (gdbarch),
14276 0, "positive");
14277 lai->primitive_type_vector [ada_primitive_type_void]
14278 = builtin->builtin_void;
14279
14280 lai->primitive_type_vector [ada_primitive_type_system_address]
14281 = lookup_pointer_type (arch_type (gdbarch, TYPE_CODE_VOID, TARGET_CHAR_BIT,
14282 "void"));
14283 TYPE_NAME (lai->primitive_type_vector [ada_primitive_type_system_address])
14284 = "system__address";
14285
14286 /* Create the equivalent of the System.Storage_Elements.Storage_Offset
14287 type. This is a signed integral type whose size is the same as
14288 the size of addresses. */
14289 {
14290 unsigned int addr_length = TYPE_LENGTH
14291 (lai->primitive_type_vector [ada_primitive_type_system_address]);
14292
14293 lai->primitive_type_vector [ada_primitive_type_storage_offset]
14294 = arch_integer_type (gdbarch, addr_length * HOST_CHAR_BIT, 0,
14295 "storage_offset");
14296 }
14297
14298 lai->bool_type_symbol = NULL;
14299 lai->bool_type_default = builtin->builtin_bool;
14300 }
14301 \f
14302 /* Language vector */
14303
14304 /* Not really used, but needed in the ada_language_defn. */
14305
14306 static void
14307 emit_char (int c, struct type *type, struct ui_file *stream, int quoter)
14308 {
14309 ada_emit_char (c, type, stream, quoter, 1);
14310 }
14311
14312 static int
14313 parse (struct parser_state *ps)
14314 {
14315 warnings_issued = 0;
14316 return ada_parse (ps);
14317 }
14318
14319 static const struct exp_descriptor ada_exp_descriptor = {
14320 ada_print_subexp,
14321 ada_operator_length,
14322 ada_operator_check,
14323 ada_op_name,
14324 ada_dump_subexp_body,
14325 ada_evaluate_subexp
14326 };
14327
14328 /* symbol_name_matcher_ftype adapter for wild_match. */
14329
14330 static bool
14331 do_wild_match (const char *symbol_search_name,
14332 const lookup_name_info &lookup_name,
14333 completion_match_result *comp_match_res)
14334 {
14335 return wild_match (symbol_search_name, ada_lookup_name (lookup_name));
14336 }
14337
14338 /* symbol_name_matcher_ftype adapter for full_match. */
14339
14340 static bool
14341 do_full_match (const char *symbol_search_name,
14342 const lookup_name_info &lookup_name,
14343 completion_match_result *comp_match_res)
14344 {
14345 return full_match (symbol_search_name, ada_lookup_name (lookup_name));
14346 }
14347
14348 /* Build the Ada lookup name for LOOKUP_NAME. */
14349
14350 ada_lookup_name_info::ada_lookup_name_info (const lookup_name_info &lookup_name)
14351 {
14352 const std::string &user_name = lookup_name.name ();
14353
14354 if (user_name[0] == '<')
14355 {
14356 if (user_name.back () == '>')
14357 m_encoded_name = user_name.substr (1, user_name.size () - 2);
14358 else
14359 m_encoded_name = user_name.substr (1, user_name.size () - 1);
14360 m_encoded_p = true;
14361 m_verbatim_p = true;
14362 m_wild_match_p = false;
14363 m_standard_p = false;
14364 }
14365 else
14366 {
14367 m_verbatim_p = false;
14368
14369 m_encoded_p = user_name.find ("__") != std::string::npos;
14370
14371 if (!m_encoded_p)
14372 {
14373 const char *folded = ada_fold_name (user_name.c_str ());
14374 const char *encoded = ada_encode_1 (folded, false);
14375 if (encoded != NULL)
14376 m_encoded_name = encoded;
14377 else
14378 m_encoded_name = user_name;
14379 }
14380 else
14381 m_encoded_name = user_name;
14382
14383 /* Handle the 'package Standard' special case. See description
14384 of m_standard_p. */
14385 if (startswith (m_encoded_name.c_str (), "standard__"))
14386 {
14387 m_encoded_name = m_encoded_name.substr (sizeof ("standard__") - 1);
14388 m_standard_p = true;
14389 }
14390 else
14391 m_standard_p = false;
14392
14393 /* If the name contains a ".", then the user is entering a fully
14394 qualified entity name, and the match must not be done in wild
14395 mode. Similarly, if the user wants to complete what looks
14396 like an encoded name, the match must not be done in wild
14397 mode. Also, in the standard__ special case always do
14398 non-wild matching. */
14399 m_wild_match_p
14400 = (lookup_name.match_type () != symbol_name_match_type::FULL
14401 && !m_encoded_p
14402 && !m_standard_p
14403 && user_name.find ('.') == std::string::npos);
14404 }
14405 }
14406
14407 /* symbol_name_matcher_ftype method for Ada. This only handles
14408 completion mode. */
14409
14410 static bool
14411 ada_symbol_name_matches (const char *symbol_search_name,
14412 const lookup_name_info &lookup_name,
14413 completion_match_result *comp_match_res)
14414 {
14415 return lookup_name.ada ().matches (symbol_search_name,
14416 lookup_name.match_type (),
14417 comp_match_res);
14418 }
14419
14420 /* Implement the "la_get_symbol_name_matcher" language_defn method for
14421 Ada. */
14422
14423 static symbol_name_matcher_ftype *
14424 ada_get_symbol_name_matcher (const lookup_name_info &lookup_name)
14425 {
14426 if (lookup_name.completion_mode ())
14427 return ada_symbol_name_matches;
14428 else
14429 {
14430 if (lookup_name.ada ().wild_match_p ())
14431 return do_wild_match;
14432 else
14433 return do_full_match;
14434 }
14435 }
14436
14437 /* Implement the "la_read_var_value" language_defn method for Ada. */
14438
14439 static struct value *
14440 ada_read_var_value (struct symbol *var, const struct block *var_block,
14441 struct frame_info *frame)
14442 {
14443 const struct block *frame_block = NULL;
14444 struct symbol *renaming_sym = NULL;
14445
14446 /* The only case where default_read_var_value is not sufficient
14447 is when VAR is a renaming... */
14448 if (frame)
14449 frame_block = get_frame_block (frame, NULL);
14450 if (frame_block)
14451 renaming_sym = ada_find_renaming_symbol (var, frame_block);
14452 if (renaming_sym != NULL)
14453 return ada_read_renaming_var_value (renaming_sym, frame_block);
14454
14455 /* This is a typical case where we expect the default_read_var_value
14456 function to work. */
14457 return default_read_var_value (var, var_block, frame);
14458 }
14459
14460 static const char *ada_extensions[] =
14461 {
14462 ".adb", ".ads", ".a", ".ada", ".dg", NULL
14463 };
14464
14465 extern const struct language_defn ada_language_defn = {
14466 "ada", /* Language name */
14467 "Ada",
14468 language_ada,
14469 range_check_off,
14470 case_sensitive_on, /* Yes, Ada is case-insensitive, but
14471 that's not quite what this means. */
14472 array_row_major,
14473 macro_expansion_no,
14474 ada_extensions,
14475 &ada_exp_descriptor,
14476 parse,
14477 ada_yyerror,
14478 resolve,
14479 ada_printchar, /* Print a character constant */
14480 ada_printstr, /* Function to print string constant */
14481 emit_char, /* Function to print single char (not used) */
14482 ada_print_type, /* Print a type using appropriate syntax */
14483 ada_print_typedef, /* Print a typedef using appropriate syntax */
14484 ada_val_print, /* Print a value using appropriate syntax */
14485 ada_value_print, /* Print a top-level value */
14486 ada_read_var_value, /* la_read_var_value */
14487 NULL, /* Language specific skip_trampoline */
14488 NULL, /* name_of_this */
14489 ada_lookup_symbol_nonlocal, /* Looking up non-local symbols. */
14490 basic_lookup_transparent_type, /* lookup_transparent_type */
14491 ada_la_decode, /* Language specific symbol demangler */
14492 ada_sniff_from_mangled_name,
14493 NULL, /* Language specific
14494 class_name_from_physname */
14495 ada_op_print_tab, /* expression operators for printing */
14496 0, /* c-style arrays */
14497 1, /* String lower bound */
14498 ada_get_gdb_completer_word_break_characters,
14499 ada_collect_symbol_completion_matches,
14500 ada_language_arch_info,
14501 ada_print_array_index,
14502 default_pass_by_reference,
14503 c_get_string,
14504 c_watch_location_expression,
14505 ada_get_symbol_name_matcher, /* la_get_symbol_name_matcher */
14506 ada_iterate_over_symbols,
14507 default_search_name_hash,
14508 &ada_varobj_ops,
14509 NULL,
14510 NULL,
14511 LANG_MAGIC
14512 };
14513
14514 /* Command-list for the "set/show ada" prefix command. */
14515 static struct cmd_list_element *set_ada_list;
14516 static struct cmd_list_element *show_ada_list;
14517
14518 /* Implement the "set ada" prefix command. */
14519
14520 static void
14521 set_ada_command (const char *arg, int from_tty)
14522 {
14523 printf_unfiltered (_(\
14524 "\"set ada\" must be followed by the name of a setting.\n"));
14525 help_list (set_ada_list, "set ada ", all_commands, gdb_stdout);
14526 }
14527
14528 /* Implement the "show ada" prefix command. */
14529
14530 static void
14531 show_ada_command (const char *args, int from_tty)
14532 {
14533 cmd_show_list (show_ada_list, from_tty, "");
14534 }
14535
14536 static void
14537 initialize_ada_catchpoint_ops (void)
14538 {
14539 struct breakpoint_ops *ops;
14540
14541 initialize_breakpoint_ops ();
14542
14543 ops = &catch_exception_breakpoint_ops;
14544 *ops = bkpt_breakpoint_ops;
14545 ops->allocate_location = allocate_location_catch_exception;
14546 ops->re_set = re_set_catch_exception;
14547 ops->check_status = check_status_catch_exception;
14548 ops->print_it = print_it_catch_exception;
14549 ops->print_one = print_one_catch_exception;
14550 ops->print_mention = print_mention_catch_exception;
14551 ops->print_recreate = print_recreate_catch_exception;
14552
14553 ops = &catch_exception_unhandled_breakpoint_ops;
14554 *ops = bkpt_breakpoint_ops;
14555 ops->allocate_location = allocate_location_catch_exception_unhandled;
14556 ops->re_set = re_set_catch_exception_unhandled;
14557 ops->check_status = check_status_catch_exception_unhandled;
14558 ops->print_it = print_it_catch_exception_unhandled;
14559 ops->print_one = print_one_catch_exception_unhandled;
14560 ops->print_mention = print_mention_catch_exception_unhandled;
14561 ops->print_recreate = print_recreate_catch_exception_unhandled;
14562
14563 ops = &catch_assert_breakpoint_ops;
14564 *ops = bkpt_breakpoint_ops;
14565 ops->allocate_location = allocate_location_catch_assert;
14566 ops->re_set = re_set_catch_assert;
14567 ops->check_status = check_status_catch_assert;
14568 ops->print_it = print_it_catch_assert;
14569 ops->print_one = print_one_catch_assert;
14570 ops->print_mention = print_mention_catch_assert;
14571 ops->print_recreate = print_recreate_catch_assert;
14572
14573 ops = &catch_handlers_breakpoint_ops;
14574 *ops = bkpt_breakpoint_ops;
14575 ops->allocate_location = allocate_location_catch_handlers;
14576 ops->re_set = re_set_catch_handlers;
14577 ops->check_status = check_status_catch_handlers;
14578 ops->print_it = print_it_catch_handlers;
14579 ops->print_one = print_one_catch_handlers;
14580 ops->print_mention = print_mention_catch_handlers;
14581 ops->print_recreate = print_recreate_catch_handlers;
14582 }
14583
14584 /* This module's 'new_objfile' observer. */
14585
14586 static void
14587 ada_new_objfile_observer (struct objfile *objfile)
14588 {
14589 ada_clear_symbol_cache ();
14590 }
14591
14592 /* This module's 'free_objfile' observer. */
14593
14594 static void
14595 ada_free_objfile_observer (struct objfile *objfile)
14596 {
14597 ada_clear_symbol_cache ();
14598 }
14599
14600 void
14601 _initialize_ada_language (void)
14602 {
14603 initialize_ada_catchpoint_ops ();
14604
14605 add_prefix_cmd ("ada", no_class, set_ada_command,
14606 _("Prefix command for changing Ada-specfic settings"),
14607 &set_ada_list, "set ada ", 0, &setlist);
14608
14609 add_prefix_cmd ("ada", no_class, show_ada_command,
14610 _("Generic command for showing Ada-specific settings."),
14611 &show_ada_list, "show ada ", 0, &showlist);
14612
14613 add_setshow_boolean_cmd ("trust-PAD-over-XVS", class_obscure,
14614 &trust_pad_over_xvs, _("\
14615 Enable or disable an optimization trusting PAD types over XVS types"), _("\
14616 Show whether an optimization trusting PAD types over XVS types is activated"),
14617 _("\
14618 This is related to the encoding used by the GNAT compiler. The debugger\n\
14619 should normally trust the contents of PAD types, but certain older versions\n\
14620 of GNAT have a bug that sometimes causes the information in the PAD type\n\
14621 to be incorrect. Turning this setting \"off\" allows the debugger to\n\
14622 work around this bug. It is always safe to turn this option \"off\", but\n\
14623 this incurs a slight performance penalty, so it is recommended to NOT change\n\
14624 this option to \"off\" unless necessary."),
14625 NULL, NULL, &set_ada_list, &show_ada_list);
14626
14627 add_setshow_boolean_cmd ("print-signatures", class_vars,
14628 &print_signatures, _("\
14629 Enable or disable the output of formal and return types for functions in the \
14630 overloads selection menu"), _("\
14631 Show whether the output of formal and return types for functions in the \
14632 overloads selection menu is activated"),
14633 NULL, NULL, NULL, &set_ada_list, &show_ada_list);
14634
14635 add_catch_command ("exception", _("\
14636 Catch Ada exceptions, when raised.\n\
14637 With an argument, catch only exceptions with the given name."),
14638 catch_ada_exception_command,
14639 NULL,
14640 CATCH_PERMANENT,
14641 CATCH_TEMPORARY);
14642
14643 add_catch_command ("handlers", _("\
14644 Catch Ada exceptions, when handled.\n\
14645 With an argument, catch only exceptions with the given name."),
14646 catch_ada_handlers_command,
14647 NULL,
14648 CATCH_PERMANENT,
14649 CATCH_TEMPORARY);
14650 add_catch_command ("assert", _("\
14651 Catch failed Ada assertions, when raised.\n\
14652 With an argument, catch only exceptions with the given name."),
14653 catch_assert_command,
14654 NULL,
14655 CATCH_PERMANENT,
14656 CATCH_TEMPORARY);
14657
14658 varsize_limit = 65536;
14659
14660 add_info ("exceptions", info_exceptions_command,
14661 _("\
14662 List all Ada exception names.\n\
14663 If a regular expression is passed as an argument, only those matching\n\
14664 the regular expression are listed."));
14665
14666 add_prefix_cmd ("ada", class_maintenance, maint_set_ada_cmd,
14667 _("Set Ada maintenance-related variables."),
14668 &maint_set_ada_cmdlist, "maintenance set ada ",
14669 0/*allow-unknown*/, &maintenance_set_cmdlist);
14670
14671 add_prefix_cmd ("ada", class_maintenance, maint_show_ada_cmd,
14672 _("Show Ada maintenance-related variables"),
14673 &maint_show_ada_cmdlist, "maintenance show ada ",
14674 0/*allow-unknown*/, &maintenance_show_cmdlist);
14675
14676 add_setshow_boolean_cmd
14677 ("ignore-descriptive-types", class_maintenance,
14678 &ada_ignore_descriptive_types_p,
14679 _("Set whether descriptive types generated by GNAT should be ignored."),
14680 _("Show whether descriptive types generated by GNAT should be ignored."),
14681 _("\
14682 When enabled, the debugger will stop using the DW_AT_GNAT_descriptive_type\n\
14683 DWARF attribute."),
14684 NULL, NULL, &maint_set_ada_cmdlist, &maint_show_ada_cmdlist);
14685
14686 decoded_names_store = htab_create_alloc
14687 (256, htab_hash_string, (int (*)(const void *, const void *)) streq,
14688 NULL, xcalloc, xfree);
14689
14690 /* The ada-lang observers. */
14691 observer_attach_new_objfile (ada_new_objfile_observer);
14692 observer_attach_free_objfile (ada_free_objfile_observer);
14693 observer_attach_inferior_exit (ada_inferior_exit);
14694
14695 /* Setup various context-specific data. */
14696 ada_inferior_data
14697 = register_inferior_data_with_cleanup (NULL, ada_inferior_data_cleanup);
14698 ada_pspace_data_handle
14699 = register_program_space_data_with_cleanup (NULL, ada_pspace_data_cleanup);
14700 }