b4eeaaf08663cde00bb5c9d5fe3e258c0bd110ec
[deliverable/binutils-gdb.git] / gdb / ada-lang.c
1 /* Ada language support routines for GDB, the GNU debugger.
2
3 Copyright (C) 1992-2020 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 "gdb_regex.h"
24 #include "frame.h"
25 #include "symtab.h"
26 #include "gdbtypes.h"
27 #include "gdbcmd.h"
28 #include "expression.h"
29 #include "parser-defs.h"
30 #include "language.h"
31 #include "varobj.h"
32 #include "inferior.h"
33 #include "symfile.h"
34 #include "objfiles.h"
35 #include "breakpoint.h"
36 #include "gdbcore.h"
37 #include "hashtab.h"
38 #include "gdb_obstack.h"
39 #include "ada-lang.h"
40 #include "completer.h"
41 #include "ui-out.h"
42 #include "block.h"
43 #include "infcall.h"
44 #include "annotate.h"
45 #include "valprint.h"
46 #include "source.h"
47 #include "observable.h"
48 #include "stack.h"
49 #include "typeprint.h"
50 #include "namespace.h"
51 #include "cli/cli-style.h"
52
53 #include "value.h"
54 #include "mi/mi-common.h"
55 #include "arch-utils.h"
56 #include "cli/cli-utils.h"
57 #include "gdbsupport/function-view.h"
58 #include "gdbsupport/byte-vector.h"
59 #include <algorithm>
60
61 /* Define whether or not the C operator '/' truncates towards zero for
62 differently signed operands (truncation direction is undefined in C).
63 Copied from valarith.c. */
64
65 #ifndef TRUNCATION_TOWARDS_ZERO
66 #define TRUNCATION_TOWARDS_ZERO ((-5 / 2) == -2)
67 #endif
68
69 static struct type *desc_base_type (struct type *);
70
71 static struct type *desc_bounds_type (struct type *);
72
73 static struct value *desc_bounds (struct value *);
74
75 static int fat_pntr_bounds_bitpos (struct type *);
76
77 static int fat_pntr_bounds_bitsize (struct type *);
78
79 static struct type *desc_data_target_type (struct type *);
80
81 static struct value *desc_data (struct value *);
82
83 static int fat_pntr_data_bitpos (struct type *);
84
85 static int fat_pntr_data_bitsize (struct type *);
86
87 static struct value *desc_one_bound (struct value *, int, int);
88
89 static int desc_bound_bitpos (struct type *, int, int);
90
91 static int desc_bound_bitsize (struct type *, int, int);
92
93 static struct type *desc_index_type (struct type *, int);
94
95 static int desc_arity (struct type *);
96
97 static int ada_type_match (struct type *, struct type *, int);
98
99 static int ada_args_match (struct symbol *, struct value **, int);
100
101 static struct value *make_array_descriptor (struct type *, struct value *);
102
103 static void ada_add_block_symbols (struct obstack *,
104 const struct block *,
105 const lookup_name_info &lookup_name,
106 domain_enum, struct objfile *);
107
108 static void ada_add_all_symbols (struct obstack *, const struct block *,
109 const lookup_name_info &lookup_name,
110 domain_enum, int, int *);
111
112 static int is_nonfunction (struct block_symbol *, int);
113
114 static void add_defn_to_vec (struct obstack *, struct symbol *,
115 const struct block *);
116
117 static int num_defns_collected (struct obstack *);
118
119 static struct block_symbol *defns_collected (struct obstack *, int);
120
121 static struct value *resolve_subexp (expression_up *, int *, int,
122 struct type *, int,
123 innermost_block_tracker *);
124
125 static void replace_operator_with_call (expression_up *, int, int, int,
126 struct symbol *, const struct block *);
127
128 static int possible_user_operator_p (enum exp_opcode, struct value **);
129
130 static const char *ada_op_name (enum exp_opcode);
131
132 static const char *ada_decoded_op_name (enum exp_opcode);
133
134 static int numeric_type_p (struct type *);
135
136 static int integer_type_p (struct type *);
137
138 static int scalar_type_p (struct type *);
139
140 static int discrete_type_p (struct type *);
141
142 static struct type *ada_lookup_struct_elt_type (struct type *, const char *,
143 int, int);
144
145 static struct value *evaluate_subexp_type (struct expression *, int *);
146
147 static struct type *ada_find_parallel_type_with_name (struct type *,
148 const char *);
149
150 static int is_dynamic_field (struct type *, int);
151
152 static struct type *to_fixed_variant_branch_type (struct type *,
153 const gdb_byte *,
154 CORE_ADDR, struct value *);
155
156 static struct type *to_fixed_array_type (struct type *, struct value *, int);
157
158 static struct type *to_fixed_range_type (struct type *, struct value *);
159
160 static struct type *to_static_fixed_type (struct type *);
161 static struct type *static_unwrap_type (struct type *type);
162
163 static struct value *unwrap_value (struct value *);
164
165 static struct type *constrained_packed_array_type (struct type *, long *);
166
167 static struct type *decode_constrained_packed_array_type (struct type *);
168
169 static long decode_packed_array_bitsize (struct type *);
170
171 static struct value *decode_constrained_packed_array (struct value *);
172
173 static int ada_is_packed_array_type (struct type *);
174
175 static int ada_is_unconstrained_packed_array_type (struct type *);
176
177 static struct value *value_subscript_packed (struct value *, int,
178 struct value **);
179
180 static struct value *coerce_unspec_val_to_type (struct value *,
181 struct type *);
182
183 static int lesseq_defined_than (struct symbol *, struct symbol *);
184
185 static int equiv_types (struct type *, struct type *);
186
187 static int is_name_suffix (const char *);
188
189 static int advance_wild_match (const char **, const char *, int);
190
191 static bool wild_match (const char *name, const char *patn);
192
193 static struct value *ada_coerce_ref (struct value *);
194
195 static LONGEST pos_atr (struct value *);
196
197 static struct value *value_pos_atr (struct type *, struct value *);
198
199 static struct value *val_atr (struct type *, LONGEST);
200
201 static struct value *value_val_atr (struct type *, struct value *);
202
203 static struct symbol *standard_lookup (const char *, const struct block *,
204 domain_enum);
205
206 static struct value *ada_search_struct_field (const char *, struct value *, int,
207 struct type *);
208
209 static int find_struct_field (const char *, struct type *, int,
210 struct type **, int *, int *, int *, int *);
211
212 static int ada_resolve_function (struct block_symbol *, int,
213 struct value **, int, const char *,
214 struct type *, int);
215
216 static int ada_is_direct_array_type (struct type *);
217
218 static void ada_language_arch_info (struct gdbarch *,
219 struct language_arch_info *);
220
221 static struct value *ada_index_struct_field (int, struct value *, int,
222 struct type *);
223
224 static struct value *assign_aggregate (struct value *, struct value *,
225 struct expression *,
226 int *, enum noside);
227
228 static void aggregate_assign_from_choices (struct value *, struct value *,
229 struct expression *,
230 int *, LONGEST *, int *,
231 int, LONGEST, LONGEST);
232
233 static void aggregate_assign_positional (struct value *, struct value *,
234 struct expression *,
235 int *, LONGEST *, int *, int,
236 LONGEST, LONGEST);
237
238
239 static void aggregate_assign_others (struct value *, struct value *,
240 struct expression *,
241 int *, LONGEST *, int, LONGEST, LONGEST);
242
243
244 static void add_component_interval (LONGEST, LONGEST, LONGEST *, int *, int);
245
246
247 static struct value *ada_evaluate_subexp (struct type *, struct expression *,
248 int *, enum noside);
249
250 static void ada_forward_operator_length (struct expression *, int, int *,
251 int *);
252
253 static struct type *ada_find_any_type (const char *name);
254
255 static symbol_name_matcher_ftype *ada_get_symbol_name_matcher
256 (const lookup_name_info &lookup_name);
257
258 \f
259
260 /* The result of a symbol lookup to be stored in our symbol cache. */
261
262 struct cache_entry
263 {
264 /* The name used to perform the lookup. */
265 const char *name;
266 /* The namespace used during the lookup. */
267 domain_enum domain;
268 /* The symbol returned by the lookup, or NULL if no matching symbol
269 was found. */
270 struct symbol *sym;
271 /* The block where the symbol was found, or NULL if no matching
272 symbol was found. */
273 const struct block *block;
274 /* A pointer to the next entry with the same hash. */
275 struct cache_entry *next;
276 };
277
278 /* The Ada symbol cache, used to store the result of Ada-mode symbol
279 lookups in the course of executing the user's commands.
280
281 The cache is implemented using a simple, fixed-sized hash.
282 The size is fixed on the grounds that there are not likely to be
283 all that many symbols looked up during any given session, regardless
284 of the size of the symbol table. If we decide to go to a resizable
285 table, let's just use the stuff from libiberty instead. */
286
287 #define HASH_SIZE 1009
288
289 struct ada_symbol_cache
290 {
291 /* An obstack used to store the entries in our cache. */
292 struct obstack cache_space;
293
294 /* The root of the hash table used to implement our symbol cache. */
295 struct cache_entry *root[HASH_SIZE];
296 };
297
298 static void ada_free_symbol_cache (struct ada_symbol_cache *sym_cache);
299
300 /* Maximum-sized dynamic type. */
301 static unsigned int varsize_limit;
302
303 static const char ada_completer_word_break_characters[] =
304 #ifdef VMS
305 " \t\n!@#%^&*()+=|~`}{[]\";:?/,-";
306 #else
307 " \t\n!@#$%^&*()+=|~`}{[]\";:?/,-";
308 #endif
309
310 /* The name of the symbol to use to get the name of the main subprogram. */
311 static const char ADA_MAIN_PROGRAM_SYMBOL_NAME[]
312 = "__gnat_ada_main_program_name";
313
314 /* Limit on the number of warnings to raise per expression evaluation. */
315 static int warning_limit = 2;
316
317 /* Number of warning messages issued; reset to 0 by cleanups after
318 expression evaluation. */
319 static int warnings_issued = 0;
320
321 static const char *known_runtime_file_name_patterns[] = {
322 ADA_KNOWN_RUNTIME_FILE_NAME_PATTERNS NULL
323 };
324
325 static const char *known_auxiliary_function_name_patterns[] = {
326 ADA_KNOWN_AUXILIARY_FUNCTION_NAME_PATTERNS NULL
327 };
328
329 /* Maintenance-related settings for this module. */
330
331 static struct cmd_list_element *maint_set_ada_cmdlist;
332 static struct cmd_list_element *maint_show_ada_cmdlist;
333
334 /* The "maintenance ada set/show ignore-descriptive-type" value. */
335
336 static bool ada_ignore_descriptive_types_p = false;
337
338 /* Inferior-specific data. */
339
340 /* Per-inferior data for this module. */
341
342 struct ada_inferior_data
343 {
344 /* The ada__tags__type_specific_data type, which is used when decoding
345 tagged types. With older versions of GNAT, this type was directly
346 accessible through a component ("tsd") in the object tag. But this
347 is no longer the case, so we cache it for each inferior. */
348 struct type *tsd_type = nullptr;
349
350 /* The exception_support_info data. This data is used to determine
351 how to implement support for Ada exception catchpoints in a given
352 inferior. */
353 const struct exception_support_info *exception_info = nullptr;
354 };
355
356 /* Our key to this module's inferior data. */
357 static const struct inferior_key<ada_inferior_data> ada_inferior_data;
358
359 /* Return our inferior data for the given inferior (INF).
360
361 This function always returns a valid pointer to an allocated
362 ada_inferior_data structure. If INF's inferior data has not
363 been previously set, this functions creates a new one with all
364 fields set to zero, sets INF's inferior to it, and then returns
365 a pointer to that newly allocated ada_inferior_data. */
366
367 static struct ada_inferior_data *
368 get_ada_inferior_data (struct inferior *inf)
369 {
370 struct ada_inferior_data *data;
371
372 data = ada_inferior_data.get (inf);
373 if (data == NULL)
374 data = ada_inferior_data.emplace (inf);
375
376 return data;
377 }
378
379 /* Perform all necessary cleanups regarding our module's inferior data
380 that is required after the inferior INF just exited. */
381
382 static void
383 ada_inferior_exit (struct inferior *inf)
384 {
385 ada_inferior_data.clear (inf);
386 }
387
388
389 /* program-space-specific data. */
390
391 /* This module's per-program-space data. */
392 struct ada_pspace_data
393 {
394 ~ada_pspace_data ()
395 {
396 if (sym_cache != NULL)
397 ada_free_symbol_cache (sym_cache);
398 }
399
400 /* The Ada symbol cache. */
401 struct ada_symbol_cache *sym_cache = nullptr;
402 };
403
404 /* Key to our per-program-space data. */
405 static const struct program_space_key<ada_pspace_data> ada_pspace_data_handle;
406
407 /* Return this module's data for the given program space (PSPACE).
408 If not is found, add a zero'ed one now.
409
410 This function always returns a valid object. */
411
412 static struct ada_pspace_data *
413 get_ada_pspace_data (struct program_space *pspace)
414 {
415 struct ada_pspace_data *data;
416
417 data = ada_pspace_data_handle.get (pspace);
418 if (data == NULL)
419 data = ada_pspace_data_handle.emplace (pspace);
420
421 return data;
422 }
423
424 /* Utilities */
425
426 /* If TYPE is a TYPE_CODE_TYPEDEF type, return the target type after
427 all typedef layers have been peeled. Otherwise, return TYPE.
428
429 Normally, we really expect a typedef type to only have 1 typedef layer.
430 In other words, we really expect the target type of a typedef type to be
431 a non-typedef type. This is particularly true for Ada units, because
432 the language does not have a typedef vs not-typedef distinction.
433 In that respect, the Ada compiler has been trying to eliminate as many
434 typedef definitions in the debugging information, since they generally
435 do not bring any extra information (we still use typedef under certain
436 circumstances related mostly to the GNAT encoding).
437
438 Unfortunately, we have seen situations where the debugging information
439 generated by the compiler leads to such multiple typedef layers. For
440 instance, consider the following example with stabs:
441
442 .stabs "pck__float_array___XUP:Tt(0,46)=s16P_ARRAY:(0,47)=[...]"[...]
443 .stabs "pck__float_array___XUP:t(0,36)=(0,46)",128,0,6,0
444
445 This is an error in the debugging information which causes type
446 pck__float_array___XUP to be defined twice, and the second time,
447 it is defined as a typedef of a typedef.
448
449 This is on the fringe of legality as far as debugging information is
450 concerned, and certainly unexpected. But it is easy to handle these
451 situations correctly, so we can afford to be lenient in this case. */
452
453 static struct type *
454 ada_typedef_target_type (struct type *type)
455 {
456 while (type->code () == TYPE_CODE_TYPEDEF)
457 type = TYPE_TARGET_TYPE (type);
458 return type;
459 }
460
461 /* Given DECODED_NAME a string holding a symbol name in its
462 decoded form (ie using the Ada dotted notation), returns
463 its unqualified name. */
464
465 static const char *
466 ada_unqualified_name (const char *decoded_name)
467 {
468 const char *result;
469
470 /* If the decoded name starts with '<', it means that the encoded
471 name does not follow standard naming conventions, and thus that
472 it is not your typical Ada symbol name. Trying to unqualify it
473 is therefore pointless and possibly erroneous. */
474 if (decoded_name[0] == '<')
475 return decoded_name;
476
477 result = strrchr (decoded_name, '.');
478 if (result != NULL)
479 result++; /* Skip the dot... */
480 else
481 result = decoded_name;
482
483 return result;
484 }
485
486 /* Return a string starting with '<', followed by STR, and '>'. */
487
488 static std::string
489 add_angle_brackets (const char *str)
490 {
491 return string_printf ("<%s>", str);
492 }
493
494 static const char *
495 ada_get_gdb_completer_word_break_characters (void)
496 {
497 return ada_completer_word_break_characters;
498 }
499
500 /* Print an array element index using the Ada syntax. */
501
502 static void
503 ada_print_array_index (struct type *index_type, LONGEST index,
504 struct ui_file *stream,
505 const struct value_print_options *options)
506 {
507 struct value *index_value = val_atr (index_type, index);
508
509 LA_VALUE_PRINT (index_value, stream, options);
510 fprintf_filtered (stream, " => ");
511 }
512
513 /* la_watch_location_expression for Ada. */
514
515 static gdb::unique_xmalloc_ptr<char>
516 ada_watch_location_expression (struct type *type, CORE_ADDR addr)
517 {
518 type = check_typedef (TYPE_TARGET_TYPE (check_typedef (type)));
519 std::string name = type_to_string (type);
520 return gdb::unique_xmalloc_ptr<char>
521 (xstrprintf ("{%s} %s", name.c_str (), core_addr_to_string (addr)));
522 }
523
524 /* Assuming V points to an array of S objects, make sure that it contains at
525 least M objects, updating V and S as necessary. */
526
527 #define GROW_VECT(v, s, m) \
528 if ((s) < (m)) (v) = (char *) grow_vect (v, &(s), m, sizeof *(v));
529
530 /* Assuming VECT points to an array of *SIZE objects of size
531 ELEMENT_SIZE, grow it to contain at least MIN_SIZE objects,
532 updating *SIZE as necessary and returning the (new) array. */
533
534 static void *
535 grow_vect (void *vect, size_t *size, size_t min_size, int element_size)
536 {
537 if (*size < min_size)
538 {
539 *size *= 2;
540 if (*size < min_size)
541 *size = min_size;
542 vect = xrealloc (vect, *size * element_size);
543 }
544 return vect;
545 }
546
547 /* True (non-zero) iff TARGET matches FIELD_NAME up to any trailing
548 suffix of FIELD_NAME beginning "___". */
549
550 static int
551 field_name_match (const char *field_name, const char *target)
552 {
553 int len = strlen (target);
554
555 return
556 (strncmp (field_name, target, len) == 0
557 && (field_name[len] == '\0'
558 || (startswith (field_name + len, "___")
559 && strcmp (field_name + strlen (field_name) - 6,
560 "___XVN") != 0)));
561 }
562
563
564 /* Assuming TYPE is a TYPE_CODE_STRUCT or a TYPE_CODE_TYPDEF to
565 a TYPE_CODE_STRUCT, find the field whose name matches FIELD_NAME,
566 and return its index. This function also handles fields whose name
567 have ___ suffixes because the compiler sometimes alters their name
568 by adding such a suffix to represent fields with certain constraints.
569 If the field could not be found, return a negative number if
570 MAYBE_MISSING is set. Otherwise raise an error. */
571
572 int
573 ada_get_field_index (const struct type *type, const char *field_name,
574 int maybe_missing)
575 {
576 int fieldno;
577 struct type *struct_type = check_typedef ((struct type *) type);
578
579 for (fieldno = 0; fieldno < struct_type->num_fields (); fieldno++)
580 if (field_name_match (TYPE_FIELD_NAME (struct_type, fieldno), field_name))
581 return fieldno;
582
583 if (!maybe_missing)
584 error (_("Unable to find field %s in struct %s. Aborting"),
585 field_name, struct_type->name ());
586
587 return -1;
588 }
589
590 /* The length of the prefix of NAME prior to any "___" suffix. */
591
592 int
593 ada_name_prefix_len (const char *name)
594 {
595 if (name == NULL)
596 return 0;
597 else
598 {
599 const char *p = strstr (name, "___");
600
601 if (p == NULL)
602 return strlen (name);
603 else
604 return p - name;
605 }
606 }
607
608 /* Return non-zero if SUFFIX is a suffix of STR.
609 Return zero if STR is null. */
610
611 static int
612 is_suffix (const char *str, const char *suffix)
613 {
614 int len1, len2;
615
616 if (str == NULL)
617 return 0;
618 len1 = strlen (str);
619 len2 = strlen (suffix);
620 return (len1 >= len2 && strcmp (str + len1 - len2, suffix) == 0);
621 }
622
623 /* The contents of value VAL, treated as a value of type TYPE. The
624 result is an lval in memory if VAL is. */
625
626 static struct value *
627 coerce_unspec_val_to_type (struct value *val, struct type *type)
628 {
629 type = ada_check_typedef (type);
630 if (value_type (val) == type)
631 return val;
632 else
633 {
634 struct value *result;
635
636 /* Make sure that the object size is not unreasonable before
637 trying to allocate some memory for it. */
638 ada_ensure_varsize_limit (type);
639
640 if (value_lazy (val)
641 || TYPE_LENGTH (type) > TYPE_LENGTH (value_type (val)))
642 result = allocate_value_lazy (type);
643 else
644 {
645 result = allocate_value (type);
646 value_contents_copy_raw (result, 0, val, 0, TYPE_LENGTH (type));
647 }
648 set_value_component_location (result, val);
649 set_value_bitsize (result, value_bitsize (val));
650 set_value_bitpos (result, value_bitpos (val));
651 if (VALUE_LVAL (result) == lval_memory)
652 set_value_address (result, value_address (val));
653 return result;
654 }
655 }
656
657 static const gdb_byte *
658 cond_offset_host (const gdb_byte *valaddr, long offset)
659 {
660 if (valaddr == NULL)
661 return NULL;
662 else
663 return valaddr + offset;
664 }
665
666 static CORE_ADDR
667 cond_offset_target (CORE_ADDR address, long offset)
668 {
669 if (address == 0)
670 return 0;
671 else
672 return address + offset;
673 }
674
675 /* Issue a warning (as for the definition of warning in utils.c, but
676 with exactly one argument rather than ...), unless the limit on the
677 number of warnings has passed during the evaluation of the current
678 expression. */
679
680 /* FIXME: cagney/2004-10-10: This function is mimicking the behavior
681 provided by "complaint". */
682 static void lim_warning (const char *format, ...) ATTRIBUTE_PRINTF (1, 2);
683
684 static void
685 lim_warning (const char *format, ...)
686 {
687 va_list args;
688
689 va_start (args, format);
690 warnings_issued += 1;
691 if (warnings_issued <= warning_limit)
692 vwarning (format, args);
693
694 va_end (args);
695 }
696
697 /* Issue an error if the size of an object of type T is unreasonable,
698 i.e. if it would be a bad idea to allocate a value of this type in
699 GDB. */
700
701 void
702 ada_ensure_varsize_limit (const struct type *type)
703 {
704 if (TYPE_LENGTH (type) > varsize_limit)
705 error (_("object size is larger than varsize-limit"));
706 }
707
708 /* Maximum value of a SIZE-byte signed integer type. */
709 static LONGEST
710 max_of_size (int size)
711 {
712 LONGEST top_bit = (LONGEST) 1 << (size * 8 - 2);
713
714 return top_bit | (top_bit - 1);
715 }
716
717 /* Minimum value of a SIZE-byte signed integer type. */
718 static LONGEST
719 min_of_size (int size)
720 {
721 return -max_of_size (size) - 1;
722 }
723
724 /* Maximum value of a SIZE-byte unsigned integer type. */
725 static ULONGEST
726 umax_of_size (int size)
727 {
728 ULONGEST top_bit = (ULONGEST) 1 << (size * 8 - 1);
729
730 return top_bit | (top_bit - 1);
731 }
732
733 /* Maximum value of integral type T, as a signed quantity. */
734 static LONGEST
735 max_of_type (struct type *t)
736 {
737 if (TYPE_UNSIGNED (t))
738 return (LONGEST) umax_of_size (TYPE_LENGTH (t));
739 else
740 return max_of_size (TYPE_LENGTH (t));
741 }
742
743 /* Minimum value of integral type T, as a signed quantity. */
744 static LONGEST
745 min_of_type (struct type *t)
746 {
747 if (TYPE_UNSIGNED (t))
748 return 0;
749 else
750 return min_of_size (TYPE_LENGTH (t));
751 }
752
753 /* The largest value in the domain of TYPE, a discrete type, as an integer. */
754 LONGEST
755 ada_discrete_type_high_bound (struct type *type)
756 {
757 type = resolve_dynamic_type (type, {}, 0);
758 switch (type->code ())
759 {
760 case TYPE_CODE_RANGE:
761 return TYPE_HIGH_BOUND (type);
762 case TYPE_CODE_ENUM:
763 return TYPE_FIELD_ENUMVAL (type, type->num_fields () - 1);
764 case TYPE_CODE_BOOL:
765 return 1;
766 case TYPE_CODE_CHAR:
767 case TYPE_CODE_INT:
768 return max_of_type (type);
769 default:
770 error (_("Unexpected type in ada_discrete_type_high_bound."));
771 }
772 }
773
774 /* The smallest value in the domain of TYPE, a discrete type, as an integer. */
775 LONGEST
776 ada_discrete_type_low_bound (struct type *type)
777 {
778 type = resolve_dynamic_type (type, {}, 0);
779 switch (type->code ())
780 {
781 case TYPE_CODE_RANGE:
782 return TYPE_LOW_BOUND (type);
783 case TYPE_CODE_ENUM:
784 return TYPE_FIELD_ENUMVAL (type, 0);
785 case TYPE_CODE_BOOL:
786 return 0;
787 case TYPE_CODE_CHAR:
788 case TYPE_CODE_INT:
789 return min_of_type (type);
790 default:
791 error (_("Unexpected type in ada_discrete_type_low_bound."));
792 }
793 }
794
795 /* The identity on non-range types. For range types, the underlying
796 non-range scalar type. */
797
798 static struct type *
799 get_base_type (struct type *type)
800 {
801 while (type != NULL && type->code () == TYPE_CODE_RANGE)
802 {
803 if (type == TYPE_TARGET_TYPE (type) || TYPE_TARGET_TYPE (type) == NULL)
804 return type;
805 type = TYPE_TARGET_TYPE (type);
806 }
807 return type;
808 }
809
810 /* Return a decoded version of the given VALUE. This means returning
811 a value whose type is obtained by applying all the GNAT-specific
812 encodings, making the resulting type a static but standard description
813 of the initial type. */
814
815 struct value *
816 ada_get_decoded_value (struct value *value)
817 {
818 struct type *type = ada_check_typedef (value_type (value));
819
820 if (ada_is_array_descriptor_type (type)
821 || (ada_is_constrained_packed_array_type (type)
822 && type->code () != TYPE_CODE_PTR))
823 {
824 if (type->code () == TYPE_CODE_TYPEDEF) /* array access type. */
825 value = ada_coerce_to_simple_array_ptr (value);
826 else
827 value = ada_coerce_to_simple_array (value);
828 }
829 else
830 value = ada_to_fixed_value (value);
831
832 return value;
833 }
834
835 /* Same as ada_get_decoded_value, but with the given TYPE.
836 Because there is no associated actual value for this type,
837 the resulting type might be a best-effort approximation in
838 the case of dynamic types. */
839
840 struct type *
841 ada_get_decoded_type (struct type *type)
842 {
843 type = to_static_fixed_type (type);
844 if (ada_is_constrained_packed_array_type (type))
845 type = ada_coerce_to_simple_array_type (type);
846 return type;
847 }
848
849 \f
850
851 /* Language Selection */
852
853 /* If the main program is in Ada, return language_ada, otherwise return LANG
854 (the main program is in Ada iif the adainit symbol is found). */
855
856 static enum language
857 ada_update_initial_language (enum language lang)
858 {
859 if (lookup_minimal_symbol ("adainit", NULL, NULL).minsym != NULL)
860 return language_ada;
861
862 return lang;
863 }
864
865 /* If the main procedure is written in Ada, then return its name.
866 The result is good until the next call. Return NULL if the main
867 procedure doesn't appear to be in Ada. */
868
869 char *
870 ada_main_name (void)
871 {
872 struct bound_minimal_symbol msym;
873 static gdb::unique_xmalloc_ptr<char> main_program_name;
874
875 /* For Ada, the name of the main procedure is stored in a specific
876 string constant, generated by the binder. Look for that symbol,
877 extract its address, and then read that string. If we didn't find
878 that string, then most probably the main procedure is not written
879 in Ada. */
880 msym = lookup_minimal_symbol (ADA_MAIN_PROGRAM_SYMBOL_NAME, NULL, NULL);
881
882 if (msym.minsym != NULL)
883 {
884 CORE_ADDR main_program_name_addr;
885 int err_code;
886
887 main_program_name_addr = BMSYMBOL_VALUE_ADDRESS (msym);
888 if (main_program_name_addr == 0)
889 error (_("Invalid address for Ada main program name."));
890
891 target_read_string (main_program_name_addr, &main_program_name,
892 1024, &err_code);
893
894 if (err_code != 0)
895 return NULL;
896 return main_program_name.get ();
897 }
898
899 /* The main procedure doesn't seem to be in Ada. */
900 return NULL;
901 }
902 \f
903 /* Symbols */
904
905 /* Table of Ada operators and their GNAT-encoded names. Last entry is pair
906 of NULLs. */
907
908 const struct ada_opname_map ada_opname_table[] = {
909 {"Oadd", "\"+\"", BINOP_ADD},
910 {"Osubtract", "\"-\"", BINOP_SUB},
911 {"Omultiply", "\"*\"", BINOP_MUL},
912 {"Odivide", "\"/\"", BINOP_DIV},
913 {"Omod", "\"mod\"", BINOP_MOD},
914 {"Orem", "\"rem\"", BINOP_REM},
915 {"Oexpon", "\"**\"", BINOP_EXP},
916 {"Olt", "\"<\"", BINOP_LESS},
917 {"Ole", "\"<=\"", BINOP_LEQ},
918 {"Ogt", "\">\"", BINOP_GTR},
919 {"Oge", "\">=\"", BINOP_GEQ},
920 {"Oeq", "\"=\"", BINOP_EQUAL},
921 {"One", "\"/=\"", BINOP_NOTEQUAL},
922 {"Oand", "\"and\"", BINOP_BITWISE_AND},
923 {"Oor", "\"or\"", BINOP_BITWISE_IOR},
924 {"Oxor", "\"xor\"", BINOP_BITWISE_XOR},
925 {"Oconcat", "\"&\"", BINOP_CONCAT},
926 {"Oabs", "\"abs\"", UNOP_ABS},
927 {"Onot", "\"not\"", UNOP_LOGICAL_NOT},
928 {"Oadd", "\"+\"", UNOP_PLUS},
929 {"Osubtract", "\"-\"", UNOP_NEG},
930 {NULL, NULL}
931 };
932
933 /* The "encoded" form of DECODED, according to GNAT conventions. The
934 result is valid until the next call to ada_encode. If
935 THROW_ERRORS, throw an error if invalid operator name is found.
936 Otherwise, return NULL in that case. */
937
938 static char *
939 ada_encode_1 (const char *decoded, bool throw_errors)
940 {
941 static char *encoding_buffer = NULL;
942 static size_t encoding_buffer_size = 0;
943 const char *p;
944 int k;
945
946 if (decoded == NULL)
947 return NULL;
948
949 GROW_VECT (encoding_buffer, encoding_buffer_size,
950 2 * strlen (decoded) + 10);
951
952 k = 0;
953 for (p = decoded; *p != '\0'; p += 1)
954 {
955 if (*p == '.')
956 {
957 encoding_buffer[k] = encoding_buffer[k + 1] = '_';
958 k += 2;
959 }
960 else if (*p == '"')
961 {
962 const struct ada_opname_map *mapping;
963
964 for (mapping = ada_opname_table;
965 mapping->encoded != NULL
966 && !startswith (p, mapping->decoded); mapping += 1)
967 ;
968 if (mapping->encoded == NULL)
969 {
970 if (throw_errors)
971 error (_("invalid Ada operator name: %s"), p);
972 else
973 return NULL;
974 }
975 strcpy (encoding_buffer + k, mapping->encoded);
976 k += strlen (mapping->encoded);
977 break;
978 }
979 else
980 {
981 encoding_buffer[k] = *p;
982 k += 1;
983 }
984 }
985
986 encoding_buffer[k] = '\0';
987 return encoding_buffer;
988 }
989
990 /* The "encoded" form of DECODED, according to GNAT conventions.
991 The result is valid until the next call to ada_encode. */
992
993 char *
994 ada_encode (const char *decoded)
995 {
996 return ada_encode_1 (decoded, true);
997 }
998
999 /* Return NAME folded to lower case, or, if surrounded by single
1000 quotes, unfolded, but with the quotes stripped away. Result good
1001 to next call. */
1002
1003 static char *
1004 ada_fold_name (gdb::string_view name)
1005 {
1006 static char *fold_buffer = NULL;
1007 static size_t fold_buffer_size = 0;
1008
1009 int len = name.size ();
1010 GROW_VECT (fold_buffer, fold_buffer_size, len + 1);
1011
1012 if (name[0] == '\'')
1013 {
1014 strncpy (fold_buffer, name.data () + 1, len - 2);
1015 fold_buffer[len - 2] = '\000';
1016 }
1017 else
1018 {
1019 int i;
1020
1021 for (i = 0; i <= len; i += 1)
1022 fold_buffer[i] = tolower (name[i]);
1023 }
1024
1025 return fold_buffer;
1026 }
1027
1028 /* Return nonzero if C is either a digit or a lowercase alphabet character. */
1029
1030 static int
1031 is_lower_alphanum (const char c)
1032 {
1033 return (isdigit (c) || (isalpha (c) && islower (c)));
1034 }
1035
1036 /* ENCODED is the linkage name of a symbol and LEN contains its length.
1037 This function saves in LEN the length of that same symbol name but
1038 without either of these suffixes:
1039 . .{DIGIT}+
1040 . ${DIGIT}+
1041 . ___{DIGIT}+
1042 . __{DIGIT}+.
1043
1044 These are suffixes introduced by the compiler for entities such as
1045 nested subprogram for instance, in order to avoid name clashes.
1046 They do not serve any purpose for the debugger. */
1047
1048 static void
1049 ada_remove_trailing_digits (const char *encoded, int *len)
1050 {
1051 if (*len > 1 && isdigit (encoded[*len - 1]))
1052 {
1053 int i = *len - 2;
1054
1055 while (i > 0 && isdigit (encoded[i]))
1056 i--;
1057 if (i >= 0 && encoded[i] == '.')
1058 *len = i;
1059 else if (i >= 0 && encoded[i] == '$')
1060 *len = i;
1061 else if (i >= 2 && startswith (encoded + i - 2, "___"))
1062 *len = i - 2;
1063 else if (i >= 1 && startswith (encoded + i - 1, "__"))
1064 *len = i - 1;
1065 }
1066 }
1067
1068 /* Remove the suffix introduced by the compiler for protected object
1069 subprograms. */
1070
1071 static void
1072 ada_remove_po_subprogram_suffix (const char *encoded, int *len)
1073 {
1074 /* Remove trailing N. */
1075
1076 /* Protected entry subprograms are broken into two
1077 separate subprograms: The first one is unprotected, and has
1078 a 'N' suffix; the second is the protected version, and has
1079 the 'P' suffix. The second calls the first one after handling
1080 the protection. Since the P subprograms are internally generated,
1081 we leave these names undecoded, giving the user a clue that this
1082 entity is internal. */
1083
1084 if (*len > 1
1085 && encoded[*len - 1] == 'N'
1086 && (isdigit (encoded[*len - 2]) || islower (encoded[*len - 2])))
1087 *len = *len - 1;
1088 }
1089
1090 /* If ENCODED follows the GNAT entity encoding conventions, then return
1091 the decoded form of ENCODED. Otherwise, return "<%s>" where "%s" is
1092 replaced by ENCODED. */
1093
1094 std::string
1095 ada_decode (const char *encoded)
1096 {
1097 int i, j;
1098 int len0;
1099 const char *p;
1100 int at_start_name;
1101 std::string decoded;
1102
1103 /* With function descriptors on PPC64, the value of a symbol named
1104 ".FN", if it exists, is the entry point of the function "FN". */
1105 if (encoded[0] == '.')
1106 encoded += 1;
1107
1108 /* The name of the Ada main procedure starts with "_ada_".
1109 This prefix is not part of the decoded name, so skip this part
1110 if we see this prefix. */
1111 if (startswith (encoded, "_ada_"))
1112 encoded += 5;
1113
1114 /* If the name starts with '_', then it is not a properly encoded
1115 name, so do not attempt to decode it. Similarly, if the name
1116 starts with '<', the name should not be decoded. */
1117 if (encoded[0] == '_' || encoded[0] == '<')
1118 goto Suppress;
1119
1120 len0 = strlen (encoded);
1121
1122 ada_remove_trailing_digits (encoded, &len0);
1123 ada_remove_po_subprogram_suffix (encoded, &len0);
1124
1125 /* Remove the ___X.* suffix if present. Do not forget to verify that
1126 the suffix is located before the current "end" of ENCODED. We want
1127 to avoid re-matching parts of ENCODED that have previously been
1128 marked as discarded (by decrementing LEN0). */
1129 p = strstr (encoded, "___");
1130 if (p != NULL && p - encoded < len0 - 3)
1131 {
1132 if (p[3] == 'X')
1133 len0 = p - encoded;
1134 else
1135 goto Suppress;
1136 }
1137
1138 /* Remove any trailing TKB suffix. It tells us that this symbol
1139 is for the body of a task, but that information does not actually
1140 appear in the decoded name. */
1141
1142 if (len0 > 3 && startswith (encoded + len0 - 3, "TKB"))
1143 len0 -= 3;
1144
1145 /* Remove any trailing TB suffix. The TB suffix is slightly different
1146 from the TKB suffix because it is used for non-anonymous task
1147 bodies. */
1148
1149 if (len0 > 2 && startswith (encoded + len0 - 2, "TB"))
1150 len0 -= 2;
1151
1152 /* Remove trailing "B" suffixes. */
1153 /* FIXME: brobecker/2006-04-19: Not sure what this are used for... */
1154
1155 if (len0 > 1 && startswith (encoded + len0 - 1, "B"))
1156 len0 -= 1;
1157
1158 /* Make decoded big enough for possible expansion by operator name. */
1159
1160 decoded.resize (2 * len0 + 1, 'X');
1161
1162 /* Remove trailing __{digit}+ or trailing ${digit}+. */
1163
1164 if (len0 > 1 && isdigit (encoded[len0 - 1]))
1165 {
1166 i = len0 - 2;
1167 while ((i >= 0 && isdigit (encoded[i]))
1168 || (i >= 1 && encoded[i] == '_' && isdigit (encoded[i - 1])))
1169 i -= 1;
1170 if (i > 1 && encoded[i] == '_' && encoded[i - 1] == '_')
1171 len0 = i - 1;
1172 else if (encoded[i] == '$')
1173 len0 = i;
1174 }
1175
1176 /* The first few characters that are not alphabetic are not part
1177 of any encoding we use, so we can copy them over verbatim. */
1178
1179 for (i = 0, j = 0; i < len0 && !isalpha (encoded[i]); i += 1, j += 1)
1180 decoded[j] = encoded[i];
1181
1182 at_start_name = 1;
1183 while (i < len0)
1184 {
1185 /* Is this a symbol function? */
1186 if (at_start_name && encoded[i] == 'O')
1187 {
1188 int k;
1189
1190 for (k = 0; ada_opname_table[k].encoded != NULL; k += 1)
1191 {
1192 int op_len = strlen (ada_opname_table[k].encoded);
1193 if ((strncmp (ada_opname_table[k].encoded + 1, encoded + i + 1,
1194 op_len - 1) == 0)
1195 && !isalnum (encoded[i + op_len]))
1196 {
1197 strcpy (&decoded.front() + j, ada_opname_table[k].decoded);
1198 at_start_name = 0;
1199 i += op_len;
1200 j += strlen (ada_opname_table[k].decoded);
1201 break;
1202 }
1203 }
1204 if (ada_opname_table[k].encoded != NULL)
1205 continue;
1206 }
1207 at_start_name = 0;
1208
1209 /* Replace "TK__" with "__", which will eventually be translated
1210 into "." (just below). */
1211
1212 if (i < len0 - 4 && startswith (encoded + i, "TK__"))
1213 i += 2;
1214
1215 /* Replace "__B_{DIGITS}+__" sequences by "__", which will eventually
1216 be translated into "." (just below). These are internal names
1217 generated for anonymous blocks inside which our symbol is nested. */
1218
1219 if (len0 - i > 5 && encoded [i] == '_' && encoded [i+1] == '_'
1220 && encoded [i+2] == 'B' && encoded [i+3] == '_'
1221 && isdigit (encoded [i+4]))
1222 {
1223 int k = i + 5;
1224
1225 while (k < len0 && isdigit (encoded[k]))
1226 k++; /* Skip any extra digit. */
1227
1228 /* Double-check that the "__B_{DIGITS}+" sequence we found
1229 is indeed followed by "__". */
1230 if (len0 - k > 2 && encoded [k] == '_' && encoded [k+1] == '_')
1231 i = k;
1232 }
1233
1234 /* Remove _E{DIGITS}+[sb] */
1235
1236 /* Just as for protected object subprograms, there are 2 categories
1237 of subprograms created by the compiler for each entry. The first
1238 one implements the actual entry code, and has a suffix following
1239 the convention above; the second one implements the barrier and
1240 uses the same convention as above, except that the 'E' is replaced
1241 by a 'B'.
1242
1243 Just as above, we do not decode the name of barrier functions
1244 to give the user a clue that the code he is debugging has been
1245 internally generated. */
1246
1247 if (len0 - i > 3 && encoded [i] == '_' && encoded[i+1] == 'E'
1248 && isdigit (encoded[i+2]))
1249 {
1250 int k = i + 3;
1251
1252 while (k < len0 && isdigit (encoded[k]))
1253 k++;
1254
1255 if (k < len0
1256 && (encoded[k] == 'b' || encoded[k] == 's'))
1257 {
1258 k++;
1259 /* Just as an extra precaution, make sure that if this
1260 suffix is followed by anything else, it is a '_'.
1261 Otherwise, we matched this sequence by accident. */
1262 if (k == len0
1263 || (k < len0 && encoded[k] == '_'))
1264 i = k;
1265 }
1266 }
1267
1268 /* Remove trailing "N" in [a-z0-9]+N__. The N is added by
1269 the GNAT front-end in protected object subprograms. */
1270
1271 if (i < len0 + 3
1272 && encoded[i] == 'N' && encoded[i+1] == '_' && encoded[i+2] == '_')
1273 {
1274 /* Backtrack a bit up until we reach either the begining of
1275 the encoded name, or "__". Make sure that we only find
1276 digits or lowercase characters. */
1277 const char *ptr = encoded + i - 1;
1278
1279 while (ptr >= encoded && is_lower_alphanum (ptr[0]))
1280 ptr--;
1281 if (ptr < encoded
1282 || (ptr > encoded && ptr[0] == '_' && ptr[-1] == '_'))
1283 i++;
1284 }
1285
1286 if (encoded[i] == 'X' && i != 0 && isalnum (encoded[i - 1]))
1287 {
1288 /* This is a X[bn]* sequence not separated from the previous
1289 part of the name with a non-alpha-numeric character (in other
1290 words, immediately following an alpha-numeric character), then
1291 verify that it is placed at the end of the encoded name. If
1292 not, then the encoding is not valid and we should abort the
1293 decoding. Otherwise, just skip it, it is used in body-nested
1294 package names. */
1295 do
1296 i += 1;
1297 while (i < len0 && (encoded[i] == 'b' || encoded[i] == 'n'));
1298 if (i < len0)
1299 goto Suppress;
1300 }
1301 else if (i < len0 - 2 && encoded[i] == '_' && encoded[i + 1] == '_')
1302 {
1303 /* Replace '__' by '.'. */
1304 decoded[j] = '.';
1305 at_start_name = 1;
1306 i += 2;
1307 j += 1;
1308 }
1309 else
1310 {
1311 /* It's a character part of the decoded name, so just copy it
1312 over. */
1313 decoded[j] = encoded[i];
1314 i += 1;
1315 j += 1;
1316 }
1317 }
1318 decoded.resize (j);
1319
1320 /* Decoded names should never contain any uppercase character.
1321 Double-check this, and abort the decoding if we find one. */
1322
1323 for (i = 0; i < decoded.length(); ++i)
1324 if (isupper (decoded[i]) || decoded[i] == ' ')
1325 goto Suppress;
1326
1327 return decoded;
1328
1329 Suppress:
1330 if (encoded[0] == '<')
1331 decoded = encoded;
1332 else
1333 decoded = '<' + std::string(encoded) + '>';
1334 return decoded;
1335
1336 }
1337
1338 /* Table for keeping permanent unique copies of decoded names. Once
1339 allocated, names in this table are never released. While this is a
1340 storage leak, it should not be significant unless there are massive
1341 changes in the set of decoded names in successive versions of a
1342 symbol table loaded during a single session. */
1343 static struct htab *decoded_names_store;
1344
1345 /* Returns the decoded name of GSYMBOL, as for ada_decode, caching it
1346 in the language-specific part of GSYMBOL, if it has not been
1347 previously computed. Tries to save the decoded name in the same
1348 obstack as GSYMBOL, if possible, and otherwise on the heap (so that,
1349 in any case, the decoded symbol has a lifetime at least that of
1350 GSYMBOL).
1351 The GSYMBOL parameter is "mutable" in the C++ sense: logically
1352 const, but nevertheless modified to a semantically equivalent form
1353 when a decoded name is cached in it. */
1354
1355 const char *
1356 ada_decode_symbol (const struct general_symbol_info *arg)
1357 {
1358 struct general_symbol_info *gsymbol = (struct general_symbol_info *) arg;
1359 const char **resultp =
1360 &gsymbol->language_specific.demangled_name;
1361
1362 if (!gsymbol->ada_mangled)
1363 {
1364 std::string decoded = ada_decode (gsymbol->linkage_name ());
1365 struct obstack *obstack = gsymbol->language_specific.obstack;
1366
1367 gsymbol->ada_mangled = 1;
1368
1369 if (obstack != NULL)
1370 *resultp = obstack_strdup (obstack, decoded.c_str ());
1371 else
1372 {
1373 /* Sometimes, we can't find a corresponding objfile, in
1374 which case, we put the result on the heap. Since we only
1375 decode when needed, we hope this usually does not cause a
1376 significant memory leak (FIXME). */
1377
1378 char **slot = (char **) htab_find_slot (decoded_names_store,
1379 decoded.c_str (), INSERT);
1380
1381 if (*slot == NULL)
1382 *slot = xstrdup (decoded.c_str ());
1383 *resultp = *slot;
1384 }
1385 }
1386
1387 return *resultp;
1388 }
1389
1390 static char *
1391 ada_la_decode (const char *encoded, int options)
1392 {
1393 return xstrdup (ada_decode (encoded).c_str ());
1394 }
1395
1396 /* Implement la_sniff_from_mangled_name for Ada. */
1397
1398 static int
1399 ada_sniff_from_mangled_name (const char *mangled, char **out)
1400 {
1401 std::string demangled = ada_decode (mangled);
1402
1403 *out = NULL;
1404
1405 if (demangled != mangled && demangled[0] != '<')
1406 {
1407 /* Set the gsymbol language to Ada, but still return 0.
1408 Two reasons for that:
1409
1410 1. For Ada, we prefer computing the symbol's decoded name
1411 on the fly rather than pre-compute it, in order to save
1412 memory (Ada projects are typically very large).
1413
1414 2. There are some areas in the definition of the GNAT
1415 encoding where, with a bit of bad luck, we might be able
1416 to decode a non-Ada symbol, generating an incorrect
1417 demangled name (Eg: names ending with "TB" for instance
1418 are identified as task bodies and so stripped from
1419 the decoded name returned).
1420
1421 Returning 1, here, but not setting *DEMANGLED, helps us get a
1422 little bit of the best of both worlds. Because we're last,
1423 we should not affect any of the other languages that were
1424 able to demangle the symbol before us; we get to correctly
1425 tag Ada symbols as such; and even if we incorrectly tagged a
1426 non-Ada symbol, which should be rare, any routing through the
1427 Ada language should be transparent (Ada tries to behave much
1428 like C/C++ with non-Ada symbols). */
1429 return 1;
1430 }
1431
1432 return 0;
1433 }
1434
1435 \f
1436
1437 /* Arrays */
1438
1439 /* Assuming that INDEX_DESC_TYPE is an ___XA structure, a structure
1440 generated by the GNAT compiler to describe the index type used
1441 for each dimension of an array, check whether it follows the latest
1442 known encoding. If not, fix it up to conform to the latest encoding.
1443 Otherwise, do nothing. This function also does nothing if
1444 INDEX_DESC_TYPE is NULL.
1445
1446 The GNAT encoding used to describe the array index type evolved a bit.
1447 Initially, the information would be provided through the name of each
1448 field of the structure type only, while the type of these fields was
1449 described as unspecified and irrelevant. The debugger was then expected
1450 to perform a global type lookup using the name of that field in order
1451 to get access to the full index type description. Because these global
1452 lookups can be very expensive, the encoding was later enhanced to make
1453 the global lookup unnecessary by defining the field type as being
1454 the full index type description.
1455
1456 The purpose of this routine is to allow us to support older versions
1457 of the compiler by detecting the use of the older encoding, and by
1458 fixing up the INDEX_DESC_TYPE to follow the new one (at this point,
1459 we essentially replace each field's meaningless type by the associated
1460 index subtype). */
1461
1462 void
1463 ada_fixup_array_indexes_type (struct type *index_desc_type)
1464 {
1465 int i;
1466
1467 if (index_desc_type == NULL)
1468 return;
1469 gdb_assert (index_desc_type->num_fields () > 0);
1470
1471 /* Check if INDEX_DESC_TYPE follows the older encoding (it is sufficient
1472 to check one field only, no need to check them all). If not, return
1473 now.
1474
1475 If our INDEX_DESC_TYPE was generated using the older encoding,
1476 the field type should be a meaningless integer type whose name
1477 is not equal to the field name. */
1478 if (TYPE_FIELD_TYPE (index_desc_type, 0)->name () != NULL
1479 && strcmp (TYPE_FIELD_TYPE (index_desc_type, 0)->name (),
1480 TYPE_FIELD_NAME (index_desc_type, 0)) == 0)
1481 return;
1482
1483 /* Fixup each field of INDEX_DESC_TYPE. */
1484 for (i = 0; i < index_desc_type->num_fields (); i++)
1485 {
1486 const char *name = TYPE_FIELD_NAME (index_desc_type, i);
1487 struct type *raw_type = ada_check_typedef (ada_find_any_type (name));
1488
1489 if (raw_type)
1490 TYPE_FIELD_TYPE (index_desc_type, i) = raw_type;
1491 }
1492 }
1493
1494 /* The desc_* routines return primitive portions of array descriptors
1495 (fat pointers). */
1496
1497 /* The descriptor or array type, if any, indicated by TYPE; removes
1498 level of indirection, if needed. */
1499
1500 static struct type *
1501 desc_base_type (struct type *type)
1502 {
1503 if (type == NULL)
1504 return NULL;
1505 type = ada_check_typedef (type);
1506 if (type->code () == TYPE_CODE_TYPEDEF)
1507 type = ada_typedef_target_type (type);
1508
1509 if (type != NULL
1510 && (type->code () == TYPE_CODE_PTR
1511 || type->code () == TYPE_CODE_REF))
1512 return ada_check_typedef (TYPE_TARGET_TYPE (type));
1513 else
1514 return type;
1515 }
1516
1517 /* True iff TYPE indicates a "thin" array pointer type. */
1518
1519 static int
1520 is_thin_pntr (struct type *type)
1521 {
1522 return
1523 is_suffix (ada_type_name (desc_base_type (type)), "___XUT")
1524 || is_suffix (ada_type_name (desc_base_type (type)), "___XUT___XVE");
1525 }
1526
1527 /* The descriptor type for thin pointer type TYPE. */
1528
1529 static struct type *
1530 thin_descriptor_type (struct type *type)
1531 {
1532 struct type *base_type = desc_base_type (type);
1533
1534 if (base_type == NULL)
1535 return NULL;
1536 if (is_suffix (ada_type_name (base_type), "___XVE"))
1537 return base_type;
1538 else
1539 {
1540 struct type *alt_type = ada_find_parallel_type (base_type, "___XVE");
1541
1542 if (alt_type == NULL)
1543 return base_type;
1544 else
1545 return alt_type;
1546 }
1547 }
1548
1549 /* A pointer to the array data for thin-pointer value VAL. */
1550
1551 static struct value *
1552 thin_data_pntr (struct value *val)
1553 {
1554 struct type *type = ada_check_typedef (value_type (val));
1555 struct type *data_type = desc_data_target_type (thin_descriptor_type (type));
1556
1557 data_type = lookup_pointer_type (data_type);
1558
1559 if (type->code () == TYPE_CODE_PTR)
1560 return value_cast (data_type, value_copy (val));
1561 else
1562 return value_from_longest (data_type, value_address (val));
1563 }
1564
1565 /* True iff TYPE indicates a "thick" array pointer type. */
1566
1567 static int
1568 is_thick_pntr (struct type *type)
1569 {
1570 type = desc_base_type (type);
1571 return (type != NULL && type->code () == TYPE_CODE_STRUCT
1572 && lookup_struct_elt_type (type, "P_BOUNDS", 1) != NULL);
1573 }
1574
1575 /* If TYPE is the type of an array descriptor (fat or thin pointer) or a
1576 pointer to one, the type of its bounds data; otherwise, NULL. */
1577
1578 static struct type *
1579 desc_bounds_type (struct type *type)
1580 {
1581 struct type *r;
1582
1583 type = desc_base_type (type);
1584
1585 if (type == NULL)
1586 return NULL;
1587 else if (is_thin_pntr (type))
1588 {
1589 type = thin_descriptor_type (type);
1590 if (type == NULL)
1591 return NULL;
1592 r = lookup_struct_elt_type (type, "BOUNDS", 1);
1593 if (r != NULL)
1594 return ada_check_typedef (r);
1595 }
1596 else if (type->code () == TYPE_CODE_STRUCT)
1597 {
1598 r = lookup_struct_elt_type (type, "P_BOUNDS", 1);
1599 if (r != NULL)
1600 return ada_check_typedef (TYPE_TARGET_TYPE (ada_check_typedef (r)));
1601 }
1602 return NULL;
1603 }
1604
1605 /* If ARR is an array descriptor (fat or thin pointer), or pointer to
1606 one, a pointer to its bounds data. Otherwise NULL. */
1607
1608 static struct value *
1609 desc_bounds (struct value *arr)
1610 {
1611 struct type *type = ada_check_typedef (value_type (arr));
1612
1613 if (is_thin_pntr (type))
1614 {
1615 struct type *bounds_type =
1616 desc_bounds_type (thin_descriptor_type (type));
1617 LONGEST addr;
1618
1619 if (bounds_type == NULL)
1620 error (_("Bad GNAT array descriptor"));
1621
1622 /* NOTE: The following calculation is not really kosher, but
1623 since desc_type is an XVE-encoded type (and shouldn't be),
1624 the correct calculation is a real pain. FIXME (and fix GCC). */
1625 if (type->code () == TYPE_CODE_PTR)
1626 addr = value_as_long (arr);
1627 else
1628 addr = value_address (arr);
1629
1630 return
1631 value_from_longest (lookup_pointer_type (bounds_type),
1632 addr - TYPE_LENGTH (bounds_type));
1633 }
1634
1635 else if (is_thick_pntr (type))
1636 {
1637 struct value *p_bounds = value_struct_elt (&arr, NULL, "P_BOUNDS", NULL,
1638 _("Bad GNAT array descriptor"));
1639 struct type *p_bounds_type = value_type (p_bounds);
1640
1641 if (p_bounds_type
1642 && p_bounds_type->code () == TYPE_CODE_PTR)
1643 {
1644 struct type *target_type = TYPE_TARGET_TYPE (p_bounds_type);
1645
1646 if (TYPE_STUB (target_type))
1647 p_bounds = value_cast (lookup_pointer_type
1648 (ada_check_typedef (target_type)),
1649 p_bounds);
1650 }
1651 else
1652 error (_("Bad GNAT array descriptor"));
1653
1654 return p_bounds;
1655 }
1656 else
1657 return NULL;
1658 }
1659
1660 /* If TYPE is the type of an array-descriptor (fat pointer), the bit
1661 position of the field containing the address of the bounds data. */
1662
1663 static int
1664 fat_pntr_bounds_bitpos (struct type *type)
1665 {
1666 return TYPE_FIELD_BITPOS (desc_base_type (type), 1);
1667 }
1668
1669 /* If TYPE is the type of an array-descriptor (fat pointer), the bit
1670 size of the field containing the address of the bounds data. */
1671
1672 static int
1673 fat_pntr_bounds_bitsize (struct type *type)
1674 {
1675 type = desc_base_type (type);
1676
1677 if (TYPE_FIELD_BITSIZE (type, 1) > 0)
1678 return TYPE_FIELD_BITSIZE (type, 1);
1679 else
1680 return 8 * TYPE_LENGTH (ada_check_typedef (TYPE_FIELD_TYPE (type, 1)));
1681 }
1682
1683 /* If TYPE is the type of an array descriptor (fat or thin pointer) or a
1684 pointer to one, the type of its array data (a array-with-no-bounds type);
1685 otherwise, NULL. Use ada_type_of_array to get an array type with bounds
1686 data. */
1687
1688 static struct type *
1689 desc_data_target_type (struct type *type)
1690 {
1691 type = desc_base_type (type);
1692
1693 /* NOTE: The following is bogus; see comment in desc_bounds. */
1694 if (is_thin_pntr (type))
1695 return desc_base_type (TYPE_FIELD_TYPE (thin_descriptor_type (type), 1));
1696 else if (is_thick_pntr (type))
1697 {
1698 struct type *data_type = lookup_struct_elt_type (type, "P_ARRAY", 1);
1699
1700 if (data_type
1701 && ada_check_typedef (data_type)->code () == TYPE_CODE_PTR)
1702 return ada_check_typedef (TYPE_TARGET_TYPE (data_type));
1703 }
1704
1705 return NULL;
1706 }
1707
1708 /* If ARR is an array descriptor (fat or thin pointer), a pointer to
1709 its array data. */
1710
1711 static struct value *
1712 desc_data (struct value *arr)
1713 {
1714 struct type *type = value_type (arr);
1715
1716 if (is_thin_pntr (type))
1717 return thin_data_pntr (arr);
1718 else if (is_thick_pntr (type))
1719 return value_struct_elt (&arr, NULL, "P_ARRAY", NULL,
1720 _("Bad GNAT array descriptor"));
1721 else
1722 return NULL;
1723 }
1724
1725
1726 /* If TYPE is the type of an array-descriptor (fat pointer), the bit
1727 position of the field containing the address of the data. */
1728
1729 static int
1730 fat_pntr_data_bitpos (struct type *type)
1731 {
1732 return TYPE_FIELD_BITPOS (desc_base_type (type), 0);
1733 }
1734
1735 /* If TYPE is the type of an array-descriptor (fat pointer), the bit
1736 size of the field containing the address of the data. */
1737
1738 static int
1739 fat_pntr_data_bitsize (struct type *type)
1740 {
1741 type = desc_base_type (type);
1742
1743 if (TYPE_FIELD_BITSIZE (type, 0) > 0)
1744 return TYPE_FIELD_BITSIZE (type, 0);
1745 else
1746 return TARGET_CHAR_BIT * TYPE_LENGTH (TYPE_FIELD_TYPE (type, 0));
1747 }
1748
1749 /* If BOUNDS is an array-bounds structure (or pointer to one), return
1750 the Ith lower bound stored in it, if WHICH is 0, and the Ith upper
1751 bound, if WHICH is 1. The first bound is I=1. */
1752
1753 static struct value *
1754 desc_one_bound (struct value *bounds, int i, int which)
1755 {
1756 char bound_name[20];
1757 xsnprintf (bound_name, sizeof (bound_name), "%cB%d",
1758 which ? 'U' : 'L', i - 1);
1759 return value_struct_elt (&bounds, NULL, bound_name, NULL,
1760 _("Bad GNAT array descriptor bounds"));
1761 }
1762
1763 /* If BOUNDS is an array-bounds structure type, return the bit position
1764 of the Ith lower bound stored in it, if WHICH is 0, and the Ith upper
1765 bound, if WHICH is 1. The first bound is I=1. */
1766
1767 static int
1768 desc_bound_bitpos (struct type *type, int i, int which)
1769 {
1770 return TYPE_FIELD_BITPOS (desc_base_type (type), 2 * i + which - 2);
1771 }
1772
1773 /* If BOUNDS is an array-bounds structure type, return the bit field size
1774 of the Ith lower bound stored in it, if WHICH is 0, and the Ith upper
1775 bound, if WHICH is 1. The first bound is I=1. */
1776
1777 static int
1778 desc_bound_bitsize (struct type *type, int i, int which)
1779 {
1780 type = desc_base_type (type);
1781
1782 if (TYPE_FIELD_BITSIZE (type, 2 * i + which - 2) > 0)
1783 return TYPE_FIELD_BITSIZE (type, 2 * i + which - 2);
1784 else
1785 return 8 * TYPE_LENGTH (TYPE_FIELD_TYPE (type, 2 * i + which - 2));
1786 }
1787
1788 /* If TYPE is the type of an array-bounds structure, the type of its
1789 Ith bound (numbering from 1). Otherwise, NULL. */
1790
1791 static struct type *
1792 desc_index_type (struct type *type, int i)
1793 {
1794 type = desc_base_type (type);
1795
1796 if (type->code () == TYPE_CODE_STRUCT)
1797 {
1798 char bound_name[20];
1799 xsnprintf (bound_name, sizeof (bound_name), "LB%d", i - 1);
1800 return lookup_struct_elt_type (type, bound_name, 1);
1801 }
1802 else
1803 return NULL;
1804 }
1805
1806 /* The number of index positions in the array-bounds type TYPE.
1807 Return 0 if TYPE is NULL. */
1808
1809 static int
1810 desc_arity (struct type *type)
1811 {
1812 type = desc_base_type (type);
1813
1814 if (type != NULL)
1815 return type->num_fields () / 2;
1816 return 0;
1817 }
1818
1819 /* Non-zero iff TYPE is a simple array type (not a pointer to one) or
1820 an array descriptor type (representing an unconstrained array
1821 type). */
1822
1823 static int
1824 ada_is_direct_array_type (struct type *type)
1825 {
1826 if (type == NULL)
1827 return 0;
1828 type = ada_check_typedef (type);
1829 return (type->code () == TYPE_CODE_ARRAY
1830 || ada_is_array_descriptor_type (type));
1831 }
1832
1833 /* Non-zero iff TYPE represents any kind of array in Ada, or a pointer
1834 * to one. */
1835
1836 static int
1837 ada_is_array_type (struct type *type)
1838 {
1839 while (type != NULL
1840 && (type->code () == TYPE_CODE_PTR
1841 || type->code () == TYPE_CODE_REF))
1842 type = TYPE_TARGET_TYPE (type);
1843 return ada_is_direct_array_type (type);
1844 }
1845
1846 /* Non-zero iff TYPE is a simple array type or pointer to one. */
1847
1848 int
1849 ada_is_simple_array_type (struct type *type)
1850 {
1851 if (type == NULL)
1852 return 0;
1853 type = ada_check_typedef (type);
1854 return (type->code () == TYPE_CODE_ARRAY
1855 || (type->code () == TYPE_CODE_PTR
1856 && (ada_check_typedef (TYPE_TARGET_TYPE (type))->code ()
1857 == TYPE_CODE_ARRAY)));
1858 }
1859
1860 /* Non-zero iff TYPE belongs to a GNAT array descriptor. */
1861
1862 int
1863 ada_is_array_descriptor_type (struct type *type)
1864 {
1865 struct type *data_type = desc_data_target_type (type);
1866
1867 if (type == NULL)
1868 return 0;
1869 type = ada_check_typedef (type);
1870 return (data_type != NULL
1871 && data_type->code () == TYPE_CODE_ARRAY
1872 && desc_arity (desc_bounds_type (type)) > 0);
1873 }
1874
1875 /* Non-zero iff type is a partially mal-formed GNAT array
1876 descriptor. FIXME: This is to compensate for some problems with
1877 debugging output from GNAT. Re-examine periodically to see if it
1878 is still needed. */
1879
1880 int
1881 ada_is_bogus_array_descriptor (struct type *type)
1882 {
1883 return
1884 type != NULL
1885 && type->code () == TYPE_CODE_STRUCT
1886 && (lookup_struct_elt_type (type, "P_BOUNDS", 1) != NULL
1887 || lookup_struct_elt_type (type, "P_ARRAY", 1) != NULL)
1888 && !ada_is_array_descriptor_type (type);
1889 }
1890
1891
1892 /* If ARR has a record type in the form of a standard GNAT array descriptor,
1893 (fat pointer) returns the type of the array data described---specifically,
1894 a pointer-to-array type. If BOUNDS is non-zero, the bounds data are filled
1895 in from the descriptor; otherwise, they are left unspecified. If
1896 the ARR denotes a null array descriptor and BOUNDS is non-zero,
1897 returns NULL. The result is simply the type of ARR if ARR is not
1898 a descriptor. */
1899
1900 static struct type *
1901 ada_type_of_array (struct value *arr, int bounds)
1902 {
1903 if (ada_is_constrained_packed_array_type (value_type (arr)))
1904 return decode_constrained_packed_array_type (value_type (arr));
1905
1906 if (!ada_is_array_descriptor_type (value_type (arr)))
1907 return value_type (arr);
1908
1909 if (!bounds)
1910 {
1911 struct type *array_type =
1912 ada_check_typedef (desc_data_target_type (value_type (arr)));
1913
1914 if (ada_is_unconstrained_packed_array_type (value_type (arr)))
1915 TYPE_FIELD_BITSIZE (array_type, 0) =
1916 decode_packed_array_bitsize (value_type (arr));
1917
1918 return array_type;
1919 }
1920 else
1921 {
1922 struct type *elt_type;
1923 int arity;
1924 struct value *descriptor;
1925
1926 elt_type = ada_array_element_type (value_type (arr), -1);
1927 arity = ada_array_arity (value_type (arr));
1928
1929 if (elt_type == NULL || arity == 0)
1930 return ada_check_typedef (value_type (arr));
1931
1932 descriptor = desc_bounds (arr);
1933 if (value_as_long (descriptor) == 0)
1934 return NULL;
1935 while (arity > 0)
1936 {
1937 struct type *range_type = alloc_type_copy (value_type (arr));
1938 struct type *array_type = alloc_type_copy (value_type (arr));
1939 struct value *low = desc_one_bound (descriptor, arity, 0);
1940 struct value *high = desc_one_bound (descriptor, arity, 1);
1941
1942 arity -= 1;
1943 create_static_range_type (range_type, value_type (low),
1944 longest_to_int (value_as_long (low)),
1945 longest_to_int (value_as_long (high)));
1946 elt_type = create_array_type (array_type, elt_type, range_type);
1947
1948 if (ada_is_unconstrained_packed_array_type (value_type (arr)))
1949 {
1950 /* We need to store the element packed bitsize, as well as
1951 recompute the array size, because it was previously
1952 computed based on the unpacked element size. */
1953 LONGEST lo = value_as_long (low);
1954 LONGEST hi = value_as_long (high);
1955
1956 TYPE_FIELD_BITSIZE (elt_type, 0) =
1957 decode_packed_array_bitsize (value_type (arr));
1958 /* If the array has no element, then the size is already
1959 zero, and does not need to be recomputed. */
1960 if (lo < hi)
1961 {
1962 int array_bitsize =
1963 (hi - lo + 1) * TYPE_FIELD_BITSIZE (elt_type, 0);
1964
1965 TYPE_LENGTH (array_type) = (array_bitsize + 7) / 8;
1966 }
1967 }
1968 }
1969
1970 return lookup_pointer_type (elt_type);
1971 }
1972 }
1973
1974 /* If ARR does not represent an array, returns ARR unchanged.
1975 Otherwise, returns either a standard GDB array with bounds set
1976 appropriately or, if ARR is a non-null fat pointer, a pointer to a standard
1977 GDB array. Returns NULL if ARR is a null fat pointer. */
1978
1979 struct value *
1980 ada_coerce_to_simple_array_ptr (struct value *arr)
1981 {
1982 if (ada_is_array_descriptor_type (value_type (arr)))
1983 {
1984 struct type *arrType = ada_type_of_array (arr, 1);
1985
1986 if (arrType == NULL)
1987 return NULL;
1988 return value_cast (arrType, value_copy (desc_data (arr)));
1989 }
1990 else if (ada_is_constrained_packed_array_type (value_type (arr)))
1991 return decode_constrained_packed_array (arr);
1992 else
1993 return arr;
1994 }
1995
1996 /* If ARR does not represent an array, returns ARR unchanged.
1997 Otherwise, returns a standard GDB array describing ARR (which may
1998 be ARR itself if it already is in the proper form). */
1999
2000 struct value *
2001 ada_coerce_to_simple_array (struct value *arr)
2002 {
2003 if (ada_is_array_descriptor_type (value_type (arr)))
2004 {
2005 struct value *arrVal = ada_coerce_to_simple_array_ptr (arr);
2006
2007 if (arrVal == NULL)
2008 error (_("Bounds unavailable for null array pointer."));
2009 ada_ensure_varsize_limit (TYPE_TARGET_TYPE (value_type (arrVal)));
2010 return value_ind (arrVal);
2011 }
2012 else if (ada_is_constrained_packed_array_type (value_type (arr)))
2013 return decode_constrained_packed_array (arr);
2014 else
2015 return arr;
2016 }
2017
2018 /* If TYPE represents a GNAT array type, return it translated to an
2019 ordinary GDB array type (possibly with BITSIZE fields indicating
2020 packing). For other types, is the identity. */
2021
2022 struct type *
2023 ada_coerce_to_simple_array_type (struct type *type)
2024 {
2025 if (ada_is_constrained_packed_array_type (type))
2026 return decode_constrained_packed_array_type (type);
2027
2028 if (ada_is_array_descriptor_type (type))
2029 return ada_check_typedef (desc_data_target_type (type));
2030
2031 return type;
2032 }
2033
2034 /* Non-zero iff TYPE represents a standard GNAT packed-array type. */
2035
2036 static int
2037 ada_is_packed_array_type (struct type *type)
2038 {
2039 if (type == NULL)
2040 return 0;
2041 type = desc_base_type (type);
2042 type = ada_check_typedef (type);
2043 return
2044 ada_type_name (type) != NULL
2045 && strstr (ada_type_name (type), "___XP") != NULL;
2046 }
2047
2048 /* Non-zero iff TYPE represents a standard GNAT constrained
2049 packed-array type. */
2050
2051 int
2052 ada_is_constrained_packed_array_type (struct type *type)
2053 {
2054 return ada_is_packed_array_type (type)
2055 && !ada_is_array_descriptor_type (type);
2056 }
2057
2058 /* Non-zero iff TYPE represents an array descriptor for a
2059 unconstrained packed-array type. */
2060
2061 static int
2062 ada_is_unconstrained_packed_array_type (struct type *type)
2063 {
2064 return ada_is_packed_array_type (type)
2065 && ada_is_array_descriptor_type (type);
2066 }
2067
2068 /* Given that TYPE encodes a packed array type (constrained or unconstrained),
2069 return the size of its elements in bits. */
2070
2071 static long
2072 decode_packed_array_bitsize (struct type *type)
2073 {
2074 const char *raw_name;
2075 const char *tail;
2076 long bits;
2077
2078 /* Access to arrays implemented as fat pointers are encoded as a typedef
2079 of the fat pointer type. We need the name of the fat pointer type
2080 to do the decoding, so strip the typedef layer. */
2081 if (type->code () == TYPE_CODE_TYPEDEF)
2082 type = ada_typedef_target_type (type);
2083
2084 raw_name = ada_type_name (ada_check_typedef (type));
2085 if (!raw_name)
2086 raw_name = ada_type_name (desc_base_type (type));
2087
2088 if (!raw_name)
2089 return 0;
2090
2091 tail = strstr (raw_name, "___XP");
2092 gdb_assert (tail != NULL);
2093
2094 if (sscanf (tail + sizeof ("___XP") - 1, "%ld", &bits) != 1)
2095 {
2096 lim_warning
2097 (_("could not understand bit size information on packed array"));
2098 return 0;
2099 }
2100
2101 return bits;
2102 }
2103
2104 /* Given that TYPE is a standard GDB array type with all bounds filled
2105 in, and that the element size of its ultimate scalar constituents
2106 (that is, either its elements, or, if it is an array of arrays, its
2107 elements' elements, etc.) is *ELT_BITS, return an identical type,
2108 but with the bit sizes of its elements (and those of any
2109 constituent arrays) recorded in the BITSIZE components of its
2110 TYPE_FIELD_BITSIZE values, and with *ELT_BITS set to its total size
2111 in bits.
2112
2113 Note that, for arrays whose index type has an XA encoding where
2114 a bound references a record discriminant, getting that discriminant,
2115 and therefore the actual value of that bound, is not possible
2116 because none of the given parameters gives us access to the record.
2117 This function assumes that it is OK in the context where it is being
2118 used to return an array whose bounds are still dynamic and where
2119 the length is arbitrary. */
2120
2121 static struct type *
2122 constrained_packed_array_type (struct type *type, long *elt_bits)
2123 {
2124 struct type *new_elt_type;
2125 struct type *new_type;
2126 struct type *index_type_desc;
2127 struct type *index_type;
2128 LONGEST low_bound, high_bound;
2129
2130 type = ada_check_typedef (type);
2131 if (type->code () != TYPE_CODE_ARRAY)
2132 return type;
2133
2134 index_type_desc = ada_find_parallel_type (type, "___XA");
2135 if (index_type_desc)
2136 index_type = to_fixed_range_type (TYPE_FIELD_TYPE (index_type_desc, 0),
2137 NULL);
2138 else
2139 index_type = TYPE_INDEX_TYPE (type);
2140
2141 new_type = alloc_type_copy (type);
2142 new_elt_type =
2143 constrained_packed_array_type (ada_check_typedef (TYPE_TARGET_TYPE (type)),
2144 elt_bits);
2145 create_array_type (new_type, new_elt_type, index_type);
2146 TYPE_FIELD_BITSIZE (new_type, 0) = *elt_bits;
2147 new_type->set_name (ada_type_name (type));
2148
2149 if ((check_typedef (index_type)->code () == TYPE_CODE_RANGE
2150 && is_dynamic_type (check_typedef (index_type)))
2151 || get_discrete_bounds (index_type, &low_bound, &high_bound) < 0)
2152 low_bound = high_bound = 0;
2153 if (high_bound < low_bound)
2154 *elt_bits = TYPE_LENGTH (new_type) = 0;
2155 else
2156 {
2157 *elt_bits *= (high_bound - low_bound + 1);
2158 TYPE_LENGTH (new_type) =
2159 (*elt_bits + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT;
2160 }
2161
2162 TYPE_FIXED_INSTANCE (new_type) = 1;
2163 return new_type;
2164 }
2165
2166 /* The array type encoded by TYPE, where
2167 ada_is_constrained_packed_array_type (TYPE). */
2168
2169 static struct type *
2170 decode_constrained_packed_array_type (struct type *type)
2171 {
2172 const char *raw_name = ada_type_name (ada_check_typedef (type));
2173 char *name;
2174 const char *tail;
2175 struct type *shadow_type;
2176 long bits;
2177
2178 if (!raw_name)
2179 raw_name = ada_type_name (desc_base_type (type));
2180
2181 if (!raw_name)
2182 return NULL;
2183
2184 name = (char *) alloca (strlen (raw_name) + 1);
2185 tail = strstr (raw_name, "___XP");
2186 type = desc_base_type (type);
2187
2188 memcpy (name, raw_name, tail - raw_name);
2189 name[tail - raw_name] = '\000';
2190
2191 shadow_type = ada_find_parallel_type_with_name (type, name);
2192
2193 if (shadow_type == NULL)
2194 {
2195 lim_warning (_("could not find bounds information on packed array"));
2196 return NULL;
2197 }
2198 shadow_type = check_typedef (shadow_type);
2199
2200 if (shadow_type->code () != TYPE_CODE_ARRAY)
2201 {
2202 lim_warning (_("could not understand bounds "
2203 "information on packed array"));
2204 return NULL;
2205 }
2206
2207 bits = decode_packed_array_bitsize (type);
2208 return constrained_packed_array_type (shadow_type, &bits);
2209 }
2210
2211 /* Given that ARR is a struct value *indicating a GNAT constrained packed
2212 array, returns a simple array that denotes that array. Its type is a
2213 standard GDB array type except that the BITSIZEs of the array
2214 target types are set to the number of bits in each element, and the
2215 type length is set appropriately. */
2216
2217 static struct value *
2218 decode_constrained_packed_array (struct value *arr)
2219 {
2220 struct type *type;
2221
2222 /* If our value is a pointer, then dereference it. Likewise if
2223 the value is a reference. Make sure that this operation does not
2224 cause the target type to be fixed, as this would indirectly cause
2225 this array to be decoded. The rest of the routine assumes that
2226 the array hasn't been decoded yet, so we use the basic "coerce_ref"
2227 and "value_ind" routines to perform the dereferencing, as opposed
2228 to using "ada_coerce_ref" or "ada_value_ind". */
2229 arr = coerce_ref (arr);
2230 if (ada_check_typedef (value_type (arr))->code () == TYPE_CODE_PTR)
2231 arr = value_ind (arr);
2232
2233 type = decode_constrained_packed_array_type (value_type (arr));
2234 if (type == NULL)
2235 {
2236 error (_("can't unpack array"));
2237 return NULL;
2238 }
2239
2240 if (type_byte_order (value_type (arr)) == BFD_ENDIAN_BIG
2241 && ada_is_modular_type (value_type (arr)))
2242 {
2243 /* This is a (right-justified) modular type representing a packed
2244 array with no wrapper. In order to interpret the value through
2245 the (left-justified) packed array type we just built, we must
2246 first left-justify it. */
2247 int bit_size, bit_pos;
2248 ULONGEST mod;
2249
2250 mod = ada_modulus (value_type (arr)) - 1;
2251 bit_size = 0;
2252 while (mod > 0)
2253 {
2254 bit_size += 1;
2255 mod >>= 1;
2256 }
2257 bit_pos = HOST_CHAR_BIT * TYPE_LENGTH (value_type (arr)) - bit_size;
2258 arr = ada_value_primitive_packed_val (arr, NULL,
2259 bit_pos / HOST_CHAR_BIT,
2260 bit_pos % HOST_CHAR_BIT,
2261 bit_size,
2262 type);
2263 }
2264
2265 return coerce_unspec_val_to_type (arr, type);
2266 }
2267
2268
2269 /* The value of the element of packed array ARR at the ARITY indices
2270 given in IND. ARR must be a simple array. */
2271
2272 static struct value *
2273 value_subscript_packed (struct value *arr, int arity, struct value **ind)
2274 {
2275 int i;
2276 int bits, elt_off, bit_off;
2277 long elt_total_bit_offset;
2278 struct type *elt_type;
2279 struct value *v;
2280
2281 bits = 0;
2282 elt_total_bit_offset = 0;
2283 elt_type = ada_check_typedef (value_type (arr));
2284 for (i = 0; i < arity; i += 1)
2285 {
2286 if (elt_type->code () != TYPE_CODE_ARRAY
2287 || TYPE_FIELD_BITSIZE (elt_type, 0) == 0)
2288 error
2289 (_("attempt to do packed indexing of "
2290 "something other than a packed array"));
2291 else
2292 {
2293 struct type *range_type = TYPE_INDEX_TYPE (elt_type);
2294 LONGEST lowerbound, upperbound;
2295 LONGEST idx;
2296
2297 if (get_discrete_bounds (range_type, &lowerbound, &upperbound) < 0)
2298 {
2299 lim_warning (_("don't know bounds of array"));
2300 lowerbound = upperbound = 0;
2301 }
2302
2303 idx = pos_atr (ind[i]);
2304 if (idx < lowerbound || idx > upperbound)
2305 lim_warning (_("packed array index %ld out of bounds"),
2306 (long) idx);
2307 bits = TYPE_FIELD_BITSIZE (elt_type, 0);
2308 elt_total_bit_offset += (idx - lowerbound) * bits;
2309 elt_type = ada_check_typedef (TYPE_TARGET_TYPE (elt_type));
2310 }
2311 }
2312 elt_off = elt_total_bit_offset / HOST_CHAR_BIT;
2313 bit_off = elt_total_bit_offset % HOST_CHAR_BIT;
2314
2315 v = ada_value_primitive_packed_val (arr, NULL, elt_off, bit_off,
2316 bits, elt_type);
2317 return v;
2318 }
2319
2320 /* Non-zero iff TYPE includes negative integer values. */
2321
2322 static int
2323 has_negatives (struct type *type)
2324 {
2325 switch (type->code ())
2326 {
2327 default:
2328 return 0;
2329 case TYPE_CODE_INT:
2330 return !TYPE_UNSIGNED (type);
2331 case TYPE_CODE_RANGE:
2332 return TYPE_LOW_BOUND (type) - TYPE_RANGE_DATA (type)->bias < 0;
2333 }
2334 }
2335
2336 /* With SRC being a buffer containing BIT_SIZE bits of data at BIT_OFFSET,
2337 unpack that data into UNPACKED. UNPACKED_LEN is the size in bytes of
2338 the unpacked buffer.
2339
2340 The size of the unpacked buffer (UNPACKED_LEN) is expected to be large
2341 enough to contain at least BIT_OFFSET bits. If not, an error is raised.
2342
2343 IS_BIG_ENDIAN is nonzero if the data is stored in big endian mode,
2344 zero otherwise.
2345
2346 IS_SIGNED_TYPE is nonzero if the data corresponds to a signed type.
2347
2348 IS_SCALAR is nonzero if the data corresponds to a signed type. */
2349
2350 static void
2351 ada_unpack_from_contents (const gdb_byte *src, int bit_offset, int bit_size,
2352 gdb_byte *unpacked, int unpacked_len,
2353 int is_big_endian, int is_signed_type,
2354 int is_scalar)
2355 {
2356 int src_len = (bit_size + bit_offset + HOST_CHAR_BIT - 1) / 8;
2357 int src_idx; /* Index into the source area */
2358 int src_bytes_left; /* Number of source bytes left to process. */
2359 int srcBitsLeft; /* Number of source bits left to move */
2360 int unusedLS; /* Number of bits in next significant
2361 byte of source that are unused */
2362
2363 int unpacked_idx; /* Index into the unpacked buffer */
2364 int unpacked_bytes_left; /* Number of bytes left to set in unpacked. */
2365
2366 unsigned long accum; /* Staging area for bits being transferred */
2367 int accumSize; /* Number of meaningful bits in accum */
2368 unsigned char sign;
2369
2370 /* Transmit bytes from least to most significant; delta is the direction
2371 the indices move. */
2372 int delta = is_big_endian ? -1 : 1;
2373
2374 /* Make sure that unpacked is large enough to receive the BIT_SIZE
2375 bits from SRC. .*/
2376 if ((bit_size + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT > unpacked_len)
2377 error (_("Cannot unpack %d bits into buffer of %d bytes"),
2378 bit_size, unpacked_len);
2379
2380 srcBitsLeft = bit_size;
2381 src_bytes_left = src_len;
2382 unpacked_bytes_left = unpacked_len;
2383 sign = 0;
2384
2385 if (is_big_endian)
2386 {
2387 src_idx = src_len - 1;
2388 if (is_signed_type
2389 && ((src[0] << bit_offset) & (1 << (HOST_CHAR_BIT - 1))))
2390 sign = ~0;
2391
2392 unusedLS =
2393 (HOST_CHAR_BIT - (bit_size + bit_offset) % HOST_CHAR_BIT)
2394 % HOST_CHAR_BIT;
2395
2396 if (is_scalar)
2397 {
2398 accumSize = 0;
2399 unpacked_idx = unpacked_len - 1;
2400 }
2401 else
2402 {
2403 /* Non-scalar values must be aligned at a byte boundary... */
2404 accumSize =
2405 (HOST_CHAR_BIT - bit_size % HOST_CHAR_BIT) % HOST_CHAR_BIT;
2406 /* ... And are placed at the beginning (most-significant) bytes
2407 of the target. */
2408 unpacked_idx = (bit_size + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT - 1;
2409 unpacked_bytes_left = unpacked_idx + 1;
2410 }
2411 }
2412 else
2413 {
2414 int sign_bit_offset = (bit_size + bit_offset - 1) % 8;
2415
2416 src_idx = unpacked_idx = 0;
2417 unusedLS = bit_offset;
2418 accumSize = 0;
2419
2420 if (is_signed_type && (src[src_len - 1] & (1 << sign_bit_offset)))
2421 sign = ~0;
2422 }
2423
2424 accum = 0;
2425 while (src_bytes_left > 0)
2426 {
2427 /* Mask for removing bits of the next source byte that are not
2428 part of the value. */
2429 unsigned int unusedMSMask =
2430 (1 << (srcBitsLeft >= HOST_CHAR_BIT ? HOST_CHAR_BIT : srcBitsLeft)) -
2431 1;
2432 /* Sign-extend bits for this byte. */
2433 unsigned int signMask = sign & ~unusedMSMask;
2434
2435 accum |=
2436 (((src[src_idx] >> unusedLS) & unusedMSMask) | signMask) << accumSize;
2437 accumSize += HOST_CHAR_BIT - unusedLS;
2438 if (accumSize >= HOST_CHAR_BIT)
2439 {
2440 unpacked[unpacked_idx] = accum & ~(~0UL << HOST_CHAR_BIT);
2441 accumSize -= HOST_CHAR_BIT;
2442 accum >>= HOST_CHAR_BIT;
2443 unpacked_bytes_left -= 1;
2444 unpacked_idx += delta;
2445 }
2446 srcBitsLeft -= HOST_CHAR_BIT - unusedLS;
2447 unusedLS = 0;
2448 src_bytes_left -= 1;
2449 src_idx += delta;
2450 }
2451 while (unpacked_bytes_left > 0)
2452 {
2453 accum |= sign << accumSize;
2454 unpacked[unpacked_idx] = accum & ~(~0UL << HOST_CHAR_BIT);
2455 accumSize -= HOST_CHAR_BIT;
2456 if (accumSize < 0)
2457 accumSize = 0;
2458 accum >>= HOST_CHAR_BIT;
2459 unpacked_bytes_left -= 1;
2460 unpacked_idx += delta;
2461 }
2462 }
2463
2464 /* Create a new value of type TYPE from the contents of OBJ starting
2465 at byte OFFSET, and bit offset BIT_OFFSET within that byte,
2466 proceeding for BIT_SIZE bits. If OBJ is an lval in memory, then
2467 assigning through the result will set the field fetched from.
2468 VALADDR is ignored unless OBJ is NULL, in which case,
2469 VALADDR+OFFSET must address the start of storage containing the
2470 packed value. The value returned in this case is never an lval.
2471 Assumes 0 <= BIT_OFFSET < HOST_CHAR_BIT. */
2472
2473 struct value *
2474 ada_value_primitive_packed_val (struct value *obj, const gdb_byte *valaddr,
2475 long offset, int bit_offset, int bit_size,
2476 struct type *type)
2477 {
2478 struct value *v;
2479 const gdb_byte *src; /* First byte containing data to unpack */
2480 gdb_byte *unpacked;
2481 const int is_scalar = is_scalar_type (type);
2482 const int is_big_endian = type_byte_order (type) == BFD_ENDIAN_BIG;
2483 gdb::byte_vector staging;
2484
2485 type = ada_check_typedef (type);
2486
2487 if (obj == NULL)
2488 src = valaddr + offset;
2489 else
2490 src = value_contents (obj) + offset;
2491
2492 if (is_dynamic_type (type))
2493 {
2494 /* The length of TYPE might by dynamic, so we need to resolve
2495 TYPE in order to know its actual size, which we then use
2496 to create the contents buffer of the value we return.
2497 The difficulty is that the data containing our object is
2498 packed, and therefore maybe not at a byte boundary. So, what
2499 we do, is unpack the data into a byte-aligned buffer, and then
2500 use that buffer as our object's value for resolving the type. */
2501 int staging_len = (bit_size + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT;
2502 staging.resize (staging_len);
2503
2504 ada_unpack_from_contents (src, bit_offset, bit_size,
2505 staging.data (), staging.size (),
2506 is_big_endian, has_negatives (type),
2507 is_scalar);
2508 type = resolve_dynamic_type (type, staging, 0);
2509 if (TYPE_LENGTH (type) < (bit_size + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT)
2510 {
2511 /* This happens when the length of the object is dynamic,
2512 and is actually smaller than the space reserved for it.
2513 For instance, in an array of variant records, the bit_size
2514 we're given is the array stride, which is constant and
2515 normally equal to the maximum size of its element.
2516 But, in reality, each element only actually spans a portion
2517 of that stride. */
2518 bit_size = TYPE_LENGTH (type) * HOST_CHAR_BIT;
2519 }
2520 }
2521
2522 if (obj == NULL)
2523 {
2524 v = allocate_value (type);
2525 src = valaddr + offset;
2526 }
2527 else if (VALUE_LVAL (obj) == lval_memory && value_lazy (obj))
2528 {
2529 int src_len = (bit_size + bit_offset + HOST_CHAR_BIT - 1) / 8;
2530 gdb_byte *buf;
2531
2532 v = value_at (type, value_address (obj) + offset);
2533 buf = (gdb_byte *) alloca (src_len);
2534 read_memory (value_address (v), buf, src_len);
2535 src = buf;
2536 }
2537 else
2538 {
2539 v = allocate_value (type);
2540 src = value_contents (obj) + offset;
2541 }
2542
2543 if (obj != NULL)
2544 {
2545 long new_offset = offset;
2546
2547 set_value_component_location (v, obj);
2548 set_value_bitpos (v, bit_offset + value_bitpos (obj));
2549 set_value_bitsize (v, bit_size);
2550 if (value_bitpos (v) >= HOST_CHAR_BIT)
2551 {
2552 ++new_offset;
2553 set_value_bitpos (v, value_bitpos (v) - HOST_CHAR_BIT);
2554 }
2555 set_value_offset (v, new_offset);
2556
2557 /* Also set the parent value. This is needed when trying to
2558 assign a new value (in inferior memory). */
2559 set_value_parent (v, obj);
2560 }
2561 else
2562 set_value_bitsize (v, bit_size);
2563 unpacked = value_contents_writeable (v);
2564
2565 if (bit_size == 0)
2566 {
2567 memset (unpacked, 0, TYPE_LENGTH (type));
2568 return v;
2569 }
2570
2571 if (staging.size () == TYPE_LENGTH (type))
2572 {
2573 /* Small short-cut: If we've unpacked the data into a buffer
2574 of the same size as TYPE's length, then we can reuse that,
2575 instead of doing the unpacking again. */
2576 memcpy (unpacked, staging.data (), staging.size ());
2577 }
2578 else
2579 ada_unpack_from_contents (src, bit_offset, bit_size,
2580 unpacked, TYPE_LENGTH (type),
2581 is_big_endian, has_negatives (type), is_scalar);
2582
2583 return v;
2584 }
2585
2586 /* Store the contents of FROMVAL into the location of TOVAL.
2587 Return a new value with the location of TOVAL and contents of
2588 FROMVAL. Handles assignment into packed fields that have
2589 floating-point or non-scalar types. */
2590
2591 static struct value *
2592 ada_value_assign (struct value *toval, struct value *fromval)
2593 {
2594 struct type *type = value_type (toval);
2595 int bits = value_bitsize (toval);
2596
2597 toval = ada_coerce_ref (toval);
2598 fromval = ada_coerce_ref (fromval);
2599
2600 if (ada_is_direct_array_type (value_type (toval)))
2601 toval = ada_coerce_to_simple_array (toval);
2602 if (ada_is_direct_array_type (value_type (fromval)))
2603 fromval = ada_coerce_to_simple_array (fromval);
2604
2605 if (!deprecated_value_modifiable (toval))
2606 error (_("Left operand of assignment is not a modifiable lvalue."));
2607
2608 if (VALUE_LVAL (toval) == lval_memory
2609 && bits > 0
2610 && (type->code () == TYPE_CODE_FLT
2611 || type->code () == TYPE_CODE_STRUCT))
2612 {
2613 int len = (value_bitpos (toval)
2614 + bits + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT;
2615 int from_size;
2616 gdb_byte *buffer = (gdb_byte *) alloca (len);
2617 struct value *val;
2618 CORE_ADDR to_addr = value_address (toval);
2619
2620 if (type->code () == TYPE_CODE_FLT)
2621 fromval = value_cast (type, fromval);
2622
2623 read_memory (to_addr, buffer, len);
2624 from_size = value_bitsize (fromval);
2625 if (from_size == 0)
2626 from_size = TYPE_LENGTH (value_type (fromval)) * TARGET_CHAR_BIT;
2627
2628 const int is_big_endian = type_byte_order (type) == BFD_ENDIAN_BIG;
2629 ULONGEST from_offset = 0;
2630 if (is_big_endian && is_scalar_type (value_type (fromval)))
2631 from_offset = from_size - bits;
2632 copy_bitwise (buffer, value_bitpos (toval),
2633 value_contents (fromval), from_offset,
2634 bits, is_big_endian);
2635 write_memory_with_notification (to_addr, buffer, len);
2636
2637 val = value_copy (toval);
2638 memcpy (value_contents_raw (val), value_contents (fromval),
2639 TYPE_LENGTH (type));
2640 deprecated_set_value_type (val, type);
2641
2642 return val;
2643 }
2644
2645 return value_assign (toval, fromval);
2646 }
2647
2648
2649 /* Given that COMPONENT is a memory lvalue that is part of the lvalue
2650 CONTAINER, assign the contents of VAL to COMPONENTS's place in
2651 CONTAINER. Modifies the VALUE_CONTENTS of CONTAINER only, not
2652 COMPONENT, and not the inferior's memory. The current contents
2653 of COMPONENT are ignored.
2654
2655 Although not part of the initial design, this function also works
2656 when CONTAINER and COMPONENT are not_lval's: it works as if CONTAINER
2657 had a null address, and COMPONENT had an address which is equal to
2658 its offset inside CONTAINER. */
2659
2660 static void
2661 value_assign_to_component (struct value *container, struct value *component,
2662 struct value *val)
2663 {
2664 LONGEST offset_in_container =
2665 (LONGEST) (value_address (component) - value_address (container));
2666 int bit_offset_in_container =
2667 value_bitpos (component) - value_bitpos (container);
2668 int bits;
2669
2670 val = value_cast (value_type (component), val);
2671
2672 if (value_bitsize (component) == 0)
2673 bits = TARGET_CHAR_BIT * TYPE_LENGTH (value_type (component));
2674 else
2675 bits = value_bitsize (component);
2676
2677 if (type_byte_order (value_type (container)) == BFD_ENDIAN_BIG)
2678 {
2679 int src_offset;
2680
2681 if (is_scalar_type (check_typedef (value_type (component))))
2682 src_offset
2683 = TYPE_LENGTH (value_type (component)) * TARGET_CHAR_BIT - bits;
2684 else
2685 src_offset = 0;
2686 copy_bitwise (value_contents_writeable (container) + offset_in_container,
2687 value_bitpos (container) + bit_offset_in_container,
2688 value_contents (val), src_offset, bits, 1);
2689 }
2690 else
2691 copy_bitwise (value_contents_writeable (container) + offset_in_container,
2692 value_bitpos (container) + bit_offset_in_container,
2693 value_contents (val), 0, bits, 0);
2694 }
2695
2696 /* Determine if TYPE is an access to an unconstrained array. */
2697
2698 bool
2699 ada_is_access_to_unconstrained_array (struct type *type)
2700 {
2701 return (type->code () == TYPE_CODE_TYPEDEF
2702 && is_thick_pntr (ada_typedef_target_type (type)));
2703 }
2704
2705 /* The value of the element of array ARR at the ARITY indices given in IND.
2706 ARR may be either a simple array, GNAT array descriptor, or pointer
2707 thereto. */
2708
2709 struct value *
2710 ada_value_subscript (struct value *arr, int arity, struct value **ind)
2711 {
2712 int k;
2713 struct value *elt;
2714 struct type *elt_type;
2715
2716 elt = ada_coerce_to_simple_array (arr);
2717
2718 elt_type = ada_check_typedef (value_type (elt));
2719 if (elt_type->code () == TYPE_CODE_ARRAY
2720 && TYPE_FIELD_BITSIZE (elt_type, 0) > 0)
2721 return value_subscript_packed (elt, arity, ind);
2722
2723 for (k = 0; k < arity; k += 1)
2724 {
2725 struct type *saved_elt_type = TYPE_TARGET_TYPE (elt_type);
2726
2727 if (elt_type->code () != TYPE_CODE_ARRAY)
2728 error (_("too many subscripts (%d expected)"), k);
2729
2730 elt = value_subscript (elt, pos_atr (ind[k]));
2731
2732 if (ada_is_access_to_unconstrained_array (saved_elt_type)
2733 && value_type (elt)->code () != TYPE_CODE_TYPEDEF)
2734 {
2735 /* The element is a typedef to an unconstrained array,
2736 except that the value_subscript call stripped the
2737 typedef layer. The typedef layer is GNAT's way to
2738 specify that the element is, at the source level, an
2739 access to the unconstrained array, rather than the
2740 unconstrained array. So, we need to restore that
2741 typedef layer, which we can do by forcing the element's
2742 type back to its original type. Otherwise, the returned
2743 value is going to be printed as the array, rather
2744 than as an access. Another symptom of the same issue
2745 would be that an expression trying to dereference the
2746 element would also be improperly rejected. */
2747 deprecated_set_value_type (elt, saved_elt_type);
2748 }
2749
2750 elt_type = ada_check_typedef (value_type (elt));
2751 }
2752
2753 return elt;
2754 }
2755
2756 /* Assuming ARR is a pointer to a GDB array, the value of the element
2757 of *ARR at the ARITY indices given in IND.
2758 Does not read the entire array into memory.
2759
2760 Note: Unlike what one would expect, this function is used instead of
2761 ada_value_subscript for basically all non-packed array types. The reason
2762 for this is that a side effect of doing our own pointer arithmetics instead
2763 of relying on value_subscript is that there is no implicit typedef peeling.
2764 This is important for arrays of array accesses, where it allows us to
2765 preserve the fact that the array's element is an array access, where the
2766 access part os encoded in a typedef layer. */
2767
2768 static struct value *
2769 ada_value_ptr_subscript (struct value *arr, int arity, struct value **ind)
2770 {
2771 int k;
2772 struct value *array_ind = ada_value_ind (arr);
2773 struct type *type
2774 = check_typedef (value_enclosing_type (array_ind));
2775
2776 if (type->code () == TYPE_CODE_ARRAY
2777 && TYPE_FIELD_BITSIZE (type, 0) > 0)
2778 return value_subscript_packed (array_ind, arity, ind);
2779
2780 for (k = 0; k < arity; k += 1)
2781 {
2782 LONGEST lwb, upb;
2783
2784 if (type->code () != TYPE_CODE_ARRAY)
2785 error (_("too many subscripts (%d expected)"), k);
2786 arr = value_cast (lookup_pointer_type (TYPE_TARGET_TYPE (type)),
2787 value_copy (arr));
2788 get_discrete_bounds (TYPE_INDEX_TYPE (type), &lwb, &upb);
2789 arr = value_ptradd (arr, pos_atr (ind[k]) - lwb);
2790 type = TYPE_TARGET_TYPE (type);
2791 }
2792
2793 return value_ind (arr);
2794 }
2795
2796 /* Given that ARRAY_PTR is a pointer or reference to an array of type TYPE (the
2797 actual type of ARRAY_PTR is ignored), returns the Ada slice of
2798 HIGH'Pos-LOW'Pos+1 elements starting at index LOW. The lower bound of
2799 this array is LOW, as per Ada rules. */
2800 static struct value *
2801 ada_value_slice_from_ptr (struct value *array_ptr, struct type *type,
2802 int low, int high)
2803 {
2804 struct type *type0 = ada_check_typedef (type);
2805 struct type *base_index_type = TYPE_TARGET_TYPE (TYPE_INDEX_TYPE (type0));
2806 struct type *index_type
2807 = create_static_range_type (NULL, base_index_type, low, high);
2808 struct type *slice_type = create_array_type_with_stride
2809 (NULL, TYPE_TARGET_TYPE (type0), index_type,
2810 type0->dyn_prop (DYN_PROP_BYTE_STRIDE),
2811 TYPE_FIELD_BITSIZE (type0, 0));
2812 int base_low = ada_discrete_type_low_bound (TYPE_INDEX_TYPE (type0));
2813 LONGEST base_low_pos, low_pos;
2814 CORE_ADDR base;
2815
2816 if (!discrete_position (base_index_type, low, &low_pos)
2817 || !discrete_position (base_index_type, base_low, &base_low_pos))
2818 {
2819 warning (_("unable to get positions in slice, use bounds instead"));
2820 low_pos = low;
2821 base_low_pos = base_low;
2822 }
2823
2824 base = value_as_address (array_ptr)
2825 + ((low_pos - base_low_pos)
2826 * TYPE_LENGTH (TYPE_TARGET_TYPE (type0)));
2827 return value_at_lazy (slice_type, base);
2828 }
2829
2830
2831 static struct value *
2832 ada_value_slice (struct value *array, int low, int high)
2833 {
2834 struct type *type = ada_check_typedef (value_type (array));
2835 struct type *base_index_type = TYPE_TARGET_TYPE (TYPE_INDEX_TYPE (type));
2836 struct type *index_type
2837 = create_static_range_type (NULL, TYPE_INDEX_TYPE (type), low, high);
2838 struct type *slice_type = create_array_type_with_stride
2839 (NULL, TYPE_TARGET_TYPE (type), index_type,
2840 type->dyn_prop (DYN_PROP_BYTE_STRIDE),
2841 TYPE_FIELD_BITSIZE (type, 0));
2842 LONGEST low_pos, high_pos;
2843
2844 if (!discrete_position (base_index_type, low, &low_pos)
2845 || !discrete_position (base_index_type, high, &high_pos))
2846 {
2847 warning (_("unable to get positions in slice, use bounds instead"));
2848 low_pos = low;
2849 high_pos = high;
2850 }
2851
2852 return value_cast (slice_type,
2853 value_slice (array, low, high_pos - low_pos + 1));
2854 }
2855
2856 /* If type is a record type in the form of a standard GNAT array
2857 descriptor, returns the number of dimensions for type. If arr is a
2858 simple array, returns the number of "array of"s that prefix its
2859 type designation. Otherwise, returns 0. */
2860
2861 int
2862 ada_array_arity (struct type *type)
2863 {
2864 int arity;
2865
2866 if (type == NULL)
2867 return 0;
2868
2869 type = desc_base_type (type);
2870
2871 arity = 0;
2872 if (type->code () == TYPE_CODE_STRUCT)
2873 return desc_arity (desc_bounds_type (type));
2874 else
2875 while (type->code () == TYPE_CODE_ARRAY)
2876 {
2877 arity += 1;
2878 type = ada_check_typedef (TYPE_TARGET_TYPE (type));
2879 }
2880
2881 return arity;
2882 }
2883
2884 /* If TYPE is a record type in the form of a standard GNAT array
2885 descriptor or a simple array type, returns the element type for
2886 TYPE after indexing by NINDICES indices, or by all indices if
2887 NINDICES is -1. Otherwise, returns NULL. */
2888
2889 struct type *
2890 ada_array_element_type (struct type *type, int nindices)
2891 {
2892 type = desc_base_type (type);
2893
2894 if (type->code () == TYPE_CODE_STRUCT)
2895 {
2896 int k;
2897 struct type *p_array_type;
2898
2899 p_array_type = desc_data_target_type (type);
2900
2901 k = ada_array_arity (type);
2902 if (k == 0)
2903 return NULL;
2904
2905 /* Initially p_array_type = elt_type(*)[]...(k times)...[]. */
2906 if (nindices >= 0 && k > nindices)
2907 k = nindices;
2908 while (k > 0 && p_array_type != NULL)
2909 {
2910 p_array_type = ada_check_typedef (TYPE_TARGET_TYPE (p_array_type));
2911 k -= 1;
2912 }
2913 return p_array_type;
2914 }
2915 else if (type->code () == TYPE_CODE_ARRAY)
2916 {
2917 while (nindices != 0 && type->code () == TYPE_CODE_ARRAY)
2918 {
2919 type = TYPE_TARGET_TYPE (type);
2920 nindices -= 1;
2921 }
2922 return type;
2923 }
2924
2925 return NULL;
2926 }
2927
2928 /* The type of nth index in arrays of given type (n numbering from 1).
2929 Does not examine memory. Throws an error if N is invalid or TYPE
2930 is not an array type. NAME is the name of the Ada attribute being
2931 evaluated ('range, 'first, 'last, or 'length); it is used in building
2932 the error message. */
2933
2934 static struct type *
2935 ada_index_type (struct type *type, int n, const char *name)
2936 {
2937 struct type *result_type;
2938
2939 type = desc_base_type (type);
2940
2941 if (n < 0 || n > ada_array_arity (type))
2942 error (_("invalid dimension number to '%s"), name);
2943
2944 if (ada_is_simple_array_type (type))
2945 {
2946 int i;
2947
2948 for (i = 1; i < n; i += 1)
2949 type = TYPE_TARGET_TYPE (type);
2950 result_type = TYPE_TARGET_TYPE (TYPE_INDEX_TYPE (type));
2951 /* FIXME: The stabs type r(0,0);bound;bound in an array type
2952 has a target type of TYPE_CODE_UNDEF. We compensate here, but
2953 perhaps stabsread.c would make more sense. */
2954 if (result_type && result_type->code () == TYPE_CODE_UNDEF)
2955 result_type = NULL;
2956 }
2957 else
2958 {
2959 result_type = desc_index_type (desc_bounds_type (type), n);
2960 if (result_type == NULL)
2961 error (_("attempt to take bound of something that is not an array"));
2962 }
2963
2964 return result_type;
2965 }
2966
2967 /* Given that arr is an array type, returns the lower bound of the
2968 Nth index (numbering from 1) if WHICH is 0, and the upper bound if
2969 WHICH is 1. This returns bounds 0 .. -1 if ARR_TYPE is an
2970 array-descriptor type. It works for other arrays with bounds supplied
2971 by run-time quantities other than discriminants. */
2972
2973 static LONGEST
2974 ada_array_bound_from_type (struct type *arr_type, int n, int which)
2975 {
2976 struct type *type, *index_type_desc, *index_type;
2977 int i;
2978
2979 gdb_assert (which == 0 || which == 1);
2980
2981 if (ada_is_constrained_packed_array_type (arr_type))
2982 arr_type = decode_constrained_packed_array_type (arr_type);
2983
2984 if (arr_type == NULL || !ada_is_simple_array_type (arr_type))
2985 return (LONGEST) - which;
2986
2987 if (arr_type->code () == TYPE_CODE_PTR)
2988 type = TYPE_TARGET_TYPE (arr_type);
2989 else
2990 type = arr_type;
2991
2992 if (TYPE_FIXED_INSTANCE (type))
2993 {
2994 /* The array has already been fixed, so we do not need to
2995 check the parallel ___XA type again. That encoding has
2996 already been applied, so ignore it now. */
2997 index_type_desc = NULL;
2998 }
2999 else
3000 {
3001 index_type_desc = ada_find_parallel_type (type, "___XA");
3002 ada_fixup_array_indexes_type (index_type_desc);
3003 }
3004
3005 if (index_type_desc != NULL)
3006 index_type = to_fixed_range_type (TYPE_FIELD_TYPE (index_type_desc, n - 1),
3007 NULL);
3008 else
3009 {
3010 struct type *elt_type = check_typedef (type);
3011
3012 for (i = 1; i < n; i++)
3013 elt_type = check_typedef (TYPE_TARGET_TYPE (elt_type));
3014
3015 index_type = TYPE_INDEX_TYPE (elt_type);
3016 }
3017
3018 return
3019 (LONGEST) (which == 0
3020 ? ada_discrete_type_low_bound (index_type)
3021 : ada_discrete_type_high_bound (index_type));
3022 }
3023
3024 /* Given that arr is an array value, returns the lower bound of the
3025 nth index (numbering from 1) if WHICH is 0, and the upper bound if
3026 WHICH is 1. This routine will also work for arrays with bounds
3027 supplied by run-time quantities other than discriminants. */
3028
3029 static LONGEST
3030 ada_array_bound (struct value *arr, int n, int which)
3031 {
3032 struct type *arr_type;
3033
3034 if (check_typedef (value_type (arr))->code () == TYPE_CODE_PTR)
3035 arr = value_ind (arr);
3036 arr_type = value_enclosing_type (arr);
3037
3038 if (ada_is_constrained_packed_array_type (arr_type))
3039 return ada_array_bound (decode_constrained_packed_array (arr), n, which);
3040 else if (ada_is_simple_array_type (arr_type))
3041 return ada_array_bound_from_type (arr_type, n, which);
3042 else
3043 return value_as_long (desc_one_bound (desc_bounds (arr), n, which));
3044 }
3045
3046 /* Given that arr is an array value, returns the length of the
3047 nth index. This routine will also work for arrays with bounds
3048 supplied by run-time quantities other than discriminants.
3049 Does not work for arrays indexed by enumeration types with representation
3050 clauses at the moment. */
3051
3052 static LONGEST
3053 ada_array_length (struct value *arr, int n)
3054 {
3055 struct type *arr_type, *index_type;
3056 int low, high;
3057
3058 if (check_typedef (value_type (arr))->code () == TYPE_CODE_PTR)
3059 arr = value_ind (arr);
3060 arr_type = value_enclosing_type (arr);
3061
3062 if (ada_is_constrained_packed_array_type (arr_type))
3063 return ada_array_length (decode_constrained_packed_array (arr), n);
3064
3065 if (ada_is_simple_array_type (arr_type))
3066 {
3067 low = ada_array_bound_from_type (arr_type, n, 0);
3068 high = ada_array_bound_from_type (arr_type, n, 1);
3069 }
3070 else
3071 {
3072 low = value_as_long (desc_one_bound (desc_bounds (arr), n, 0));
3073 high = value_as_long (desc_one_bound (desc_bounds (arr), n, 1));
3074 }
3075
3076 arr_type = check_typedef (arr_type);
3077 index_type = ada_index_type (arr_type, n, "length");
3078 if (index_type != NULL)
3079 {
3080 struct type *base_type;
3081 if (index_type->code () == TYPE_CODE_RANGE)
3082 base_type = TYPE_TARGET_TYPE (index_type);
3083 else
3084 base_type = index_type;
3085
3086 low = pos_atr (value_from_longest (base_type, low));
3087 high = pos_atr (value_from_longest (base_type, high));
3088 }
3089 return high - low + 1;
3090 }
3091
3092 /* An array whose type is that of ARR_TYPE (an array type), with
3093 bounds LOW to HIGH, but whose contents are unimportant. If HIGH is
3094 less than LOW, then LOW-1 is used. */
3095
3096 static struct value *
3097 empty_array (struct type *arr_type, int low, int high)
3098 {
3099 struct type *arr_type0 = ada_check_typedef (arr_type);
3100 struct type *index_type
3101 = create_static_range_type
3102 (NULL, TYPE_TARGET_TYPE (TYPE_INDEX_TYPE (arr_type0)), low,
3103 high < low ? low - 1 : high);
3104 struct type *elt_type = ada_array_element_type (arr_type0, 1);
3105
3106 return allocate_value (create_array_type (NULL, elt_type, index_type));
3107 }
3108 \f
3109
3110 /* Name resolution */
3111
3112 /* The "decoded" name for the user-definable Ada operator corresponding
3113 to OP. */
3114
3115 static const char *
3116 ada_decoded_op_name (enum exp_opcode op)
3117 {
3118 int i;
3119
3120 for (i = 0; ada_opname_table[i].encoded != NULL; i += 1)
3121 {
3122 if (ada_opname_table[i].op == op)
3123 return ada_opname_table[i].decoded;
3124 }
3125 error (_("Could not find operator name for opcode"));
3126 }
3127
3128 /* Returns true (non-zero) iff decoded name N0 should appear before N1
3129 in a listing of choices during disambiguation (see sort_choices, below).
3130 The idea is that overloadings of a subprogram name from the
3131 same package should sort in their source order. We settle for ordering
3132 such symbols by their trailing number (__N or $N). */
3133
3134 static int
3135 encoded_ordered_before (const char *N0, const char *N1)
3136 {
3137 if (N1 == NULL)
3138 return 0;
3139 else if (N0 == NULL)
3140 return 1;
3141 else
3142 {
3143 int k0, k1;
3144
3145 for (k0 = strlen (N0) - 1; k0 > 0 && isdigit (N0[k0]); k0 -= 1)
3146 ;
3147 for (k1 = strlen (N1) - 1; k1 > 0 && isdigit (N1[k1]); k1 -= 1)
3148 ;
3149 if ((N0[k0] == '_' || N0[k0] == '$') && N0[k0 + 1] != '\000'
3150 && (N1[k1] == '_' || N1[k1] == '$') && N1[k1 + 1] != '\000')
3151 {
3152 int n0, n1;
3153
3154 n0 = k0;
3155 while (N0[n0] == '_' && n0 > 0 && N0[n0 - 1] == '_')
3156 n0 -= 1;
3157 n1 = k1;
3158 while (N1[n1] == '_' && n1 > 0 && N1[n1 - 1] == '_')
3159 n1 -= 1;
3160 if (n0 == n1 && strncmp (N0, N1, n0) == 0)
3161 return (atoi (N0 + k0 + 1) < atoi (N1 + k1 + 1));
3162 }
3163 return (strcmp (N0, N1) < 0);
3164 }
3165 }
3166
3167 /* Sort SYMS[0..NSYMS-1] to put the choices in a canonical order by the
3168 encoded names. */
3169
3170 static void
3171 sort_choices (struct block_symbol syms[], int nsyms)
3172 {
3173 int i;
3174
3175 for (i = 1; i < nsyms; i += 1)
3176 {
3177 struct block_symbol sym = syms[i];
3178 int j;
3179
3180 for (j = i - 1; j >= 0; j -= 1)
3181 {
3182 if (encoded_ordered_before (syms[j].symbol->linkage_name (),
3183 sym.symbol->linkage_name ()))
3184 break;
3185 syms[j + 1] = syms[j];
3186 }
3187 syms[j + 1] = sym;
3188 }
3189 }
3190
3191 /* Whether GDB should display formals and return types for functions in the
3192 overloads selection menu. */
3193 static bool print_signatures = true;
3194
3195 /* Print the signature for SYM on STREAM according to the FLAGS options. For
3196 all but functions, the signature is just the name of the symbol. For
3197 functions, this is the name of the function, the list of types for formals
3198 and the return type (if any). */
3199
3200 static void
3201 ada_print_symbol_signature (struct ui_file *stream, struct symbol *sym,
3202 const struct type_print_options *flags)
3203 {
3204 struct type *type = SYMBOL_TYPE (sym);
3205
3206 fprintf_filtered (stream, "%s", sym->print_name ());
3207 if (!print_signatures
3208 || type == NULL
3209 || type->code () != TYPE_CODE_FUNC)
3210 return;
3211
3212 if (type->num_fields () > 0)
3213 {
3214 int i;
3215
3216 fprintf_filtered (stream, " (");
3217 for (i = 0; i < type->num_fields (); ++i)
3218 {
3219 if (i > 0)
3220 fprintf_filtered (stream, "; ");
3221 ada_print_type (TYPE_FIELD_TYPE (type, i), NULL, stream, -1, 0,
3222 flags);
3223 }
3224 fprintf_filtered (stream, ")");
3225 }
3226 if (TYPE_TARGET_TYPE (type) != NULL
3227 && TYPE_TARGET_TYPE (type)->code () != TYPE_CODE_VOID)
3228 {
3229 fprintf_filtered (stream, " return ");
3230 ada_print_type (TYPE_TARGET_TYPE (type), NULL, stream, -1, 0, flags);
3231 }
3232 }
3233
3234 /* Read and validate a set of numeric choices from the user in the
3235 range 0 .. N_CHOICES-1. Place the results in increasing
3236 order in CHOICES[0 .. N-1], and return N.
3237
3238 The user types choices as a sequence of numbers on one line
3239 separated by blanks, encoding them as follows:
3240
3241 + A choice of 0 means to cancel the selection, throwing an error.
3242 + If IS_ALL_CHOICE, a choice of 1 selects the entire set 0 .. N_CHOICES-1.
3243 + The user chooses k by typing k+IS_ALL_CHOICE+1.
3244
3245 The user is not allowed to choose more than MAX_RESULTS values.
3246
3247 ANNOTATION_SUFFIX, if present, is used to annotate the input
3248 prompts (for use with the -f switch). */
3249
3250 static int
3251 get_selections (int *choices, int n_choices, int max_results,
3252 int is_all_choice, const char *annotation_suffix)
3253 {
3254 const char *args;
3255 const char *prompt;
3256 int n_chosen;
3257 int first_choice = is_all_choice ? 2 : 1;
3258
3259 prompt = getenv ("PS2");
3260 if (prompt == NULL)
3261 prompt = "> ";
3262
3263 args = command_line_input (prompt, annotation_suffix);
3264
3265 if (args == NULL)
3266 error_no_arg (_("one or more choice numbers"));
3267
3268 n_chosen = 0;
3269
3270 /* Set choices[0 .. n_chosen-1] to the users' choices in ascending
3271 order, as given in args. Choices are validated. */
3272 while (1)
3273 {
3274 char *args2;
3275 int choice, j;
3276
3277 args = skip_spaces (args);
3278 if (*args == '\0' && n_chosen == 0)
3279 error_no_arg (_("one or more choice numbers"));
3280 else if (*args == '\0')
3281 break;
3282
3283 choice = strtol (args, &args2, 10);
3284 if (args == args2 || choice < 0
3285 || choice > n_choices + first_choice - 1)
3286 error (_("Argument must be choice number"));
3287 args = args2;
3288
3289 if (choice == 0)
3290 error (_("cancelled"));
3291
3292 if (choice < first_choice)
3293 {
3294 n_chosen = n_choices;
3295 for (j = 0; j < n_choices; j += 1)
3296 choices[j] = j;
3297 break;
3298 }
3299 choice -= first_choice;
3300
3301 for (j = n_chosen - 1; j >= 0 && choice < choices[j]; j -= 1)
3302 {
3303 }
3304
3305 if (j < 0 || choice != choices[j])
3306 {
3307 int k;
3308
3309 for (k = n_chosen - 1; k > j; k -= 1)
3310 choices[k + 1] = choices[k];
3311 choices[j + 1] = choice;
3312 n_chosen += 1;
3313 }
3314 }
3315
3316 if (n_chosen > max_results)
3317 error (_("Select no more than %d of the above"), max_results);
3318
3319 return n_chosen;
3320 }
3321
3322 /* Given a list of NSYMS symbols in SYMS, select up to MAX_RESULTS>0
3323 by asking the user (if necessary), returning the number selected,
3324 and setting the first elements of SYMS items. Error if no symbols
3325 selected. */
3326
3327 /* NOTE: Adapted from decode_line_2 in symtab.c, with which it ought
3328 to be re-integrated one of these days. */
3329
3330 static int
3331 user_select_syms (struct block_symbol *syms, int nsyms, int max_results)
3332 {
3333 int i;
3334 int *chosen = XALLOCAVEC (int , nsyms);
3335 int n_chosen;
3336 int first_choice = (max_results == 1) ? 1 : 2;
3337 const char *select_mode = multiple_symbols_select_mode ();
3338
3339 if (max_results < 1)
3340 error (_("Request to select 0 symbols!"));
3341 if (nsyms <= 1)
3342 return nsyms;
3343
3344 if (select_mode == multiple_symbols_cancel)
3345 error (_("\
3346 canceled because the command is ambiguous\n\
3347 See set/show multiple-symbol."));
3348
3349 /* If select_mode is "all", then return all possible symbols.
3350 Only do that if more than one symbol can be selected, of course.
3351 Otherwise, display the menu as usual. */
3352 if (select_mode == multiple_symbols_all && max_results > 1)
3353 return nsyms;
3354
3355 printf_filtered (_("[0] cancel\n"));
3356 if (max_results > 1)
3357 printf_filtered (_("[1] all\n"));
3358
3359 sort_choices (syms, nsyms);
3360
3361 for (i = 0; i < nsyms; i += 1)
3362 {
3363 if (syms[i].symbol == NULL)
3364 continue;
3365
3366 if (SYMBOL_CLASS (syms[i].symbol) == LOC_BLOCK)
3367 {
3368 struct symtab_and_line sal =
3369 find_function_start_sal (syms[i].symbol, 1);
3370
3371 printf_filtered ("[%d] ", i + first_choice);
3372 ada_print_symbol_signature (gdb_stdout, syms[i].symbol,
3373 &type_print_raw_options);
3374 if (sal.symtab == NULL)
3375 printf_filtered (_(" at %p[<no source file available>%p]:%d\n"),
3376 metadata_style.style ().ptr (), nullptr, sal.line);
3377 else
3378 printf_filtered
3379 (_(" at %ps:%d\n"),
3380 styled_string (file_name_style.style (),
3381 symtab_to_filename_for_display (sal.symtab)),
3382 sal.line);
3383 continue;
3384 }
3385 else
3386 {
3387 int is_enumeral =
3388 (SYMBOL_CLASS (syms[i].symbol) == LOC_CONST
3389 && SYMBOL_TYPE (syms[i].symbol) != NULL
3390 && SYMBOL_TYPE (syms[i].symbol)->code () == TYPE_CODE_ENUM);
3391 struct symtab *symtab = NULL;
3392
3393 if (SYMBOL_OBJFILE_OWNED (syms[i].symbol))
3394 symtab = symbol_symtab (syms[i].symbol);
3395
3396 if (SYMBOL_LINE (syms[i].symbol) != 0 && symtab != NULL)
3397 {
3398 printf_filtered ("[%d] ", i + first_choice);
3399 ada_print_symbol_signature (gdb_stdout, syms[i].symbol,
3400 &type_print_raw_options);
3401 printf_filtered (_(" at %s:%d\n"),
3402 symtab_to_filename_for_display (symtab),
3403 SYMBOL_LINE (syms[i].symbol));
3404 }
3405 else if (is_enumeral
3406 && SYMBOL_TYPE (syms[i].symbol)->name () != NULL)
3407 {
3408 printf_filtered (("[%d] "), i + first_choice);
3409 ada_print_type (SYMBOL_TYPE (syms[i].symbol), NULL,
3410 gdb_stdout, -1, 0, &type_print_raw_options);
3411 printf_filtered (_("'(%s) (enumeral)\n"),
3412 syms[i].symbol->print_name ());
3413 }
3414 else
3415 {
3416 printf_filtered ("[%d] ", i + first_choice);
3417 ada_print_symbol_signature (gdb_stdout, syms[i].symbol,
3418 &type_print_raw_options);
3419
3420 if (symtab != NULL)
3421 printf_filtered (is_enumeral
3422 ? _(" in %s (enumeral)\n")
3423 : _(" at %s:?\n"),
3424 symtab_to_filename_for_display (symtab));
3425 else
3426 printf_filtered (is_enumeral
3427 ? _(" (enumeral)\n")
3428 : _(" at ?\n"));
3429 }
3430 }
3431 }
3432
3433 n_chosen = get_selections (chosen, nsyms, max_results, max_results > 1,
3434 "overload-choice");
3435
3436 for (i = 0; i < n_chosen; i += 1)
3437 syms[i] = syms[chosen[i]];
3438
3439 return n_chosen;
3440 }
3441
3442 /* Same as evaluate_type (*EXP), but resolves ambiguous symbol
3443 references (marked by OP_VAR_VALUE nodes in which the symbol has an
3444 undefined namespace) and converts operators that are
3445 user-defined into appropriate function calls. If CONTEXT_TYPE is
3446 non-null, it provides a preferred result type [at the moment, only
3447 type void has any effect---causing procedures to be preferred over
3448 functions in calls]. A null CONTEXT_TYPE indicates that a non-void
3449 return type is preferred. May change (expand) *EXP. */
3450
3451 static void
3452 resolve (expression_up *expp, int void_context_p, int parse_completion,
3453 innermost_block_tracker *tracker)
3454 {
3455 struct type *context_type = NULL;
3456 int pc = 0;
3457
3458 if (void_context_p)
3459 context_type = builtin_type ((*expp)->gdbarch)->builtin_void;
3460
3461 resolve_subexp (expp, &pc, 1, context_type, parse_completion, tracker);
3462 }
3463
3464 /* Resolve the operator of the subexpression beginning at
3465 position *POS of *EXPP. "Resolving" consists of replacing
3466 the symbols that have undefined namespaces in OP_VAR_VALUE nodes
3467 with their resolutions, replacing built-in operators with
3468 function calls to user-defined operators, where appropriate, and,
3469 when DEPROCEDURE_P is non-zero, converting function-valued variables
3470 into parameterless calls. May expand *EXPP. The CONTEXT_TYPE functions
3471 are as in ada_resolve, above. */
3472
3473 static struct value *
3474 resolve_subexp (expression_up *expp, int *pos, int deprocedure_p,
3475 struct type *context_type, int parse_completion,
3476 innermost_block_tracker *tracker)
3477 {
3478 int pc = *pos;
3479 int i;
3480 struct expression *exp; /* Convenience: == *expp. */
3481 enum exp_opcode op = (*expp)->elts[pc].opcode;
3482 struct value **argvec; /* Vector of operand types (alloca'ed). */
3483 int nargs; /* Number of operands. */
3484 int oplen;
3485
3486 argvec = NULL;
3487 nargs = 0;
3488 exp = expp->get ();
3489
3490 /* Pass one: resolve operands, saving their types and updating *pos,
3491 if needed. */
3492 switch (op)
3493 {
3494 case OP_FUNCALL:
3495 if (exp->elts[pc + 3].opcode == OP_VAR_VALUE
3496 && SYMBOL_DOMAIN (exp->elts[pc + 5].symbol) == UNDEF_DOMAIN)
3497 *pos += 7;
3498 else
3499 {
3500 *pos += 3;
3501 resolve_subexp (expp, pos, 0, NULL, parse_completion, tracker);
3502 }
3503 nargs = longest_to_int (exp->elts[pc + 1].longconst);
3504 break;
3505
3506 case UNOP_ADDR:
3507 *pos += 1;
3508 resolve_subexp (expp, pos, 0, NULL, parse_completion, tracker);
3509 break;
3510
3511 case UNOP_QUAL:
3512 *pos += 3;
3513 resolve_subexp (expp, pos, 1, check_typedef (exp->elts[pc + 1].type),
3514 parse_completion, tracker);
3515 break;
3516
3517 case OP_ATR_MODULUS:
3518 case OP_ATR_SIZE:
3519 case OP_ATR_TAG:
3520 case OP_ATR_FIRST:
3521 case OP_ATR_LAST:
3522 case OP_ATR_LENGTH:
3523 case OP_ATR_POS:
3524 case OP_ATR_VAL:
3525 case OP_ATR_MIN:
3526 case OP_ATR_MAX:
3527 case TERNOP_IN_RANGE:
3528 case BINOP_IN_BOUNDS:
3529 case UNOP_IN_RANGE:
3530 case OP_AGGREGATE:
3531 case OP_OTHERS:
3532 case OP_CHOICES:
3533 case OP_POSITIONAL:
3534 case OP_DISCRETE_RANGE:
3535 case OP_NAME:
3536 ada_forward_operator_length (exp, pc, &oplen, &nargs);
3537 *pos += oplen;
3538 break;
3539
3540 case BINOP_ASSIGN:
3541 {
3542 struct value *arg1;
3543
3544 *pos += 1;
3545 arg1 = resolve_subexp (expp, pos, 0, NULL, parse_completion, tracker);
3546 if (arg1 == NULL)
3547 resolve_subexp (expp, pos, 1, NULL, parse_completion, tracker);
3548 else
3549 resolve_subexp (expp, pos, 1, value_type (arg1), parse_completion,
3550 tracker);
3551 break;
3552 }
3553
3554 case UNOP_CAST:
3555 *pos += 3;
3556 nargs = 1;
3557 break;
3558
3559 case BINOP_ADD:
3560 case BINOP_SUB:
3561 case BINOP_MUL:
3562 case BINOP_DIV:
3563 case BINOP_REM:
3564 case BINOP_MOD:
3565 case BINOP_EXP:
3566 case BINOP_CONCAT:
3567 case BINOP_LOGICAL_AND:
3568 case BINOP_LOGICAL_OR:
3569 case BINOP_BITWISE_AND:
3570 case BINOP_BITWISE_IOR:
3571 case BINOP_BITWISE_XOR:
3572
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
3580 case BINOP_REPEAT:
3581 case BINOP_SUBSCRIPT:
3582 case BINOP_COMMA:
3583 *pos += 1;
3584 nargs = 2;
3585 break;
3586
3587 case UNOP_NEG:
3588 case UNOP_PLUS:
3589 case UNOP_LOGICAL_NOT:
3590 case UNOP_ABS:
3591 case UNOP_IND:
3592 *pos += 1;
3593 nargs = 1;
3594 break;
3595
3596 case OP_LONG:
3597 case OP_FLOAT:
3598 case OP_VAR_VALUE:
3599 case OP_VAR_MSYM_VALUE:
3600 *pos += 4;
3601 break;
3602
3603 case OP_TYPE:
3604 case OP_BOOL:
3605 case OP_LAST:
3606 case OP_INTERNALVAR:
3607 *pos += 3;
3608 break;
3609
3610 case UNOP_MEMVAL:
3611 *pos += 3;
3612 nargs = 1;
3613 break;
3614
3615 case OP_REGISTER:
3616 *pos += 4 + BYTES_TO_EXP_ELEM (exp->elts[pc + 1].longconst + 1);
3617 break;
3618
3619 case STRUCTOP_STRUCT:
3620 *pos += 4 + BYTES_TO_EXP_ELEM (exp->elts[pc + 1].longconst + 1);
3621 nargs = 1;
3622 break;
3623
3624 case TERNOP_SLICE:
3625 *pos += 1;
3626 nargs = 3;
3627 break;
3628
3629 case OP_STRING:
3630 break;
3631
3632 default:
3633 error (_("Unexpected operator during name resolution"));
3634 }
3635
3636 argvec = XALLOCAVEC (struct value *, nargs + 1);
3637 for (i = 0; i < nargs; i += 1)
3638 argvec[i] = resolve_subexp (expp, pos, 1, NULL, parse_completion,
3639 tracker);
3640 argvec[i] = NULL;
3641 exp = expp->get ();
3642
3643 /* Pass two: perform any resolution on principal operator. */
3644 switch (op)
3645 {
3646 default:
3647 break;
3648
3649 case OP_VAR_VALUE:
3650 if (SYMBOL_DOMAIN (exp->elts[pc + 2].symbol) == UNDEF_DOMAIN)
3651 {
3652 std::vector<struct block_symbol> candidates;
3653 int n_candidates;
3654
3655 n_candidates =
3656 ada_lookup_symbol_list (exp->elts[pc + 2].symbol->linkage_name (),
3657 exp->elts[pc + 1].block, VAR_DOMAIN,
3658 &candidates);
3659
3660 if (n_candidates > 1)
3661 {
3662 /* Types tend to get re-introduced locally, so if there
3663 are any local symbols that are not types, first filter
3664 out all types. */
3665 int j;
3666 for (j = 0; j < n_candidates; j += 1)
3667 switch (SYMBOL_CLASS (candidates[j].symbol))
3668 {
3669 case LOC_REGISTER:
3670 case LOC_ARG:
3671 case LOC_REF_ARG:
3672 case LOC_REGPARM_ADDR:
3673 case LOC_LOCAL:
3674 case LOC_COMPUTED:
3675 goto FoundNonType;
3676 default:
3677 break;
3678 }
3679 FoundNonType:
3680 if (j < n_candidates)
3681 {
3682 j = 0;
3683 while (j < n_candidates)
3684 {
3685 if (SYMBOL_CLASS (candidates[j].symbol) == LOC_TYPEDEF)
3686 {
3687 candidates[j] = candidates[n_candidates - 1];
3688 n_candidates -= 1;
3689 }
3690 else
3691 j += 1;
3692 }
3693 }
3694 }
3695
3696 if (n_candidates == 0)
3697 error (_("No definition found for %s"),
3698 exp->elts[pc + 2].symbol->print_name ());
3699 else if (n_candidates == 1)
3700 i = 0;
3701 else if (deprocedure_p
3702 && !is_nonfunction (candidates.data (), n_candidates))
3703 {
3704 i = ada_resolve_function
3705 (candidates.data (), n_candidates, NULL, 0,
3706 exp->elts[pc + 2].symbol->linkage_name (),
3707 context_type, parse_completion);
3708 if (i < 0)
3709 error (_("Could not find a match for %s"),
3710 exp->elts[pc + 2].symbol->print_name ());
3711 }
3712 else
3713 {
3714 printf_filtered (_("Multiple matches for %s\n"),
3715 exp->elts[pc + 2].symbol->print_name ());
3716 user_select_syms (candidates.data (), n_candidates, 1);
3717 i = 0;
3718 }
3719
3720 exp->elts[pc + 1].block = candidates[i].block;
3721 exp->elts[pc + 2].symbol = candidates[i].symbol;
3722 tracker->update (candidates[i]);
3723 }
3724
3725 if (deprocedure_p
3726 && (SYMBOL_TYPE (exp->elts[pc + 2].symbol)->code ()
3727 == TYPE_CODE_FUNC))
3728 {
3729 replace_operator_with_call (expp, pc, 0, 4,
3730 exp->elts[pc + 2].symbol,
3731 exp->elts[pc + 1].block);
3732 exp = expp->get ();
3733 }
3734 break;
3735
3736 case OP_FUNCALL:
3737 {
3738 if (exp->elts[pc + 3].opcode == OP_VAR_VALUE
3739 && SYMBOL_DOMAIN (exp->elts[pc + 5].symbol) == UNDEF_DOMAIN)
3740 {
3741 std::vector<struct block_symbol> candidates;
3742 int n_candidates;
3743
3744 n_candidates =
3745 ada_lookup_symbol_list (exp->elts[pc + 5].symbol->linkage_name (),
3746 exp->elts[pc + 4].block, VAR_DOMAIN,
3747 &candidates);
3748
3749 if (n_candidates == 1)
3750 i = 0;
3751 else
3752 {
3753 i = ada_resolve_function
3754 (candidates.data (), n_candidates,
3755 argvec, nargs,
3756 exp->elts[pc + 5].symbol->linkage_name (),
3757 context_type, parse_completion);
3758 if (i < 0)
3759 error (_("Could not find a match for %s"),
3760 exp->elts[pc + 5].symbol->print_name ());
3761 }
3762
3763 exp->elts[pc + 4].block = candidates[i].block;
3764 exp->elts[pc + 5].symbol = candidates[i].symbol;
3765 tracker->update (candidates[i]);
3766 }
3767 }
3768 break;
3769 case BINOP_ADD:
3770 case BINOP_SUB:
3771 case BINOP_MUL:
3772 case BINOP_DIV:
3773 case BINOP_REM:
3774 case BINOP_MOD:
3775 case BINOP_CONCAT:
3776 case BINOP_BITWISE_AND:
3777 case BINOP_BITWISE_IOR:
3778 case BINOP_BITWISE_XOR:
3779 case BINOP_EQUAL:
3780 case BINOP_NOTEQUAL:
3781 case BINOP_LESS:
3782 case BINOP_GTR:
3783 case BINOP_LEQ:
3784 case BINOP_GEQ:
3785 case BINOP_EXP:
3786 case UNOP_NEG:
3787 case UNOP_PLUS:
3788 case UNOP_LOGICAL_NOT:
3789 case UNOP_ABS:
3790 if (possible_user_operator_p (op, argvec))
3791 {
3792 std::vector<struct block_symbol> candidates;
3793 int n_candidates;
3794
3795 n_candidates =
3796 ada_lookup_symbol_list (ada_decoded_op_name (op),
3797 NULL, VAR_DOMAIN,
3798 &candidates);
3799
3800 i = ada_resolve_function (candidates.data (), n_candidates, argvec,
3801 nargs, ada_decoded_op_name (op), NULL,
3802 parse_completion);
3803 if (i < 0)
3804 break;
3805
3806 replace_operator_with_call (expp, pc, nargs, 1,
3807 candidates[i].symbol,
3808 candidates[i].block);
3809 exp = expp->get ();
3810 }
3811 break;
3812
3813 case OP_TYPE:
3814 case OP_REGISTER:
3815 return NULL;
3816 }
3817
3818 *pos = pc;
3819 if (exp->elts[pc].opcode == OP_VAR_MSYM_VALUE)
3820 return evaluate_var_msym_value (EVAL_AVOID_SIDE_EFFECTS,
3821 exp->elts[pc + 1].objfile,
3822 exp->elts[pc + 2].msymbol);
3823 else
3824 return evaluate_subexp_type (exp, pos);
3825 }
3826
3827 /* Return non-zero if formal type FTYPE matches actual type ATYPE. If
3828 MAY_DEREF is non-zero, the formal may be a pointer and the actual
3829 a non-pointer. */
3830 /* The term "match" here is rather loose. The match is heuristic and
3831 liberal. */
3832
3833 static int
3834 ada_type_match (struct type *ftype, struct type *atype, int may_deref)
3835 {
3836 ftype = ada_check_typedef (ftype);
3837 atype = ada_check_typedef (atype);
3838
3839 if (ftype->code () == TYPE_CODE_REF)
3840 ftype = TYPE_TARGET_TYPE (ftype);
3841 if (atype->code () == TYPE_CODE_REF)
3842 atype = TYPE_TARGET_TYPE (atype);
3843
3844 switch (ftype->code ())
3845 {
3846 default:
3847 return ftype->code () == atype->code ();
3848 case TYPE_CODE_PTR:
3849 if (atype->code () == TYPE_CODE_PTR)
3850 return ada_type_match (TYPE_TARGET_TYPE (ftype),
3851 TYPE_TARGET_TYPE (atype), 0);
3852 else
3853 return (may_deref
3854 && ada_type_match (TYPE_TARGET_TYPE (ftype), atype, 0));
3855 case TYPE_CODE_INT:
3856 case TYPE_CODE_ENUM:
3857 case TYPE_CODE_RANGE:
3858 switch (atype->code ())
3859 {
3860 case TYPE_CODE_INT:
3861 case TYPE_CODE_ENUM:
3862 case TYPE_CODE_RANGE:
3863 return 1;
3864 default:
3865 return 0;
3866 }
3867
3868 case TYPE_CODE_ARRAY:
3869 return (atype->code () == TYPE_CODE_ARRAY
3870 || ada_is_array_descriptor_type (atype));
3871
3872 case TYPE_CODE_STRUCT:
3873 if (ada_is_array_descriptor_type (ftype))
3874 return (atype->code () == TYPE_CODE_ARRAY
3875 || ada_is_array_descriptor_type (atype));
3876 else
3877 return (atype->code () == TYPE_CODE_STRUCT
3878 && !ada_is_array_descriptor_type (atype));
3879
3880 case TYPE_CODE_UNION:
3881 case TYPE_CODE_FLT:
3882 return (atype->code () == ftype->code ());
3883 }
3884 }
3885
3886 /* Return non-zero if the formals of FUNC "sufficiently match" the
3887 vector of actual argument types ACTUALS of size N_ACTUALS. FUNC
3888 may also be an enumeral, in which case it is treated as a 0-
3889 argument function. */
3890
3891 static int
3892 ada_args_match (struct symbol *func, struct value **actuals, int n_actuals)
3893 {
3894 int i;
3895 struct type *func_type = SYMBOL_TYPE (func);
3896
3897 if (SYMBOL_CLASS (func) == LOC_CONST
3898 && func_type->code () == TYPE_CODE_ENUM)
3899 return (n_actuals == 0);
3900 else if (func_type == NULL || func_type->code () != TYPE_CODE_FUNC)
3901 return 0;
3902
3903 if (func_type->num_fields () != n_actuals)
3904 return 0;
3905
3906 for (i = 0; i < n_actuals; i += 1)
3907 {
3908 if (actuals[i] == NULL)
3909 return 0;
3910 else
3911 {
3912 struct type *ftype = ada_check_typedef (TYPE_FIELD_TYPE (func_type,
3913 i));
3914 struct type *atype = ada_check_typedef (value_type (actuals[i]));
3915
3916 if (!ada_type_match (ftype, atype, 1))
3917 return 0;
3918 }
3919 }
3920 return 1;
3921 }
3922
3923 /* False iff function type FUNC_TYPE definitely does not produce a value
3924 compatible with type CONTEXT_TYPE. Conservatively returns 1 if
3925 FUNC_TYPE is not a valid function type with a non-null return type
3926 or an enumerated type. A null CONTEXT_TYPE indicates any non-void type. */
3927
3928 static int
3929 return_match (struct type *func_type, struct type *context_type)
3930 {
3931 struct type *return_type;
3932
3933 if (func_type == NULL)
3934 return 1;
3935
3936 if (func_type->code () == TYPE_CODE_FUNC)
3937 return_type = get_base_type (TYPE_TARGET_TYPE (func_type));
3938 else
3939 return_type = get_base_type (func_type);
3940 if (return_type == NULL)
3941 return 1;
3942
3943 context_type = get_base_type (context_type);
3944
3945 if (return_type->code () == TYPE_CODE_ENUM)
3946 return context_type == NULL || return_type == context_type;
3947 else if (context_type == NULL)
3948 return return_type->code () != TYPE_CODE_VOID;
3949 else
3950 return return_type->code () == context_type->code ();
3951 }
3952
3953
3954 /* Returns the index in SYMS[0..NSYMS-1] that contains the symbol for the
3955 function (if any) that matches the types of the NARGS arguments in
3956 ARGS. If CONTEXT_TYPE is non-null and there is at least one match
3957 that returns that type, then eliminate matches that don't. If
3958 CONTEXT_TYPE is void and there is at least one match that does not
3959 return void, eliminate all matches that do.
3960
3961 Asks the user if there is more than one match remaining. Returns -1
3962 if there is no such symbol or none is selected. NAME is used
3963 solely for messages. May re-arrange and modify SYMS in
3964 the process; the index returned is for the modified vector. */
3965
3966 static int
3967 ada_resolve_function (struct block_symbol syms[],
3968 int nsyms, struct value **args, int nargs,
3969 const char *name, struct type *context_type,
3970 int parse_completion)
3971 {
3972 int fallback;
3973 int k;
3974 int m; /* Number of hits */
3975
3976 m = 0;
3977 /* In the first pass of the loop, we only accept functions matching
3978 context_type. If none are found, we add a second pass of the loop
3979 where every function is accepted. */
3980 for (fallback = 0; m == 0 && fallback < 2; fallback++)
3981 {
3982 for (k = 0; k < nsyms; k += 1)
3983 {
3984 struct type *type = ada_check_typedef (SYMBOL_TYPE (syms[k].symbol));
3985
3986 if (ada_args_match (syms[k].symbol, args, nargs)
3987 && (fallback || return_match (type, context_type)))
3988 {
3989 syms[m] = syms[k];
3990 m += 1;
3991 }
3992 }
3993 }
3994
3995 /* If we got multiple matches, ask the user which one to use. Don't do this
3996 interactive thing during completion, though, as the purpose of the
3997 completion is providing a list of all possible matches. Prompting the
3998 user to filter it down would be completely unexpected in this case. */
3999 if (m == 0)
4000 return -1;
4001 else if (m > 1 && !parse_completion)
4002 {
4003 printf_filtered (_("Multiple matches for %s\n"), name);
4004 user_select_syms (syms, m, 1);
4005 return 0;
4006 }
4007 return 0;
4008 }
4009
4010 /* Replace the operator of length OPLEN at position PC in *EXPP with a call
4011 on the function identified by SYM and BLOCK, and taking NARGS
4012 arguments. Update *EXPP as needed to hold more space. */
4013
4014 static void
4015 replace_operator_with_call (expression_up *expp, int pc, int nargs,
4016 int oplen, struct symbol *sym,
4017 const struct block *block)
4018 {
4019 /* A new expression, with 6 more elements (3 for funcall, 4 for function
4020 symbol, -oplen for operator being replaced). */
4021 struct expression *newexp = (struct expression *)
4022 xzalloc (sizeof (struct expression)
4023 + EXP_ELEM_TO_BYTES ((*expp)->nelts + 7 - oplen));
4024 struct expression *exp = expp->get ();
4025
4026 newexp->nelts = exp->nelts + 7 - oplen;
4027 newexp->language_defn = exp->language_defn;
4028 newexp->gdbarch = exp->gdbarch;
4029 memcpy (newexp->elts, exp->elts, EXP_ELEM_TO_BYTES (pc));
4030 memcpy (newexp->elts + pc + 7, exp->elts + pc + oplen,
4031 EXP_ELEM_TO_BYTES (exp->nelts - pc - oplen));
4032
4033 newexp->elts[pc].opcode = newexp->elts[pc + 2].opcode = OP_FUNCALL;
4034 newexp->elts[pc + 1].longconst = (LONGEST) nargs;
4035
4036 newexp->elts[pc + 3].opcode = newexp->elts[pc + 6].opcode = OP_VAR_VALUE;
4037 newexp->elts[pc + 4].block = block;
4038 newexp->elts[pc + 5].symbol = sym;
4039
4040 expp->reset (newexp);
4041 }
4042
4043 /* Type-class predicates */
4044
4045 /* True iff TYPE is numeric (i.e., an INT, RANGE (of numeric type),
4046 or FLOAT). */
4047
4048 static int
4049 numeric_type_p (struct type *type)
4050 {
4051 if (type == NULL)
4052 return 0;
4053 else
4054 {
4055 switch (type->code ())
4056 {
4057 case TYPE_CODE_INT:
4058 case TYPE_CODE_FLT:
4059 return 1;
4060 case TYPE_CODE_RANGE:
4061 return (type == TYPE_TARGET_TYPE (type)
4062 || numeric_type_p (TYPE_TARGET_TYPE (type)));
4063 default:
4064 return 0;
4065 }
4066 }
4067 }
4068
4069 /* True iff TYPE is integral (an INT or RANGE of INTs). */
4070
4071 static int
4072 integer_type_p (struct type *type)
4073 {
4074 if (type == NULL)
4075 return 0;
4076 else
4077 {
4078 switch (type->code ())
4079 {
4080 case TYPE_CODE_INT:
4081 return 1;
4082 case TYPE_CODE_RANGE:
4083 return (type == TYPE_TARGET_TYPE (type)
4084 || integer_type_p (TYPE_TARGET_TYPE (type)));
4085 default:
4086 return 0;
4087 }
4088 }
4089 }
4090
4091 /* True iff TYPE is scalar (INT, RANGE, FLOAT, ENUM). */
4092
4093 static int
4094 scalar_type_p (struct type *type)
4095 {
4096 if (type == NULL)
4097 return 0;
4098 else
4099 {
4100 switch (type->code ())
4101 {
4102 case TYPE_CODE_INT:
4103 case TYPE_CODE_RANGE:
4104 case TYPE_CODE_ENUM:
4105 case TYPE_CODE_FLT:
4106 return 1;
4107 default:
4108 return 0;
4109 }
4110 }
4111 }
4112
4113 /* True iff TYPE is discrete (INT, RANGE, ENUM). */
4114
4115 static int
4116 discrete_type_p (struct type *type)
4117 {
4118 if (type == NULL)
4119 return 0;
4120 else
4121 {
4122 switch (type->code ())
4123 {
4124 case TYPE_CODE_INT:
4125 case TYPE_CODE_RANGE:
4126 case TYPE_CODE_ENUM:
4127 case TYPE_CODE_BOOL:
4128 return 1;
4129 default:
4130 return 0;
4131 }
4132 }
4133 }
4134
4135 /* Returns non-zero if OP with operands in the vector ARGS could be
4136 a user-defined function. Errs on the side of pre-defined operators
4137 (i.e., result 0). */
4138
4139 static int
4140 possible_user_operator_p (enum exp_opcode op, struct value *args[])
4141 {
4142 struct type *type0 =
4143 (args[0] == NULL) ? NULL : ada_check_typedef (value_type (args[0]));
4144 struct type *type1 =
4145 (args[1] == NULL) ? NULL : ada_check_typedef (value_type (args[1]));
4146
4147 if (type0 == NULL)
4148 return 0;
4149
4150 switch (op)
4151 {
4152 default:
4153 return 0;
4154
4155 case BINOP_ADD:
4156 case BINOP_SUB:
4157 case BINOP_MUL:
4158 case BINOP_DIV:
4159 return (!(numeric_type_p (type0) && numeric_type_p (type1)));
4160
4161 case BINOP_REM:
4162 case BINOP_MOD:
4163 case BINOP_BITWISE_AND:
4164 case BINOP_BITWISE_IOR:
4165 case BINOP_BITWISE_XOR:
4166 return (!(integer_type_p (type0) && integer_type_p (type1)));
4167
4168 case BINOP_EQUAL:
4169 case BINOP_NOTEQUAL:
4170 case BINOP_LESS:
4171 case BINOP_GTR:
4172 case BINOP_LEQ:
4173 case BINOP_GEQ:
4174 return (!(scalar_type_p (type0) && scalar_type_p (type1)));
4175
4176 case BINOP_CONCAT:
4177 return !ada_is_array_type (type0) || !ada_is_array_type (type1);
4178
4179 case BINOP_EXP:
4180 return (!(numeric_type_p (type0) && integer_type_p (type1)));
4181
4182 case UNOP_NEG:
4183 case UNOP_PLUS:
4184 case UNOP_LOGICAL_NOT:
4185 case UNOP_ABS:
4186 return (!numeric_type_p (type0));
4187
4188 }
4189 }
4190 \f
4191 /* Renaming */
4192
4193 /* NOTES:
4194
4195 1. In the following, we assume that a renaming type's name may
4196 have an ___XD suffix. It would be nice if this went away at some
4197 point.
4198 2. We handle both the (old) purely type-based representation of
4199 renamings and the (new) variable-based encoding. At some point,
4200 it is devoutly to be hoped that the former goes away
4201 (FIXME: hilfinger-2007-07-09).
4202 3. Subprogram renamings are not implemented, although the XRS
4203 suffix is recognized (FIXME: hilfinger-2007-07-09). */
4204
4205 /* If SYM encodes a renaming,
4206
4207 <renaming> renames <renamed entity>,
4208
4209 sets *LEN to the length of the renamed entity's name,
4210 *RENAMED_ENTITY to that name (not null-terminated), and *RENAMING_EXPR to
4211 the string describing the subcomponent selected from the renamed
4212 entity. Returns ADA_NOT_RENAMING if SYM does not encode a renaming
4213 (in which case, the values of *RENAMED_ENTITY, *LEN, and *RENAMING_EXPR
4214 are undefined). Otherwise, returns a value indicating the category
4215 of entity renamed: an object (ADA_OBJECT_RENAMING), exception
4216 (ADA_EXCEPTION_RENAMING), package (ADA_PACKAGE_RENAMING), or
4217 subprogram (ADA_SUBPROGRAM_RENAMING). Does no allocation; the
4218 strings returned in *RENAMED_ENTITY and *RENAMING_EXPR should not be
4219 deallocated. The values of RENAMED_ENTITY, LEN, or RENAMING_EXPR
4220 may be NULL, in which case they are not assigned.
4221
4222 [Currently, however, GCC does not generate subprogram renamings.] */
4223
4224 enum ada_renaming_category
4225 ada_parse_renaming (struct symbol *sym,
4226 const char **renamed_entity, int *len,
4227 const char **renaming_expr)
4228 {
4229 enum ada_renaming_category kind;
4230 const char *info;
4231 const char *suffix;
4232
4233 if (sym == NULL)
4234 return ADA_NOT_RENAMING;
4235 switch (SYMBOL_CLASS (sym))
4236 {
4237 default:
4238 return ADA_NOT_RENAMING;
4239 case LOC_LOCAL:
4240 case LOC_STATIC:
4241 case LOC_COMPUTED:
4242 case LOC_OPTIMIZED_OUT:
4243 info = strstr (sym->linkage_name (), "___XR");
4244 if (info == NULL)
4245 return ADA_NOT_RENAMING;
4246 switch (info[5])
4247 {
4248 case '_':
4249 kind = ADA_OBJECT_RENAMING;
4250 info += 6;
4251 break;
4252 case 'E':
4253 kind = ADA_EXCEPTION_RENAMING;
4254 info += 7;
4255 break;
4256 case 'P':
4257 kind = ADA_PACKAGE_RENAMING;
4258 info += 7;
4259 break;
4260 case 'S':
4261 kind = ADA_SUBPROGRAM_RENAMING;
4262 info += 7;
4263 break;
4264 default:
4265 return ADA_NOT_RENAMING;
4266 }
4267 }
4268
4269 if (renamed_entity != NULL)
4270 *renamed_entity = info;
4271 suffix = strstr (info, "___XE");
4272 if (suffix == NULL || suffix == info)
4273 return ADA_NOT_RENAMING;
4274 if (len != NULL)
4275 *len = strlen (info) - strlen (suffix);
4276 suffix += 5;
4277 if (renaming_expr != NULL)
4278 *renaming_expr = suffix;
4279 return kind;
4280 }
4281
4282 /* Compute the value of the given RENAMING_SYM, which is expected to
4283 be a symbol encoding a renaming expression. BLOCK is the block
4284 used to evaluate the renaming. */
4285
4286 static struct value *
4287 ada_read_renaming_var_value (struct symbol *renaming_sym,
4288 const struct block *block)
4289 {
4290 const char *sym_name;
4291
4292 sym_name = renaming_sym->linkage_name ();
4293 expression_up expr = parse_exp_1 (&sym_name, 0, block, 0);
4294 return evaluate_expression (expr.get ());
4295 }
4296 \f
4297
4298 /* Evaluation: Function Calls */
4299
4300 /* Return an lvalue containing the value VAL. This is the identity on
4301 lvalues, and otherwise has the side-effect of allocating memory
4302 in the inferior where a copy of the value contents is copied. */
4303
4304 static struct value *
4305 ensure_lval (struct value *val)
4306 {
4307 if (VALUE_LVAL (val) == not_lval
4308 || VALUE_LVAL (val) == lval_internalvar)
4309 {
4310 int len = TYPE_LENGTH (ada_check_typedef (value_type (val)));
4311 const CORE_ADDR addr =
4312 value_as_long (value_allocate_space_in_inferior (len));
4313
4314 VALUE_LVAL (val) = lval_memory;
4315 set_value_address (val, addr);
4316 write_memory (addr, value_contents (val), len);
4317 }
4318
4319 return val;
4320 }
4321
4322 /* Given ARG, a value of type (pointer or reference to a)*
4323 structure/union, extract the component named NAME from the ultimate
4324 target structure/union and return it as a value with its
4325 appropriate type.
4326
4327 The routine searches for NAME among all members of the structure itself
4328 and (recursively) among all members of any wrapper members
4329 (e.g., '_parent').
4330
4331 If NO_ERR, then simply return NULL in case of error, rather than
4332 calling error. */
4333
4334 static struct value *
4335 ada_value_struct_elt (struct value *arg, const char *name, int no_err)
4336 {
4337 struct type *t, *t1;
4338 struct value *v;
4339 int check_tag;
4340
4341 v = NULL;
4342 t1 = t = ada_check_typedef (value_type (arg));
4343 if (t->code () == TYPE_CODE_REF)
4344 {
4345 t1 = TYPE_TARGET_TYPE (t);
4346 if (t1 == NULL)
4347 goto BadValue;
4348 t1 = ada_check_typedef (t1);
4349 if (t1->code () == TYPE_CODE_PTR)
4350 {
4351 arg = coerce_ref (arg);
4352 t = t1;
4353 }
4354 }
4355
4356 while (t->code () == TYPE_CODE_PTR)
4357 {
4358 t1 = TYPE_TARGET_TYPE (t);
4359 if (t1 == NULL)
4360 goto BadValue;
4361 t1 = ada_check_typedef (t1);
4362 if (t1->code () == TYPE_CODE_PTR)
4363 {
4364 arg = value_ind (arg);
4365 t = t1;
4366 }
4367 else
4368 break;
4369 }
4370
4371 if (t1->code () != TYPE_CODE_STRUCT && t1->code () != TYPE_CODE_UNION)
4372 goto BadValue;
4373
4374 if (t1 == t)
4375 v = ada_search_struct_field (name, arg, 0, t);
4376 else
4377 {
4378 int bit_offset, bit_size, byte_offset;
4379 struct type *field_type;
4380 CORE_ADDR address;
4381
4382 if (t->code () == TYPE_CODE_PTR)
4383 address = value_address (ada_value_ind (arg));
4384 else
4385 address = value_address (ada_coerce_ref (arg));
4386
4387 /* Check to see if this is a tagged type. We also need to handle
4388 the case where the type is a reference to a tagged type, but
4389 we have to be careful to exclude pointers to tagged types.
4390 The latter should be shown as usual (as a pointer), whereas
4391 a reference should mostly be transparent to the user. */
4392
4393 if (ada_is_tagged_type (t1, 0)
4394 || (t1->code () == TYPE_CODE_REF
4395 && ada_is_tagged_type (TYPE_TARGET_TYPE (t1), 0)))
4396 {
4397 /* We first try to find the searched field in the current type.
4398 If not found then let's look in the fixed type. */
4399
4400 if (!find_struct_field (name, t1, 0,
4401 &field_type, &byte_offset, &bit_offset,
4402 &bit_size, NULL))
4403 check_tag = 1;
4404 else
4405 check_tag = 0;
4406 }
4407 else
4408 check_tag = 0;
4409
4410 /* Convert to fixed type in all cases, so that we have proper
4411 offsets to each field in unconstrained record types. */
4412 t1 = ada_to_fixed_type (ada_get_base_type (t1), NULL,
4413 address, NULL, check_tag);
4414
4415 if (find_struct_field (name, t1, 0,
4416 &field_type, &byte_offset, &bit_offset,
4417 &bit_size, NULL))
4418 {
4419 if (bit_size != 0)
4420 {
4421 if (t->code () == TYPE_CODE_REF)
4422 arg = ada_coerce_ref (arg);
4423 else
4424 arg = ada_value_ind (arg);
4425 v = ada_value_primitive_packed_val (arg, NULL, byte_offset,
4426 bit_offset, bit_size,
4427 field_type);
4428 }
4429 else
4430 v = value_at_lazy (field_type, address + byte_offset);
4431 }
4432 }
4433
4434 if (v != NULL || no_err)
4435 return v;
4436 else
4437 error (_("There is no member named %s."), name);
4438
4439 BadValue:
4440 if (no_err)
4441 return NULL;
4442 else
4443 error (_("Attempt to extract a component of "
4444 "a value that is not a record."));
4445 }
4446
4447 /* Return the value ACTUAL, converted to be an appropriate value for a
4448 formal of type FORMAL_TYPE. Use *SP as a stack pointer for
4449 allocating any necessary descriptors (fat pointers), or copies of
4450 values not residing in memory, updating it as needed. */
4451
4452 struct value *
4453 ada_convert_actual (struct value *actual, struct type *formal_type0)
4454 {
4455 struct type *actual_type = ada_check_typedef (value_type (actual));
4456 struct type *formal_type = ada_check_typedef (formal_type0);
4457 struct type *formal_target =
4458 formal_type->code () == TYPE_CODE_PTR
4459 ? ada_check_typedef (TYPE_TARGET_TYPE (formal_type)) : formal_type;
4460 struct type *actual_target =
4461 actual_type->code () == TYPE_CODE_PTR
4462 ? ada_check_typedef (TYPE_TARGET_TYPE (actual_type)) : actual_type;
4463
4464 if (ada_is_array_descriptor_type (formal_target)
4465 && actual_target->code () == TYPE_CODE_ARRAY)
4466 return make_array_descriptor (formal_type, actual);
4467 else if (formal_type->code () == TYPE_CODE_PTR
4468 || formal_type->code () == TYPE_CODE_REF)
4469 {
4470 struct value *result;
4471
4472 if (formal_target->code () == TYPE_CODE_ARRAY
4473 && ada_is_array_descriptor_type (actual_target))
4474 result = desc_data (actual);
4475 else if (formal_type->code () != TYPE_CODE_PTR)
4476 {
4477 if (VALUE_LVAL (actual) != lval_memory)
4478 {
4479 struct value *val;
4480
4481 actual_type = ada_check_typedef (value_type (actual));
4482 val = allocate_value (actual_type);
4483 memcpy ((char *) value_contents_raw (val),
4484 (char *) value_contents (actual),
4485 TYPE_LENGTH (actual_type));
4486 actual = ensure_lval (val);
4487 }
4488 result = value_addr (actual);
4489 }
4490 else
4491 return actual;
4492 return value_cast_pointers (formal_type, result, 0);
4493 }
4494 else if (actual_type->code () == TYPE_CODE_PTR)
4495 return ada_value_ind (actual);
4496 else if (ada_is_aligner_type (formal_type))
4497 {
4498 /* We need to turn this parameter into an aligner type
4499 as well. */
4500 struct value *aligner = allocate_value (formal_type);
4501 struct value *component = ada_value_struct_elt (aligner, "F", 0);
4502
4503 value_assign_to_component (aligner, component, actual);
4504 return aligner;
4505 }
4506
4507 return actual;
4508 }
4509
4510 /* Convert VALUE (which must be an address) to a CORE_ADDR that is a pointer of
4511 type TYPE. This is usually an inefficient no-op except on some targets
4512 (such as AVR) where the representation of a pointer and an address
4513 differs. */
4514
4515 static CORE_ADDR
4516 value_pointer (struct value *value, struct type *type)
4517 {
4518 struct gdbarch *gdbarch = get_type_arch (type);
4519 unsigned len = TYPE_LENGTH (type);
4520 gdb_byte *buf = (gdb_byte *) alloca (len);
4521 CORE_ADDR addr;
4522
4523 addr = value_address (value);
4524 gdbarch_address_to_pointer (gdbarch, type, buf, addr);
4525 addr = extract_unsigned_integer (buf, len, type_byte_order (type));
4526 return addr;
4527 }
4528
4529
4530 /* Push a descriptor of type TYPE for array value ARR on the stack at
4531 *SP, updating *SP to reflect the new descriptor. Return either
4532 an lvalue representing the new descriptor, or (if TYPE is a pointer-
4533 to-descriptor type rather than a descriptor type), a struct value *
4534 representing a pointer to this descriptor. */
4535
4536 static struct value *
4537 make_array_descriptor (struct type *type, struct value *arr)
4538 {
4539 struct type *bounds_type = desc_bounds_type (type);
4540 struct type *desc_type = desc_base_type (type);
4541 struct value *descriptor = allocate_value (desc_type);
4542 struct value *bounds = allocate_value (bounds_type);
4543 int i;
4544
4545 for (i = ada_array_arity (ada_check_typedef (value_type (arr)));
4546 i > 0; i -= 1)
4547 {
4548 modify_field (value_type (bounds), value_contents_writeable (bounds),
4549 ada_array_bound (arr, i, 0),
4550 desc_bound_bitpos (bounds_type, i, 0),
4551 desc_bound_bitsize (bounds_type, i, 0));
4552 modify_field (value_type (bounds), value_contents_writeable (bounds),
4553 ada_array_bound (arr, i, 1),
4554 desc_bound_bitpos (bounds_type, i, 1),
4555 desc_bound_bitsize (bounds_type, i, 1));
4556 }
4557
4558 bounds = ensure_lval (bounds);
4559
4560 modify_field (value_type (descriptor),
4561 value_contents_writeable (descriptor),
4562 value_pointer (ensure_lval (arr),
4563 TYPE_FIELD_TYPE (desc_type, 0)),
4564 fat_pntr_data_bitpos (desc_type),
4565 fat_pntr_data_bitsize (desc_type));
4566
4567 modify_field (value_type (descriptor),
4568 value_contents_writeable (descriptor),
4569 value_pointer (bounds,
4570 TYPE_FIELD_TYPE (desc_type, 1)),
4571 fat_pntr_bounds_bitpos (desc_type),
4572 fat_pntr_bounds_bitsize (desc_type));
4573
4574 descriptor = ensure_lval (descriptor);
4575
4576 if (type->code () == TYPE_CODE_PTR)
4577 return value_addr (descriptor);
4578 else
4579 return descriptor;
4580 }
4581 \f
4582 /* Symbol Cache Module */
4583
4584 /* Performance measurements made as of 2010-01-15 indicate that
4585 this cache does bring some noticeable improvements. Depending
4586 on the type of entity being printed, the cache can make it as much
4587 as an order of magnitude faster than without it.
4588
4589 The descriptive type DWARF extension has significantly reduced
4590 the need for this cache, at least when DWARF is being used. However,
4591 even in this case, some expensive name-based symbol searches are still
4592 sometimes necessary - to find an XVZ variable, mostly. */
4593
4594 /* Initialize the contents of SYM_CACHE. */
4595
4596 static void
4597 ada_init_symbol_cache (struct ada_symbol_cache *sym_cache)
4598 {
4599 obstack_init (&sym_cache->cache_space);
4600 memset (sym_cache->root, '\000', sizeof (sym_cache->root));
4601 }
4602
4603 /* Free the memory used by SYM_CACHE. */
4604
4605 static void
4606 ada_free_symbol_cache (struct ada_symbol_cache *sym_cache)
4607 {
4608 obstack_free (&sym_cache->cache_space, NULL);
4609 xfree (sym_cache);
4610 }
4611
4612 /* Return the symbol cache associated to the given program space PSPACE.
4613 If not allocated for this PSPACE yet, allocate and initialize one. */
4614
4615 static struct ada_symbol_cache *
4616 ada_get_symbol_cache (struct program_space *pspace)
4617 {
4618 struct ada_pspace_data *pspace_data = get_ada_pspace_data (pspace);
4619
4620 if (pspace_data->sym_cache == NULL)
4621 {
4622 pspace_data->sym_cache = XCNEW (struct ada_symbol_cache);
4623 ada_init_symbol_cache (pspace_data->sym_cache);
4624 }
4625
4626 return pspace_data->sym_cache;
4627 }
4628
4629 /* Clear all entries from the symbol cache. */
4630
4631 static void
4632 ada_clear_symbol_cache (void)
4633 {
4634 struct ada_symbol_cache *sym_cache
4635 = ada_get_symbol_cache (current_program_space);
4636
4637 obstack_free (&sym_cache->cache_space, NULL);
4638 ada_init_symbol_cache (sym_cache);
4639 }
4640
4641 /* Search our cache for an entry matching NAME and DOMAIN.
4642 Return it if found, or NULL otherwise. */
4643
4644 static struct cache_entry **
4645 find_entry (const char *name, domain_enum domain)
4646 {
4647 struct ada_symbol_cache *sym_cache
4648 = ada_get_symbol_cache (current_program_space);
4649 int h = msymbol_hash (name) % HASH_SIZE;
4650 struct cache_entry **e;
4651
4652 for (e = &sym_cache->root[h]; *e != NULL; e = &(*e)->next)
4653 {
4654 if (domain == (*e)->domain && strcmp (name, (*e)->name) == 0)
4655 return e;
4656 }
4657 return NULL;
4658 }
4659
4660 /* Search the symbol cache for an entry matching NAME and DOMAIN.
4661 Return 1 if found, 0 otherwise.
4662
4663 If an entry was found and SYM is not NULL, set *SYM to the entry's
4664 SYM. Same principle for BLOCK if not NULL. */
4665
4666 static int
4667 lookup_cached_symbol (const char *name, domain_enum domain,
4668 struct symbol **sym, const struct block **block)
4669 {
4670 struct cache_entry **e = find_entry (name, domain);
4671
4672 if (e == NULL)
4673 return 0;
4674 if (sym != NULL)
4675 *sym = (*e)->sym;
4676 if (block != NULL)
4677 *block = (*e)->block;
4678 return 1;
4679 }
4680
4681 /* Assuming that (SYM, BLOCK) is the result of the lookup of NAME
4682 in domain DOMAIN, save this result in our symbol cache. */
4683
4684 static void
4685 cache_symbol (const char *name, domain_enum domain, struct symbol *sym,
4686 const struct block *block)
4687 {
4688 struct ada_symbol_cache *sym_cache
4689 = ada_get_symbol_cache (current_program_space);
4690 int h;
4691 struct cache_entry *e;
4692
4693 /* Symbols for builtin types don't have a block.
4694 For now don't cache such symbols. */
4695 if (sym != NULL && !SYMBOL_OBJFILE_OWNED (sym))
4696 return;
4697
4698 /* If the symbol is a local symbol, then do not cache it, as a search
4699 for that symbol depends on the context. To determine whether
4700 the symbol is local or not, we check the block where we found it
4701 against the global and static blocks of its associated symtab. */
4702 if (sym
4703 && BLOCKVECTOR_BLOCK (SYMTAB_BLOCKVECTOR (symbol_symtab (sym)),
4704 GLOBAL_BLOCK) != block
4705 && BLOCKVECTOR_BLOCK (SYMTAB_BLOCKVECTOR (symbol_symtab (sym)),
4706 STATIC_BLOCK) != block)
4707 return;
4708
4709 h = msymbol_hash (name) % HASH_SIZE;
4710 e = XOBNEW (&sym_cache->cache_space, cache_entry);
4711 e->next = sym_cache->root[h];
4712 sym_cache->root[h] = e;
4713 e->name = obstack_strdup (&sym_cache->cache_space, name);
4714 e->sym = sym;
4715 e->domain = domain;
4716 e->block = block;
4717 }
4718 \f
4719 /* Symbol Lookup */
4720
4721 /* Return the symbol name match type that should be used used when
4722 searching for all symbols matching LOOKUP_NAME.
4723
4724 LOOKUP_NAME is expected to be a symbol name after transformation
4725 for Ada lookups. */
4726
4727 static symbol_name_match_type
4728 name_match_type_from_name (const char *lookup_name)
4729 {
4730 return (strstr (lookup_name, "__") == NULL
4731 ? symbol_name_match_type::WILD
4732 : symbol_name_match_type::FULL);
4733 }
4734
4735 /* Return the result of a standard (literal, C-like) lookup of NAME in
4736 given DOMAIN, visible from lexical block BLOCK. */
4737
4738 static struct symbol *
4739 standard_lookup (const char *name, const struct block *block,
4740 domain_enum domain)
4741 {
4742 /* Initialize it just to avoid a GCC false warning. */
4743 struct block_symbol sym = {};
4744
4745 if (lookup_cached_symbol (name, domain, &sym.symbol, NULL))
4746 return sym.symbol;
4747 ada_lookup_encoded_symbol (name, block, domain, &sym);
4748 cache_symbol (name, domain, sym.symbol, sym.block);
4749 return sym.symbol;
4750 }
4751
4752
4753 /* Non-zero iff there is at least one non-function/non-enumeral symbol
4754 in the symbol fields of SYMS[0..N-1]. We treat enumerals as functions,
4755 since they contend in overloading in the same way. */
4756 static int
4757 is_nonfunction (struct block_symbol syms[], int n)
4758 {
4759 int i;
4760
4761 for (i = 0; i < n; i += 1)
4762 if (SYMBOL_TYPE (syms[i].symbol)->code () != TYPE_CODE_FUNC
4763 && (SYMBOL_TYPE (syms[i].symbol)->code () != TYPE_CODE_ENUM
4764 || SYMBOL_CLASS (syms[i].symbol) != LOC_CONST))
4765 return 1;
4766
4767 return 0;
4768 }
4769
4770 /* If true (non-zero), then TYPE0 and TYPE1 represent equivalent
4771 struct types. Otherwise, they may not. */
4772
4773 static int
4774 equiv_types (struct type *type0, struct type *type1)
4775 {
4776 if (type0 == type1)
4777 return 1;
4778 if (type0 == NULL || type1 == NULL
4779 || type0->code () != type1->code ())
4780 return 0;
4781 if ((type0->code () == TYPE_CODE_STRUCT
4782 || type0->code () == TYPE_CODE_ENUM)
4783 && ada_type_name (type0) != NULL && ada_type_name (type1) != NULL
4784 && strcmp (ada_type_name (type0), ada_type_name (type1)) == 0)
4785 return 1;
4786
4787 return 0;
4788 }
4789
4790 /* True iff SYM0 represents the same entity as SYM1, or one that is
4791 no more defined than that of SYM1. */
4792
4793 static int
4794 lesseq_defined_than (struct symbol *sym0, struct symbol *sym1)
4795 {
4796 if (sym0 == sym1)
4797 return 1;
4798 if (SYMBOL_DOMAIN (sym0) != SYMBOL_DOMAIN (sym1)
4799 || SYMBOL_CLASS (sym0) != SYMBOL_CLASS (sym1))
4800 return 0;
4801
4802 switch (SYMBOL_CLASS (sym0))
4803 {
4804 case LOC_UNDEF:
4805 return 1;
4806 case LOC_TYPEDEF:
4807 {
4808 struct type *type0 = SYMBOL_TYPE (sym0);
4809 struct type *type1 = SYMBOL_TYPE (sym1);
4810 const char *name0 = sym0->linkage_name ();
4811 const char *name1 = sym1->linkage_name ();
4812 int len0 = strlen (name0);
4813
4814 return
4815 type0->code () == type1->code ()
4816 && (equiv_types (type0, type1)
4817 || (len0 < strlen (name1) && strncmp (name0, name1, len0) == 0
4818 && startswith (name1 + len0, "___XV")));
4819 }
4820 case LOC_CONST:
4821 return SYMBOL_VALUE (sym0) == SYMBOL_VALUE (sym1)
4822 && equiv_types (SYMBOL_TYPE (sym0), SYMBOL_TYPE (sym1));
4823
4824 case LOC_STATIC:
4825 {
4826 const char *name0 = sym0->linkage_name ();
4827 const char *name1 = sym1->linkage_name ();
4828 return (strcmp (name0, name1) == 0
4829 && SYMBOL_VALUE_ADDRESS (sym0) == SYMBOL_VALUE_ADDRESS (sym1));
4830 }
4831
4832 default:
4833 return 0;
4834 }
4835 }
4836
4837 /* Append (SYM,BLOCK,SYMTAB) to the end of the array of struct block_symbol
4838 records in OBSTACKP. Do nothing if SYM is a duplicate. */
4839
4840 static void
4841 add_defn_to_vec (struct obstack *obstackp,
4842 struct symbol *sym,
4843 const struct block *block)
4844 {
4845 int i;
4846 struct block_symbol *prevDefns = defns_collected (obstackp, 0);
4847
4848 /* Do not try to complete stub types, as the debugger is probably
4849 already scanning all symbols matching a certain name at the
4850 time when this function is called. Trying to replace the stub
4851 type by its associated full type will cause us to restart a scan
4852 which may lead to an infinite recursion. Instead, the client
4853 collecting the matching symbols will end up collecting several
4854 matches, with at least one of them complete. It can then filter
4855 out the stub ones if needed. */
4856
4857 for (i = num_defns_collected (obstackp) - 1; i >= 0; i -= 1)
4858 {
4859 if (lesseq_defined_than (sym, prevDefns[i].symbol))
4860 return;
4861 else if (lesseq_defined_than (prevDefns[i].symbol, sym))
4862 {
4863 prevDefns[i].symbol = sym;
4864 prevDefns[i].block = block;
4865 return;
4866 }
4867 }
4868
4869 {
4870 struct block_symbol info;
4871
4872 info.symbol = sym;
4873 info.block = block;
4874 obstack_grow (obstackp, &info, sizeof (struct block_symbol));
4875 }
4876 }
4877
4878 /* Number of block_symbol structures currently collected in current vector in
4879 OBSTACKP. */
4880
4881 static int
4882 num_defns_collected (struct obstack *obstackp)
4883 {
4884 return obstack_object_size (obstackp) / sizeof (struct block_symbol);
4885 }
4886
4887 /* Vector of block_symbol structures currently collected in current vector in
4888 OBSTACKP. If FINISH, close off the vector and return its final address. */
4889
4890 static struct block_symbol *
4891 defns_collected (struct obstack *obstackp, int finish)
4892 {
4893 if (finish)
4894 return (struct block_symbol *) obstack_finish (obstackp);
4895 else
4896 return (struct block_symbol *) obstack_base (obstackp);
4897 }
4898
4899 /* Return a bound minimal symbol matching NAME according to Ada
4900 decoding rules. Returns an invalid symbol if there is no such
4901 minimal symbol. Names prefixed with "standard__" are handled
4902 specially: "standard__" is first stripped off, and only static and
4903 global symbols are searched. */
4904
4905 struct bound_minimal_symbol
4906 ada_lookup_simple_minsym (const char *name)
4907 {
4908 struct bound_minimal_symbol result;
4909
4910 memset (&result, 0, sizeof (result));
4911
4912 symbol_name_match_type match_type = name_match_type_from_name (name);
4913 lookup_name_info lookup_name (name, match_type);
4914
4915 symbol_name_matcher_ftype *match_name
4916 = ada_get_symbol_name_matcher (lookup_name);
4917
4918 for (objfile *objfile : current_program_space->objfiles ())
4919 {
4920 for (minimal_symbol *msymbol : objfile->msymbols ())
4921 {
4922 if (match_name (msymbol->linkage_name (), lookup_name, NULL)
4923 && MSYMBOL_TYPE (msymbol) != mst_solib_trampoline)
4924 {
4925 result.minsym = msymbol;
4926 result.objfile = objfile;
4927 break;
4928 }
4929 }
4930 }
4931
4932 return result;
4933 }
4934
4935 /* For all subprograms that statically enclose the subprogram of the
4936 selected frame, add symbols matching identifier NAME in DOMAIN
4937 and their blocks to the list of data in OBSTACKP, as for
4938 ada_add_block_symbols (q.v.). If WILD_MATCH_P, treat as NAME
4939 with a wildcard prefix. */
4940
4941 static void
4942 add_symbols_from_enclosing_procs (struct obstack *obstackp,
4943 const lookup_name_info &lookup_name,
4944 domain_enum domain)
4945 {
4946 }
4947
4948 /* True if TYPE is definitely an artificial type supplied to a symbol
4949 for which no debugging information was given in the symbol file. */
4950
4951 static int
4952 is_nondebugging_type (struct type *type)
4953 {
4954 const char *name = ada_type_name (type);
4955
4956 return (name != NULL && strcmp (name, "<variable, no debug info>") == 0);
4957 }
4958
4959 /* Return nonzero if TYPE1 and TYPE2 are two enumeration types
4960 that are deemed "identical" for practical purposes.
4961
4962 This function assumes that TYPE1 and TYPE2 are both TYPE_CODE_ENUM
4963 types and that their number of enumerals is identical (in other
4964 words, type1->num_fields () == type2->num_fields ()). */
4965
4966 static int
4967 ada_identical_enum_types_p (struct type *type1, struct type *type2)
4968 {
4969 int i;
4970
4971 /* The heuristic we use here is fairly conservative. We consider
4972 that 2 enumerate types are identical if they have the same
4973 number of enumerals and that all enumerals have the same
4974 underlying value and name. */
4975
4976 /* All enums in the type should have an identical underlying value. */
4977 for (i = 0; i < type1->num_fields (); i++)
4978 if (TYPE_FIELD_ENUMVAL (type1, i) != TYPE_FIELD_ENUMVAL (type2, i))
4979 return 0;
4980
4981 /* All enumerals should also have the same name (modulo any numerical
4982 suffix). */
4983 for (i = 0; i < type1->num_fields (); i++)
4984 {
4985 const char *name_1 = TYPE_FIELD_NAME (type1, i);
4986 const char *name_2 = TYPE_FIELD_NAME (type2, i);
4987 int len_1 = strlen (name_1);
4988 int len_2 = strlen (name_2);
4989
4990 ada_remove_trailing_digits (TYPE_FIELD_NAME (type1, i), &len_1);
4991 ada_remove_trailing_digits (TYPE_FIELD_NAME (type2, i), &len_2);
4992 if (len_1 != len_2
4993 || strncmp (TYPE_FIELD_NAME (type1, i),
4994 TYPE_FIELD_NAME (type2, i),
4995 len_1) != 0)
4996 return 0;
4997 }
4998
4999 return 1;
5000 }
5001
5002 /* Return nonzero if all the symbols in SYMS are all enumeral symbols
5003 that are deemed "identical" for practical purposes. Sometimes,
5004 enumerals are not strictly identical, but their types are so similar
5005 that they can be considered identical.
5006
5007 For instance, consider the following code:
5008
5009 type Color is (Black, Red, Green, Blue, White);
5010 type RGB_Color is new Color range Red .. Blue;
5011
5012 Type RGB_Color is a subrange of an implicit type which is a copy
5013 of type Color. If we call that implicit type RGB_ColorB ("B" is
5014 for "Base Type"), then type RGB_ColorB is a copy of type Color.
5015 As a result, when an expression references any of the enumeral
5016 by name (Eg. "print green"), the expression is technically
5017 ambiguous and the user should be asked to disambiguate. But
5018 doing so would only hinder the user, since it wouldn't matter
5019 what choice he makes, the outcome would always be the same.
5020 So, for practical purposes, we consider them as the same. */
5021
5022 static int
5023 symbols_are_identical_enums (const std::vector<struct block_symbol> &syms)
5024 {
5025 int i;
5026
5027 /* Before performing a thorough comparison check of each type,
5028 we perform a series of inexpensive checks. We expect that these
5029 checks will quickly fail in the vast majority of cases, and thus
5030 help prevent the unnecessary use of a more expensive comparison.
5031 Said comparison also expects us to make some of these checks
5032 (see ada_identical_enum_types_p). */
5033
5034 /* Quick check: All symbols should have an enum type. */
5035 for (i = 0; i < syms.size (); i++)
5036 if (SYMBOL_TYPE (syms[i].symbol)->code () != TYPE_CODE_ENUM)
5037 return 0;
5038
5039 /* Quick check: They should all have the same value. */
5040 for (i = 1; i < syms.size (); i++)
5041 if (SYMBOL_VALUE (syms[i].symbol) != SYMBOL_VALUE (syms[0].symbol))
5042 return 0;
5043
5044 /* Quick check: They should all have the same number of enumerals. */
5045 for (i = 1; i < syms.size (); i++)
5046 if (SYMBOL_TYPE (syms[i].symbol)->num_fields ()
5047 != SYMBOL_TYPE (syms[0].symbol)->num_fields ())
5048 return 0;
5049
5050 /* All the sanity checks passed, so we might have a set of
5051 identical enumeration types. Perform a more complete
5052 comparison of the type of each symbol. */
5053 for (i = 1; i < syms.size (); i++)
5054 if (!ada_identical_enum_types_p (SYMBOL_TYPE (syms[i].symbol),
5055 SYMBOL_TYPE (syms[0].symbol)))
5056 return 0;
5057
5058 return 1;
5059 }
5060
5061 /* Remove any non-debugging symbols in SYMS that definitely
5062 duplicate other symbols in the list (The only case I know of where
5063 this happens is when object files containing stabs-in-ecoff are
5064 linked with files containing ordinary ecoff debugging symbols (or no
5065 debugging symbols)). Modifies SYMS to squeeze out deleted entries.
5066 Returns the number of items in the modified list. */
5067
5068 static int
5069 remove_extra_symbols (std::vector<struct block_symbol> *syms)
5070 {
5071 int i, j;
5072
5073 /* We should never be called with less than 2 symbols, as there
5074 cannot be any extra symbol in that case. But it's easy to
5075 handle, since we have nothing to do in that case. */
5076 if (syms->size () < 2)
5077 return syms->size ();
5078
5079 i = 0;
5080 while (i < syms->size ())
5081 {
5082 int remove_p = 0;
5083
5084 /* If two symbols have the same name and one of them is a stub type,
5085 the get rid of the stub. */
5086
5087 if (TYPE_STUB (SYMBOL_TYPE ((*syms)[i].symbol))
5088 && (*syms)[i].symbol->linkage_name () != NULL)
5089 {
5090 for (j = 0; j < syms->size (); j++)
5091 {
5092 if (j != i
5093 && !TYPE_STUB (SYMBOL_TYPE ((*syms)[j].symbol))
5094 && (*syms)[j].symbol->linkage_name () != NULL
5095 && strcmp ((*syms)[i].symbol->linkage_name (),
5096 (*syms)[j].symbol->linkage_name ()) == 0)
5097 remove_p = 1;
5098 }
5099 }
5100
5101 /* Two symbols with the same name, same class and same address
5102 should be identical. */
5103
5104 else if ((*syms)[i].symbol->linkage_name () != NULL
5105 && SYMBOL_CLASS ((*syms)[i].symbol) == LOC_STATIC
5106 && is_nondebugging_type (SYMBOL_TYPE ((*syms)[i].symbol)))
5107 {
5108 for (j = 0; j < syms->size (); j += 1)
5109 {
5110 if (i != j
5111 && (*syms)[j].symbol->linkage_name () != NULL
5112 && strcmp ((*syms)[i].symbol->linkage_name (),
5113 (*syms)[j].symbol->linkage_name ()) == 0
5114 && SYMBOL_CLASS ((*syms)[i].symbol)
5115 == SYMBOL_CLASS ((*syms)[j].symbol)
5116 && SYMBOL_VALUE_ADDRESS ((*syms)[i].symbol)
5117 == SYMBOL_VALUE_ADDRESS ((*syms)[j].symbol))
5118 remove_p = 1;
5119 }
5120 }
5121
5122 if (remove_p)
5123 syms->erase (syms->begin () + i);
5124
5125 i += 1;
5126 }
5127
5128 /* If all the remaining symbols are identical enumerals, then
5129 just keep the first one and discard the rest.
5130
5131 Unlike what we did previously, we do not discard any entry
5132 unless they are ALL identical. This is because the symbol
5133 comparison is not a strict comparison, but rather a practical
5134 comparison. If all symbols are considered identical, then
5135 we can just go ahead and use the first one and discard the rest.
5136 But if we cannot reduce the list to a single element, we have
5137 to ask the user to disambiguate anyways. And if we have to
5138 present a multiple-choice menu, it's less confusing if the list
5139 isn't missing some choices that were identical and yet distinct. */
5140 if (symbols_are_identical_enums (*syms))
5141 syms->resize (1);
5142
5143 return syms->size ();
5144 }
5145
5146 /* Given a type that corresponds to a renaming entity, use the type name
5147 to extract the scope (package name or function name, fully qualified,
5148 and following the GNAT encoding convention) where this renaming has been
5149 defined. */
5150
5151 static std::string
5152 xget_renaming_scope (struct type *renaming_type)
5153 {
5154 /* The renaming types adhere to the following convention:
5155 <scope>__<rename>___<XR extension>.
5156 So, to extract the scope, we search for the "___XR" extension,
5157 and then backtrack until we find the first "__". */
5158
5159 const char *name = renaming_type->name ();
5160 const char *suffix = strstr (name, "___XR");
5161 const char *last;
5162
5163 /* Now, backtrack a bit until we find the first "__". Start looking
5164 at suffix - 3, as the <rename> part is at least one character long. */
5165
5166 for (last = suffix - 3; last > name; last--)
5167 if (last[0] == '_' && last[1] == '_')
5168 break;
5169
5170 /* Make a copy of scope and return it. */
5171 return std::string (name, last);
5172 }
5173
5174 /* Return nonzero if NAME corresponds to a package name. */
5175
5176 static int
5177 is_package_name (const char *name)
5178 {
5179 /* Here, We take advantage of the fact that no symbols are generated
5180 for packages, while symbols are generated for each function.
5181 So the condition for NAME represent a package becomes equivalent
5182 to NAME not existing in our list of symbols. There is only one
5183 small complication with library-level functions (see below). */
5184
5185 /* If it is a function that has not been defined at library level,
5186 then we should be able to look it up in the symbols. */
5187 if (standard_lookup (name, NULL, VAR_DOMAIN) != NULL)
5188 return 0;
5189
5190 /* Library-level function names start with "_ada_". See if function
5191 "_ada_" followed by NAME can be found. */
5192
5193 /* Do a quick check that NAME does not contain "__", since library-level
5194 functions names cannot contain "__" in them. */
5195 if (strstr (name, "__") != NULL)
5196 return 0;
5197
5198 std::string fun_name = string_printf ("_ada_%s", name);
5199
5200 return (standard_lookup (fun_name.c_str (), NULL, VAR_DOMAIN) == NULL);
5201 }
5202
5203 /* Return nonzero if SYM corresponds to a renaming entity that is
5204 not visible from FUNCTION_NAME. */
5205
5206 static int
5207 old_renaming_is_invisible (const struct symbol *sym, const char *function_name)
5208 {
5209 if (SYMBOL_CLASS (sym) != LOC_TYPEDEF)
5210 return 0;
5211
5212 std::string scope = xget_renaming_scope (SYMBOL_TYPE (sym));
5213
5214 /* If the rename has been defined in a package, then it is visible. */
5215 if (is_package_name (scope.c_str ()))
5216 return 0;
5217
5218 /* Check that the rename is in the current function scope by checking
5219 that its name starts with SCOPE. */
5220
5221 /* If the function name starts with "_ada_", it means that it is
5222 a library-level function. Strip this prefix before doing the
5223 comparison, as the encoding for the renaming does not contain
5224 this prefix. */
5225 if (startswith (function_name, "_ada_"))
5226 function_name += 5;
5227
5228 return !startswith (function_name, scope.c_str ());
5229 }
5230
5231 /* Remove entries from SYMS that corresponds to a renaming entity that
5232 is not visible from the function associated with CURRENT_BLOCK or
5233 that is superfluous due to the presence of more specific renaming
5234 information. Places surviving symbols in the initial entries of
5235 SYMS and returns the number of surviving symbols.
5236
5237 Rationale:
5238 First, in cases where an object renaming is implemented as a
5239 reference variable, GNAT may produce both the actual reference
5240 variable and the renaming encoding. In this case, we discard the
5241 latter.
5242
5243 Second, GNAT emits a type following a specified encoding for each renaming
5244 entity. Unfortunately, STABS currently does not support the definition
5245 of types that are local to a given lexical block, so all renamings types
5246 are emitted at library level. As a consequence, if an application
5247 contains two renaming entities using the same name, and a user tries to
5248 print the value of one of these entities, the result of the ada symbol
5249 lookup will also contain the wrong renaming type.
5250
5251 This function partially covers for this limitation by attempting to
5252 remove from the SYMS list renaming symbols that should be visible
5253 from CURRENT_BLOCK. However, there does not seem be a 100% reliable
5254 method with the current information available. The implementation
5255 below has a couple of limitations (FIXME: brobecker-2003-05-12):
5256
5257 - When the user tries to print a rename in a function while there
5258 is another rename entity defined in a package: Normally, the
5259 rename in the function has precedence over the rename in the
5260 package, so the latter should be removed from the list. This is
5261 currently not the case.
5262
5263 - This function will incorrectly remove valid renames if
5264 the CURRENT_BLOCK corresponds to a function which symbol name
5265 has been changed by an "Export" pragma. As a consequence,
5266 the user will be unable to print such rename entities. */
5267
5268 static int
5269 remove_irrelevant_renamings (std::vector<struct block_symbol> *syms,
5270 const struct block *current_block)
5271 {
5272 struct symbol *current_function;
5273 const char *current_function_name;
5274 int i;
5275 int is_new_style_renaming;
5276
5277 /* If there is both a renaming foo___XR... encoded as a variable and
5278 a simple variable foo in the same block, discard the latter.
5279 First, zero out such symbols, then compress. */
5280 is_new_style_renaming = 0;
5281 for (i = 0; i < syms->size (); i += 1)
5282 {
5283 struct symbol *sym = (*syms)[i].symbol;
5284 const struct block *block = (*syms)[i].block;
5285 const char *name;
5286 const char *suffix;
5287
5288 if (sym == NULL || SYMBOL_CLASS (sym) == LOC_TYPEDEF)
5289 continue;
5290 name = sym->linkage_name ();
5291 suffix = strstr (name, "___XR");
5292
5293 if (suffix != NULL)
5294 {
5295 int name_len = suffix - name;
5296 int j;
5297
5298 is_new_style_renaming = 1;
5299 for (j = 0; j < syms->size (); j += 1)
5300 if (i != j && (*syms)[j].symbol != NULL
5301 && strncmp (name, (*syms)[j].symbol->linkage_name (),
5302 name_len) == 0
5303 && block == (*syms)[j].block)
5304 (*syms)[j].symbol = NULL;
5305 }
5306 }
5307 if (is_new_style_renaming)
5308 {
5309 int j, k;
5310
5311 for (j = k = 0; j < syms->size (); j += 1)
5312 if ((*syms)[j].symbol != NULL)
5313 {
5314 (*syms)[k] = (*syms)[j];
5315 k += 1;
5316 }
5317 return k;
5318 }
5319
5320 /* Extract the function name associated to CURRENT_BLOCK.
5321 Abort if unable to do so. */
5322
5323 if (current_block == NULL)
5324 return syms->size ();
5325
5326 current_function = block_linkage_function (current_block);
5327 if (current_function == NULL)
5328 return syms->size ();
5329
5330 current_function_name = current_function->linkage_name ();
5331 if (current_function_name == NULL)
5332 return syms->size ();
5333
5334 /* Check each of the symbols, and remove it from the list if it is
5335 a type corresponding to a renaming that is out of the scope of
5336 the current block. */
5337
5338 i = 0;
5339 while (i < syms->size ())
5340 {
5341 if (ada_parse_renaming ((*syms)[i].symbol, NULL, NULL, NULL)
5342 == ADA_OBJECT_RENAMING
5343 && old_renaming_is_invisible ((*syms)[i].symbol,
5344 current_function_name))
5345 syms->erase (syms->begin () + i);
5346 else
5347 i += 1;
5348 }
5349
5350 return syms->size ();
5351 }
5352
5353 /* Add to OBSTACKP all symbols from BLOCK (and its super-blocks)
5354 whose name and domain match NAME and DOMAIN respectively.
5355 If no match was found, then extend the search to "enclosing"
5356 routines (in other words, if we're inside a nested function,
5357 search the symbols defined inside the enclosing functions).
5358 If WILD_MATCH_P is nonzero, perform the naming matching in
5359 "wild" mode (see function "wild_match" for more info).
5360
5361 Note: This function assumes that OBSTACKP has 0 (zero) element in it. */
5362
5363 static void
5364 ada_add_local_symbols (struct obstack *obstackp,
5365 const lookup_name_info &lookup_name,
5366 const struct block *block, domain_enum domain)
5367 {
5368 int block_depth = 0;
5369
5370 while (block != NULL)
5371 {
5372 block_depth += 1;
5373 ada_add_block_symbols (obstackp, block, lookup_name, domain, NULL);
5374
5375 /* If we found a non-function match, assume that's the one. */
5376 if (is_nonfunction (defns_collected (obstackp, 0),
5377 num_defns_collected (obstackp)))
5378 return;
5379
5380 block = BLOCK_SUPERBLOCK (block);
5381 }
5382
5383 /* If no luck so far, try to find NAME as a local symbol in some lexically
5384 enclosing subprogram. */
5385 if (num_defns_collected (obstackp) == 0 && block_depth > 2)
5386 add_symbols_from_enclosing_procs (obstackp, lookup_name, domain);
5387 }
5388
5389 /* An object of this type is used as the user_data argument when
5390 calling the map_matching_symbols method. */
5391
5392 struct match_data
5393 {
5394 struct objfile *objfile;
5395 struct obstack *obstackp;
5396 struct symbol *arg_sym;
5397 int found_sym;
5398 };
5399
5400 /* A callback for add_nonlocal_symbols that adds symbol, found in BSYM,
5401 to a list of symbols. DATA is a pointer to a struct match_data *
5402 containing the obstack that collects the symbol list, the file that SYM
5403 must come from, a flag indicating whether a non-argument symbol has
5404 been found in the current block, and the last argument symbol
5405 passed in SYM within the current block (if any). When SYM is null,
5406 marking the end of a block, the argument symbol is added if no
5407 other has been found. */
5408
5409 static bool
5410 aux_add_nonlocal_symbols (struct block_symbol *bsym,
5411 struct match_data *data)
5412 {
5413 const struct block *block = bsym->block;
5414 struct symbol *sym = bsym->symbol;
5415
5416 if (sym == NULL)
5417 {
5418 if (!data->found_sym && data->arg_sym != NULL)
5419 add_defn_to_vec (data->obstackp,
5420 fixup_symbol_section (data->arg_sym, data->objfile),
5421 block);
5422 data->found_sym = 0;
5423 data->arg_sym = NULL;
5424 }
5425 else
5426 {
5427 if (SYMBOL_CLASS (sym) == LOC_UNRESOLVED)
5428 return true;
5429 else if (SYMBOL_IS_ARGUMENT (sym))
5430 data->arg_sym = sym;
5431 else
5432 {
5433 data->found_sym = 1;
5434 add_defn_to_vec (data->obstackp,
5435 fixup_symbol_section (sym, data->objfile),
5436 block);
5437 }
5438 }
5439 return true;
5440 }
5441
5442 /* Helper for add_nonlocal_symbols. Find symbols in DOMAIN which are
5443 targeted by renamings matching LOOKUP_NAME in BLOCK. Add these
5444 symbols to OBSTACKP. Return whether we found such symbols. */
5445
5446 static int
5447 ada_add_block_renamings (struct obstack *obstackp,
5448 const struct block *block,
5449 const lookup_name_info &lookup_name,
5450 domain_enum domain)
5451 {
5452 struct using_direct *renaming;
5453 int defns_mark = num_defns_collected (obstackp);
5454
5455 symbol_name_matcher_ftype *name_match
5456 = ada_get_symbol_name_matcher (lookup_name);
5457
5458 for (renaming = block_using (block);
5459 renaming != NULL;
5460 renaming = renaming->next)
5461 {
5462 const char *r_name;
5463
5464 /* Avoid infinite recursions: skip this renaming if we are actually
5465 already traversing it.
5466
5467 Currently, symbol lookup in Ada don't use the namespace machinery from
5468 C++/Fortran support: skip namespace imports that use them. */
5469 if (renaming->searched
5470 || (renaming->import_src != NULL
5471 && renaming->import_src[0] != '\0')
5472 || (renaming->import_dest != NULL
5473 && renaming->import_dest[0] != '\0'))
5474 continue;
5475 renaming->searched = 1;
5476
5477 /* TODO: here, we perform another name-based symbol lookup, which can
5478 pull its own multiple overloads. In theory, we should be able to do
5479 better in this case since, in DWARF, DW_AT_import is a DIE reference,
5480 not a simple name. But in order to do this, we would need to enhance
5481 the DWARF reader to associate a symbol to this renaming, instead of a
5482 name. So, for now, we do something simpler: re-use the C++/Fortran
5483 namespace machinery. */
5484 r_name = (renaming->alias != NULL
5485 ? renaming->alias
5486 : renaming->declaration);
5487 if (name_match (r_name, lookup_name, NULL))
5488 {
5489 lookup_name_info decl_lookup_name (renaming->declaration,
5490 lookup_name.match_type ());
5491 ada_add_all_symbols (obstackp, block, decl_lookup_name, domain,
5492 1, NULL);
5493 }
5494 renaming->searched = 0;
5495 }
5496 return num_defns_collected (obstackp) != defns_mark;
5497 }
5498
5499 /* Implements compare_names, but only applying the comparision using
5500 the given CASING. */
5501
5502 static int
5503 compare_names_with_case (const char *string1, const char *string2,
5504 enum case_sensitivity casing)
5505 {
5506 while (*string1 != '\0' && *string2 != '\0')
5507 {
5508 char c1, c2;
5509
5510 if (isspace (*string1) || isspace (*string2))
5511 return strcmp_iw_ordered (string1, string2);
5512
5513 if (casing == case_sensitive_off)
5514 {
5515 c1 = tolower (*string1);
5516 c2 = tolower (*string2);
5517 }
5518 else
5519 {
5520 c1 = *string1;
5521 c2 = *string2;
5522 }
5523 if (c1 != c2)
5524 break;
5525
5526 string1 += 1;
5527 string2 += 1;
5528 }
5529
5530 switch (*string1)
5531 {
5532 case '(':
5533 return strcmp_iw_ordered (string1, string2);
5534 case '_':
5535 if (*string2 == '\0')
5536 {
5537 if (is_name_suffix (string1))
5538 return 0;
5539 else
5540 return 1;
5541 }
5542 /* FALLTHROUGH */
5543 default:
5544 if (*string2 == '(')
5545 return strcmp_iw_ordered (string1, string2);
5546 else
5547 {
5548 if (casing == case_sensitive_off)
5549 return tolower (*string1) - tolower (*string2);
5550 else
5551 return *string1 - *string2;
5552 }
5553 }
5554 }
5555
5556 /* Compare STRING1 to STRING2, with results as for strcmp.
5557 Compatible with strcmp_iw_ordered in that...
5558
5559 strcmp_iw_ordered (STRING1, STRING2) <= 0
5560
5561 ... implies...
5562
5563 compare_names (STRING1, STRING2) <= 0
5564
5565 (they may differ as to what symbols compare equal). */
5566
5567 static int
5568 compare_names (const char *string1, const char *string2)
5569 {
5570 int result;
5571
5572 /* Similar to what strcmp_iw_ordered does, we need to perform
5573 a case-insensitive comparison first, and only resort to
5574 a second, case-sensitive, comparison if the first one was
5575 not sufficient to differentiate the two strings. */
5576
5577 result = compare_names_with_case (string1, string2, case_sensitive_off);
5578 if (result == 0)
5579 result = compare_names_with_case (string1, string2, case_sensitive_on);
5580
5581 return result;
5582 }
5583
5584 /* Convenience function to get at the Ada encoded lookup name for
5585 LOOKUP_NAME, as a C string. */
5586
5587 static const char *
5588 ada_lookup_name (const lookup_name_info &lookup_name)
5589 {
5590 return lookup_name.ada ().lookup_name ().c_str ();
5591 }
5592
5593 /* Add to OBSTACKP all non-local symbols whose name and domain match
5594 LOOKUP_NAME and DOMAIN respectively. The search is performed on
5595 GLOBAL_BLOCK symbols if GLOBAL is non-zero, or on STATIC_BLOCK
5596 symbols otherwise. */
5597
5598 static void
5599 add_nonlocal_symbols (struct obstack *obstackp,
5600 const lookup_name_info &lookup_name,
5601 domain_enum domain, int global)
5602 {
5603 struct match_data data;
5604
5605 memset (&data, 0, sizeof data);
5606 data.obstackp = obstackp;
5607
5608 bool is_wild_match = lookup_name.ada ().wild_match_p ();
5609
5610 auto callback = [&] (struct block_symbol *bsym)
5611 {
5612 return aux_add_nonlocal_symbols (bsym, &data);
5613 };
5614
5615 for (objfile *objfile : current_program_space->objfiles ())
5616 {
5617 data.objfile = objfile;
5618
5619 objfile->sf->qf->map_matching_symbols (objfile, lookup_name,
5620 domain, global, callback,
5621 (is_wild_match
5622 ? NULL : compare_names));
5623
5624 for (compunit_symtab *cu : objfile->compunits ())
5625 {
5626 const struct block *global_block
5627 = BLOCKVECTOR_BLOCK (COMPUNIT_BLOCKVECTOR (cu), GLOBAL_BLOCK);
5628
5629 if (ada_add_block_renamings (obstackp, global_block, lookup_name,
5630 domain))
5631 data.found_sym = 1;
5632 }
5633 }
5634
5635 if (num_defns_collected (obstackp) == 0 && global && !is_wild_match)
5636 {
5637 const char *name = ada_lookup_name (lookup_name);
5638 std::string bracket_name = std::string ("<_ada_") + name + '>';
5639 lookup_name_info name1 (bracket_name, symbol_name_match_type::FULL);
5640
5641 for (objfile *objfile : current_program_space->objfiles ())
5642 {
5643 data.objfile = objfile;
5644 objfile->sf->qf->map_matching_symbols (objfile, name1,
5645 domain, global, callback,
5646 compare_names);
5647 }
5648 }
5649 }
5650
5651 /* Find symbols in DOMAIN matching LOOKUP_NAME, in BLOCK and, if
5652 FULL_SEARCH is non-zero, enclosing scope and in global scopes,
5653 returning the number of matches. Add these to OBSTACKP.
5654
5655 When FULL_SEARCH is non-zero, any non-function/non-enumeral
5656 symbol match within the nest of blocks whose innermost member is BLOCK,
5657 is the one match returned (no other matches in that or
5658 enclosing blocks is returned). If there are any matches in or
5659 surrounding BLOCK, then these alone are returned.
5660
5661 Names prefixed with "standard__" are handled specially:
5662 "standard__" is first stripped off (by the lookup_name
5663 constructor), and only static and global symbols are searched.
5664
5665 If MADE_GLOBAL_LOOKUP_P is non-null, set it before return to whether we had
5666 to lookup global symbols. */
5667
5668 static void
5669 ada_add_all_symbols (struct obstack *obstackp,
5670 const struct block *block,
5671 const lookup_name_info &lookup_name,
5672 domain_enum domain,
5673 int full_search,
5674 int *made_global_lookup_p)
5675 {
5676 struct symbol *sym;
5677
5678 if (made_global_lookup_p)
5679 *made_global_lookup_p = 0;
5680
5681 /* Special case: If the user specifies a symbol name inside package
5682 Standard, do a non-wild matching of the symbol name without
5683 the "standard__" prefix. This was primarily introduced in order
5684 to allow the user to specifically access the standard exceptions
5685 using, for instance, Standard.Constraint_Error when Constraint_Error
5686 is ambiguous (due to the user defining its own Constraint_Error
5687 entity inside its program). */
5688 if (lookup_name.ada ().standard_p ())
5689 block = NULL;
5690
5691 /* Check the non-global symbols. If we have ANY match, then we're done. */
5692
5693 if (block != NULL)
5694 {
5695 if (full_search)
5696 ada_add_local_symbols (obstackp, lookup_name, block, domain);
5697 else
5698 {
5699 /* In the !full_search case we're are being called by
5700 ada_iterate_over_symbols, and we don't want to search
5701 superblocks. */
5702 ada_add_block_symbols (obstackp, block, lookup_name, domain, NULL);
5703 }
5704 if (num_defns_collected (obstackp) > 0 || !full_search)
5705 return;
5706 }
5707
5708 /* No non-global symbols found. Check our cache to see if we have
5709 already performed this search before. If we have, then return
5710 the same result. */
5711
5712 if (lookup_cached_symbol (ada_lookup_name (lookup_name),
5713 domain, &sym, &block))
5714 {
5715 if (sym != NULL)
5716 add_defn_to_vec (obstackp, sym, block);
5717 return;
5718 }
5719
5720 if (made_global_lookup_p)
5721 *made_global_lookup_p = 1;
5722
5723 /* Search symbols from all global blocks. */
5724
5725 add_nonlocal_symbols (obstackp, lookup_name, domain, 1);
5726
5727 /* Now add symbols from all per-file blocks if we've gotten no hits
5728 (not strictly correct, but perhaps better than an error). */
5729
5730 if (num_defns_collected (obstackp) == 0)
5731 add_nonlocal_symbols (obstackp, lookup_name, domain, 0);
5732 }
5733
5734 /* Find symbols in DOMAIN matching LOOKUP_NAME, in BLOCK and, if FULL_SEARCH
5735 is non-zero, enclosing scope and in global scopes, returning the number of
5736 matches.
5737 Fills *RESULTS with (SYM,BLOCK) tuples, indicating the symbols
5738 found and the blocks and symbol tables (if any) in which they were
5739 found.
5740
5741 When full_search is non-zero, any non-function/non-enumeral
5742 symbol match within the nest of blocks whose innermost member is BLOCK,
5743 is the one match returned (no other matches in that or
5744 enclosing blocks is returned). If there are any matches in or
5745 surrounding BLOCK, then these alone are returned.
5746
5747 Names prefixed with "standard__" are handled specially: "standard__"
5748 is first stripped off, and only static and global symbols are searched. */
5749
5750 static int
5751 ada_lookup_symbol_list_worker (const lookup_name_info &lookup_name,
5752 const struct block *block,
5753 domain_enum domain,
5754 std::vector<struct block_symbol> *results,
5755 int full_search)
5756 {
5757 int syms_from_global_search;
5758 int ndefns;
5759 auto_obstack obstack;
5760
5761 ada_add_all_symbols (&obstack, block, lookup_name,
5762 domain, full_search, &syms_from_global_search);
5763
5764 ndefns = num_defns_collected (&obstack);
5765
5766 struct block_symbol *base = defns_collected (&obstack, 1);
5767 for (int i = 0; i < ndefns; ++i)
5768 results->push_back (base[i]);
5769
5770 ndefns = remove_extra_symbols (results);
5771
5772 if (ndefns == 0 && full_search && syms_from_global_search)
5773 cache_symbol (ada_lookup_name (lookup_name), domain, NULL, NULL);
5774
5775 if (ndefns == 1 && full_search && syms_from_global_search)
5776 cache_symbol (ada_lookup_name (lookup_name), domain,
5777 (*results)[0].symbol, (*results)[0].block);
5778
5779 ndefns = remove_irrelevant_renamings (results, block);
5780
5781 return ndefns;
5782 }
5783
5784 /* Find symbols in DOMAIN matching NAME, in BLOCK and enclosing scope and
5785 in global scopes, returning the number of matches, and filling *RESULTS
5786 with (SYM,BLOCK) tuples.
5787
5788 See ada_lookup_symbol_list_worker for further details. */
5789
5790 int
5791 ada_lookup_symbol_list (const char *name, const struct block *block,
5792 domain_enum domain,
5793 std::vector<struct block_symbol> *results)
5794 {
5795 symbol_name_match_type name_match_type = name_match_type_from_name (name);
5796 lookup_name_info lookup_name (name, name_match_type);
5797
5798 return ada_lookup_symbol_list_worker (lookup_name, block, domain, results, 1);
5799 }
5800
5801 /* Implementation of the la_iterate_over_symbols method. */
5802
5803 static bool
5804 ada_iterate_over_symbols
5805 (const struct block *block, const lookup_name_info &name,
5806 domain_enum domain,
5807 gdb::function_view<symbol_found_callback_ftype> callback)
5808 {
5809 int ndefs, i;
5810 std::vector<struct block_symbol> results;
5811
5812 ndefs = ada_lookup_symbol_list_worker (name, block, domain, &results, 0);
5813
5814 for (i = 0; i < ndefs; ++i)
5815 {
5816 if (!callback (&results[i]))
5817 return false;
5818 }
5819
5820 return true;
5821 }
5822
5823 /* The result is as for ada_lookup_symbol_list with FULL_SEARCH set
5824 to 1, but choosing the first symbol found if there are multiple
5825 choices.
5826
5827 The result is stored in *INFO, which must be non-NULL.
5828 If no match is found, INFO->SYM is set to NULL. */
5829
5830 void
5831 ada_lookup_encoded_symbol (const char *name, const struct block *block,
5832 domain_enum domain,
5833 struct block_symbol *info)
5834 {
5835 /* Since we already have an encoded name, wrap it in '<>' to force a
5836 verbatim match. Otherwise, if the name happens to not look like
5837 an encoded name (because it doesn't include a "__"),
5838 ada_lookup_name_info would re-encode/fold it again, and that
5839 would e.g., incorrectly lowercase object renaming names like
5840 "R28b" -> "r28b". */
5841 std::string verbatim = std::string ("<") + name + '>';
5842
5843 gdb_assert (info != NULL);
5844 *info = ada_lookup_symbol (verbatim.c_str (), block, domain);
5845 }
5846
5847 /* Return a symbol in DOMAIN matching NAME, in BLOCK0 and enclosing
5848 scope and in global scopes, or NULL if none. NAME is folded and
5849 encoded first. Otherwise, the result is as for ada_lookup_symbol_list,
5850 choosing the first symbol if there are multiple choices. */
5851
5852 struct block_symbol
5853 ada_lookup_symbol (const char *name, const struct block *block0,
5854 domain_enum domain)
5855 {
5856 std::vector<struct block_symbol> candidates;
5857 int n_candidates;
5858
5859 n_candidates = ada_lookup_symbol_list (name, block0, domain, &candidates);
5860
5861 if (n_candidates == 0)
5862 return {};
5863
5864 block_symbol info = candidates[0];
5865 info.symbol = fixup_symbol_section (info.symbol, NULL);
5866 return info;
5867 }
5868
5869 static struct block_symbol
5870 ada_lookup_symbol_nonlocal (const struct language_defn *langdef,
5871 const char *name,
5872 const struct block *block,
5873 const domain_enum domain)
5874 {
5875 struct block_symbol sym;
5876
5877 sym = ada_lookup_symbol (name, block_static_block (block), domain);
5878 if (sym.symbol != NULL)
5879 return sym;
5880
5881 /* If we haven't found a match at this point, try the primitive
5882 types. In other languages, this search is performed before
5883 searching for global symbols in order to short-circuit that
5884 global-symbol search if it happens that the name corresponds
5885 to a primitive type. But we cannot do the same in Ada, because
5886 it is perfectly legitimate for a program to declare a type which
5887 has the same name as a standard type. If looking up a type in
5888 that situation, we have traditionally ignored the primitive type
5889 in favor of user-defined types. This is why, unlike most other
5890 languages, we search the primitive types this late and only after
5891 having searched the global symbols without success. */
5892
5893 if (domain == VAR_DOMAIN)
5894 {
5895 struct gdbarch *gdbarch;
5896
5897 if (block == NULL)
5898 gdbarch = target_gdbarch ();
5899 else
5900 gdbarch = block_gdbarch (block);
5901 sym.symbol = language_lookup_primitive_type_as_symbol (langdef, gdbarch, name);
5902 if (sym.symbol != NULL)
5903 return sym;
5904 }
5905
5906 return {};
5907 }
5908
5909
5910 /* True iff STR is a possible encoded suffix of a normal Ada name
5911 that is to be ignored for matching purposes. Suffixes of parallel
5912 names (e.g., XVE) are not included here. Currently, the possible suffixes
5913 are given by any of the regular expressions:
5914
5915 [.$][0-9]+ [nested subprogram suffix, on platforms such as GNU/Linux]
5916 ___[0-9]+ [nested subprogram suffix, on platforms such as HP/UX]
5917 TKB [subprogram suffix for task bodies]
5918 _E[0-9]+[bs]$ [protected object entry suffixes]
5919 (X[nb]*)?((\$|__)[0-9](_?[0-9]+)|___(JM|LJM|X([FDBUP].*|R[^T]?)))?$
5920
5921 Also, any leading "__[0-9]+" sequence is skipped before the suffix
5922 match is performed. This sequence is used to differentiate homonyms,
5923 is an optional part of a valid name suffix. */
5924
5925 static int
5926 is_name_suffix (const char *str)
5927 {
5928 int k;
5929 const char *matching;
5930 const int len = strlen (str);
5931
5932 /* Skip optional leading __[0-9]+. */
5933
5934 if (len > 3 && str[0] == '_' && str[1] == '_' && isdigit (str[2]))
5935 {
5936 str += 3;
5937 while (isdigit (str[0]))
5938 str += 1;
5939 }
5940
5941 /* [.$][0-9]+ */
5942
5943 if (str[0] == '.' || str[0] == '$')
5944 {
5945 matching = str + 1;
5946 while (isdigit (matching[0]))
5947 matching += 1;
5948 if (matching[0] == '\0')
5949 return 1;
5950 }
5951
5952 /* ___[0-9]+ */
5953
5954 if (len > 3 && str[0] == '_' && str[1] == '_' && str[2] == '_')
5955 {
5956 matching = str + 3;
5957 while (isdigit (matching[0]))
5958 matching += 1;
5959 if (matching[0] == '\0')
5960 return 1;
5961 }
5962
5963 /* "TKB" suffixes are used for subprograms implementing task bodies. */
5964
5965 if (strcmp (str, "TKB") == 0)
5966 return 1;
5967
5968 #if 0
5969 /* FIXME: brobecker/2005-09-23: Protected Object subprograms end
5970 with a N at the end. Unfortunately, the compiler uses the same
5971 convention for other internal types it creates. So treating
5972 all entity names that end with an "N" as a name suffix causes
5973 some regressions. For instance, consider the case of an enumerated
5974 type. To support the 'Image attribute, it creates an array whose
5975 name ends with N.
5976 Having a single character like this as a suffix carrying some
5977 information is a bit risky. Perhaps we should change the encoding
5978 to be something like "_N" instead. In the meantime, do not do
5979 the following check. */
5980 /* Protected Object Subprograms */
5981 if (len == 1 && str [0] == 'N')
5982 return 1;
5983 #endif
5984
5985 /* _E[0-9]+[bs]$ */
5986 if (len > 3 && str[0] == '_' && str [1] == 'E' && isdigit (str[2]))
5987 {
5988 matching = str + 3;
5989 while (isdigit (matching[0]))
5990 matching += 1;
5991 if ((matching[0] == 'b' || matching[0] == 's')
5992 && matching [1] == '\0')
5993 return 1;
5994 }
5995
5996 /* ??? We should not modify STR directly, as we are doing below. This
5997 is fine in this case, but may become problematic later if we find
5998 that this alternative did not work, and want to try matching
5999 another one from the begining of STR. Since we modified it, we
6000 won't be able to find the begining of the string anymore! */
6001 if (str[0] == 'X')
6002 {
6003 str += 1;
6004 while (str[0] != '_' && str[0] != '\0')
6005 {
6006 if (str[0] != 'n' && str[0] != 'b')
6007 return 0;
6008 str += 1;
6009 }
6010 }
6011
6012 if (str[0] == '\000')
6013 return 1;
6014
6015 if (str[0] == '_')
6016 {
6017 if (str[1] != '_' || str[2] == '\000')
6018 return 0;
6019 if (str[2] == '_')
6020 {
6021 if (strcmp (str + 3, "JM") == 0)
6022 return 1;
6023 /* FIXME: brobecker/2004-09-30: GNAT will soon stop using
6024 the LJM suffix in favor of the JM one. But we will
6025 still accept LJM as a valid suffix for a reasonable
6026 amount of time, just to allow ourselves to debug programs
6027 compiled using an older version of GNAT. */
6028 if (strcmp (str + 3, "LJM") == 0)
6029 return 1;
6030 if (str[3] != 'X')
6031 return 0;
6032 if (str[4] == 'F' || str[4] == 'D' || str[4] == 'B'
6033 || str[4] == 'U' || str[4] == 'P')
6034 return 1;
6035 if (str[4] == 'R' && str[5] != 'T')
6036 return 1;
6037 return 0;
6038 }
6039 if (!isdigit (str[2]))
6040 return 0;
6041 for (k = 3; str[k] != '\0'; k += 1)
6042 if (!isdigit (str[k]) && str[k] != '_')
6043 return 0;
6044 return 1;
6045 }
6046 if (str[0] == '$' && isdigit (str[1]))
6047 {
6048 for (k = 2; str[k] != '\0'; k += 1)
6049 if (!isdigit (str[k]) && str[k] != '_')
6050 return 0;
6051 return 1;
6052 }
6053 return 0;
6054 }
6055
6056 /* Return non-zero if the string starting at NAME and ending before
6057 NAME_END contains no capital letters. */
6058
6059 static int
6060 is_valid_name_for_wild_match (const char *name0)
6061 {
6062 std::string decoded_name = ada_decode (name0);
6063 int i;
6064
6065 /* If the decoded name starts with an angle bracket, it means that
6066 NAME0 does not follow the GNAT encoding format. It should then
6067 not be allowed as a possible wild match. */
6068 if (decoded_name[0] == '<')
6069 return 0;
6070
6071 for (i=0; decoded_name[i] != '\0'; i++)
6072 if (isalpha (decoded_name[i]) && !islower (decoded_name[i]))
6073 return 0;
6074
6075 return 1;
6076 }
6077
6078 /* Advance *NAMEP to next occurrence of TARGET0 in the string NAME0
6079 that could start a simple name. Assumes that *NAMEP points into
6080 the string beginning at NAME0. */
6081
6082 static int
6083 advance_wild_match (const char **namep, const char *name0, int target0)
6084 {
6085 const char *name = *namep;
6086
6087 while (1)
6088 {
6089 int t0, t1;
6090
6091 t0 = *name;
6092 if (t0 == '_')
6093 {
6094 t1 = name[1];
6095 if ((t1 >= 'a' && t1 <= 'z') || (t1 >= '0' && t1 <= '9'))
6096 {
6097 name += 1;
6098 if (name == name0 + 5 && startswith (name0, "_ada"))
6099 break;
6100 else
6101 name += 1;
6102 }
6103 else if (t1 == '_' && ((name[2] >= 'a' && name[2] <= 'z')
6104 || name[2] == target0))
6105 {
6106 name += 2;
6107 break;
6108 }
6109 else
6110 return 0;
6111 }
6112 else if ((t0 >= 'a' && t0 <= 'z') || (t0 >= '0' && t0 <= '9'))
6113 name += 1;
6114 else
6115 return 0;
6116 }
6117
6118 *namep = name;
6119 return 1;
6120 }
6121
6122 /* Return true iff NAME encodes a name of the form prefix.PATN.
6123 Ignores any informational suffixes of NAME (i.e., for which
6124 is_name_suffix is true). Assumes that PATN is a lower-cased Ada
6125 simple name. */
6126
6127 static bool
6128 wild_match (const char *name, const char *patn)
6129 {
6130 const char *p;
6131 const char *name0 = name;
6132
6133 while (1)
6134 {
6135 const char *match = name;
6136
6137 if (*name == *patn)
6138 {
6139 for (name += 1, p = patn + 1; *p != '\0'; name += 1, p += 1)
6140 if (*p != *name)
6141 break;
6142 if (*p == '\0' && is_name_suffix (name))
6143 return match == name0 || is_valid_name_for_wild_match (name0);
6144
6145 if (name[-1] == '_')
6146 name -= 1;
6147 }
6148 if (!advance_wild_match (&name, name0, *patn))
6149 return false;
6150 }
6151 }
6152
6153 /* Returns true iff symbol name SYM_NAME matches SEARCH_NAME, ignoring
6154 any trailing suffixes that encode debugging information or leading
6155 _ada_ on SYM_NAME (see is_name_suffix commentary for the debugging
6156 information that is ignored). */
6157
6158 static bool
6159 full_match (const char *sym_name, const char *search_name)
6160 {
6161 size_t search_name_len = strlen (search_name);
6162
6163 if (strncmp (sym_name, search_name, search_name_len) == 0
6164 && is_name_suffix (sym_name + search_name_len))
6165 return true;
6166
6167 if (startswith (sym_name, "_ada_")
6168 && strncmp (sym_name + 5, search_name, search_name_len) == 0
6169 && is_name_suffix (sym_name + search_name_len + 5))
6170 return true;
6171
6172 return false;
6173 }
6174
6175 /* Add symbols from BLOCK matching LOOKUP_NAME in DOMAIN to vector
6176 *defn_symbols, updating the list of symbols in OBSTACKP (if
6177 necessary). OBJFILE is the section containing BLOCK. */
6178
6179 static void
6180 ada_add_block_symbols (struct obstack *obstackp,
6181 const struct block *block,
6182 const lookup_name_info &lookup_name,
6183 domain_enum domain, struct objfile *objfile)
6184 {
6185 struct block_iterator iter;
6186 /* A matching argument symbol, if any. */
6187 struct symbol *arg_sym;
6188 /* Set true when we find a matching non-argument symbol. */
6189 int found_sym;
6190 struct symbol *sym;
6191
6192 arg_sym = NULL;
6193 found_sym = 0;
6194 for (sym = block_iter_match_first (block, lookup_name, &iter);
6195 sym != NULL;
6196 sym = block_iter_match_next (lookup_name, &iter))
6197 {
6198 if (symbol_matches_domain (sym->language (), SYMBOL_DOMAIN (sym), domain))
6199 {
6200 if (SYMBOL_CLASS (sym) != LOC_UNRESOLVED)
6201 {
6202 if (SYMBOL_IS_ARGUMENT (sym))
6203 arg_sym = sym;
6204 else
6205 {
6206 found_sym = 1;
6207 add_defn_to_vec (obstackp,
6208 fixup_symbol_section (sym, objfile),
6209 block);
6210 }
6211 }
6212 }
6213 }
6214
6215 /* Handle renamings. */
6216
6217 if (ada_add_block_renamings (obstackp, block, lookup_name, domain))
6218 found_sym = 1;
6219
6220 if (!found_sym && arg_sym != NULL)
6221 {
6222 add_defn_to_vec (obstackp,
6223 fixup_symbol_section (arg_sym, objfile),
6224 block);
6225 }
6226
6227 if (!lookup_name.ada ().wild_match_p ())
6228 {
6229 arg_sym = NULL;
6230 found_sym = 0;
6231 const std::string &ada_lookup_name = lookup_name.ada ().lookup_name ();
6232 const char *name = ada_lookup_name.c_str ();
6233 size_t name_len = ada_lookup_name.size ();
6234
6235 ALL_BLOCK_SYMBOLS (block, iter, sym)
6236 {
6237 if (symbol_matches_domain (sym->language (),
6238 SYMBOL_DOMAIN (sym), domain))
6239 {
6240 int cmp;
6241
6242 cmp = (int) '_' - (int) sym->linkage_name ()[0];
6243 if (cmp == 0)
6244 {
6245 cmp = !startswith (sym->linkage_name (), "_ada_");
6246 if (cmp == 0)
6247 cmp = strncmp (name, sym->linkage_name () + 5,
6248 name_len);
6249 }
6250
6251 if (cmp == 0
6252 && is_name_suffix (sym->linkage_name () + name_len + 5))
6253 {
6254 if (SYMBOL_CLASS (sym) != LOC_UNRESOLVED)
6255 {
6256 if (SYMBOL_IS_ARGUMENT (sym))
6257 arg_sym = sym;
6258 else
6259 {
6260 found_sym = 1;
6261 add_defn_to_vec (obstackp,
6262 fixup_symbol_section (sym, objfile),
6263 block);
6264 }
6265 }
6266 }
6267 }
6268 }
6269
6270 /* NOTE: This really shouldn't be needed for _ada_ symbols.
6271 They aren't parameters, right? */
6272 if (!found_sym && arg_sym != NULL)
6273 {
6274 add_defn_to_vec (obstackp,
6275 fixup_symbol_section (arg_sym, objfile),
6276 block);
6277 }
6278 }
6279 }
6280 \f
6281
6282 /* Symbol Completion */
6283
6284 /* See symtab.h. */
6285
6286 bool
6287 ada_lookup_name_info::matches
6288 (const char *sym_name,
6289 symbol_name_match_type match_type,
6290 completion_match_result *comp_match_res) const
6291 {
6292 bool match = false;
6293 const char *text = m_encoded_name.c_str ();
6294 size_t text_len = m_encoded_name.size ();
6295
6296 /* First, test against the fully qualified name of the symbol. */
6297
6298 if (strncmp (sym_name, text, text_len) == 0)
6299 match = true;
6300
6301 std::string decoded_name = ada_decode (sym_name);
6302 if (match && !m_encoded_p)
6303 {
6304 /* One needed check before declaring a positive match is to verify
6305 that iff we are doing a verbatim match, the decoded version
6306 of the symbol name starts with '<'. Otherwise, this symbol name
6307 is not a suitable completion. */
6308
6309 bool has_angle_bracket = (decoded_name[0] == '<');
6310 match = (has_angle_bracket == m_verbatim_p);
6311 }
6312
6313 if (match && !m_verbatim_p)
6314 {
6315 /* When doing non-verbatim match, another check that needs to
6316 be done is to verify that the potentially matching symbol name
6317 does not include capital letters, because the ada-mode would
6318 not be able to understand these symbol names without the
6319 angle bracket notation. */
6320 const char *tmp;
6321
6322 for (tmp = sym_name; *tmp != '\0' && !isupper (*tmp); tmp++);
6323 if (*tmp != '\0')
6324 match = false;
6325 }
6326
6327 /* Second: Try wild matching... */
6328
6329 if (!match && m_wild_match_p)
6330 {
6331 /* Since we are doing wild matching, this means that TEXT
6332 may represent an unqualified symbol name. We therefore must
6333 also compare TEXT against the unqualified name of the symbol. */
6334 sym_name = ada_unqualified_name (decoded_name.c_str ());
6335
6336 if (strncmp (sym_name, text, text_len) == 0)
6337 match = true;
6338 }
6339
6340 /* Finally: If we found a match, prepare the result to return. */
6341
6342 if (!match)
6343 return false;
6344
6345 if (comp_match_res != NULL)
6346 {
6347 std::string &match_str = comp_match_res->match.storage ();
6348
6349 if (!m_encoded_p)
6350 match_str = ada_decode (sym_name);
6351 else
6352 {
6353 if (m_verbatim_p)
6354 match_str = add_angle_brackets (sym_name);
6355 else
6356 match_str = sym_name;
6357
6358 }
6359
6360 comp_match_res->set_match (match_str.c_str ());
6361 }
6362
6363 return true;
6364 }
6365
6366 /* Add the list of possible symbol names completing TEXT to TRACKER.
6367 WORD is the entire command on which completion is made. */
6368
6369 static void
6370 ada_collect_symbol_completion_matches (completion_tracker &tracker,
6371 complete_symbol_mode mode,
6372 symbol_name_match_type name_match_type,
6373 const char *text, const char *word,
6374 enum type_code code)
6375 {
6376 struct symbol *sym;
6377 const struct block *b, *surrounding_static_block = 0;
6378 struct block_iterator iter;
6379
6380 gdb_assert (code == TYPE_CODE_UNDEF);
6381
6382 lookup_name_info lookup_name (text, name_match_type, true);
6383
6384 /* First, look at the partial symtab symbols. */
6385 expand_symtabs_matching (NULL,
6386 lookup_name,
6387 NULL,
6388 NULL,
6389 ALL_DOMAIN);
6390
6391 /* At this point scan through the misc symbol vectors and add each
6392 symbol you find to the list. Eventually we want to ignore
6393 anything that isn't a text symbol (everything else will be
6394 handled by the psymtab code above). */
6395
6396 for (objfile *objfile : current_program_space->objfiles ())
6397 {
6398 for (minimal_symbol *msymbol : objfile->msymbols ())
6399 {
6400 QUIT;
6401
6402 if (completion_skip_symbol (mode, msymbol))
6403 continue;
6404
6405 language symbol_language = msymbol->language ();
6406
6407 /* Ada minimal symbols won't have their language set to Ada. If
6408 we let completion_list_add_name compare using the
6409 default/C-like matcher, then when completing e.g., symbols in a
6410 package named "pck", we'd match internal Ada symbols like
6411 "pckS", which are invalid in an Ada expression, unless you wrap
6412 them in '<' '>' to request a verbatim match.
6413
6414 Unfortunately, some Ada encoded names successfully demangle as
6415 C++ symbols (using an old mangling scheme), such as "name__2Xn"
6416 -> "Xn::name(void)" and thus some Ada minimal symbols end up
6417 with the wrong language set. Paper over that issue here. */
6418 if (symbol_language == language_auto
6419 || symbol_language == language_cplus)
6420 symbol_language = language_ada;
6421
6422 completion_list_add_name (tracker,
6423 symbol_language,
6424 msymbol->linkage_name (),
6425 lookup_name, text, word);
6426 }
6427 }
6428
6429 /* Search upwards from currently selected frame (so that we can
6430 complete on local vars. */
6431
6432 for (b = get_selected_block (0); b != NULL; b = BLOCK_SUPERBLOCK (b))
6433 {
6434 if (!BLOCK_SUPERBLOCK (b))
6435 surrounding_static_block = b; /* For elmin of dups */
6436
6437 ALL_BLOCK_SYMBOLS (b, iter, sym)
6438 {
6439 if (completion_skip_symbol (mode, sym))
6440 continue;
6441
6442 completion_list_add_name (tracker,
6443 sym->language (),
6444 sym->linkage_name (),
6445 lookup_name, text, word);
6446 }
6447 }
6448
6449 /* Go through the symtabs and check the externs and statics for
6450 symbols which match. */
6451
6452 for (objfile *objfile : current_program_space->objfiles ())
6453 {
6454 for (compunit_symtab *s : objfile->compunits ())
6455 {
6456 QUIT;
6457 b = BLOCKVECTOR_BLOCK (COMPUNIT_BLOCKVECTOR (s), GLOBAL_BLOCK);
6458 ALL_BLOCK_SYMBOLS (b, iter, sym)
6459 {
6460 if (completion_skip_symbol (mode, sym))
6461 continue;
6462
6463 completion_list_add_name (tracker,
6464 sym->language (),
6465 sym->linkage_name (),
6466 lookup_name, text, word);
6467 }
6468 }
6469 }
6470
6471 for (objfile *objfile : current_program_space->objfiles ())
6472 {
6473 for (compunit_symtab *s : objfile->compunits ())
6474 {
6475 QUIT;
6476 b = BLOCKVECTOR_BLOCK (COMPUNIT_BLOCKVECTOR (s), STATIC_BLOCK);
6477 /* Don't do this block twice. */
6478 if (b == surrounding_static_block)
6479 continue;
6480 ALL_BLOCK_SYMBOLS (b, iter, sym)
6481 {
6482 if (completion_skip_symbol (mode, sym))
6483 continue;
6484
6485 completion_list_add_name (tracker,
6486 sym->language (),
6487 sym->linkage_name (),
6488 lookup_name, text, word);
6489 }
6490 }
6491 }
6492 }
6493
6494 /* Field Access */
6495
6496 /* Return non-zero if TYPE is a pointer to the GNAT dispatch table used
6497 for tagged types. */
6498
6499 static int
6500 ada_is_dispatch_table_ptr_type (struct type *type)
6501 {
6502 const char *name;
6503
6504 if (type->code () != TYPE_CODE_PTR)
6505 return 0;
6506
6507 name = TYPE_TARGET_TYPE (type)->name ();
6508 if (name == NULL)
6509 return 0;
6510
6511 return (strcmp (name, "ada__tags__dispatch_table") == 0);
6512 }
6513
6514 /* Return non-zero if TYPE is an interface tag. */
6515
6516 static int
6517 ada_is_interface_tag (struct type *type)
6518 {
6519 const char *name = type->name ();
6520
6521 if (name == NULL)
6522 return 0;
6523
6524 return (strcmp (name, "ada__tags__interface_tag") == 0);
6525 }
6526
6527 /* True if field number FIELD_NUM in struct or union type TYPE is supposed
6528 to be invisible to users. */
6529
6530 int
6531 ada_is_ignored_field (struct type *type, int field_num)
6532 {
6533 if (field_num < 0 || field_num > type->num_fields ())
6534 return 1;
6535
6536 /* Check the name of that field. */
6537 {
6538 const char *name = TYPE_FIELD_NAME (type, field_num);
6539
6540 /* Anonymous field names should not be printed.
6541 brobecker/2007-02-20: I don't think this can actually happen
6542 but we don't want to print the value of anonymous fields anyway. */
6543 if (name == NULL)
6544 return 1;
6545
6546 /* Normally, fields whose name start with an underscore ("_")
6547 are fields that have been internally generated by the compiler,
6548 and thus should not be printed. The "_parent" field is special,
6549 however: This is a field internally generated by the compiler
6550 for tagged types, and it contains the components inherited from
6551 the parent type. This field should not be printed as is, but
6552 should not be ignored either. */
6553 if (name[0] == '_' && !startswith (name, "_parent"))
6554 return 1;
6555 }
6556
6557 /* If this is the dispatch table of a tagged type or an interface tag,
6558 then ignore. */
6559 if (ada_is_tagged_type (type, 1)
6560 && (ada_is_dispatch_table_ptr_type (TYPE_FIELD_TYPE (type, field_num))
6561 || ada_is_interface_tag (TYPE_FIELD_TYPE (type, field_num))))
6562 return 1;
6563
6564 /* Not a special field, so it should not be ignored. */
6565 return 0;
6566 }
6567
6568 /* True iff TYPE has a tag field. If REFOK, then TYPE may also be a
6569 pointer or reference type whose ultimate target has a tag field. */
6570
6571 int
6572 ada_is_tagged_type (struct type *type, int refok)
6573 {
6574 return (ada_lookup_struct_elt_type (type, "_tag", refok, 1) != NULL);
6575 }
6576
6577 /* True iff TYPE represents the type of X'Tag */
6578
6579 int
6580 ada_is_tag_type (struct type *type)
6581 {
6582 type = ada_check_typedef (type);
6583
6584 if (type == NULL || type->code () != TYPE_CODE_PTR)
6585 return 0;
6586 else
6587 {
6588 const char *name = ada_type_name (TYPE_TARGET_TYPE (type));
6589
6590 return (name != NULL
6591 && strcmp (name, "ada__tags__dispatch_table") == 0);
6592 }
6593 }
6594
6595 /* The type of the tag on VAL. */
6596
6597 static struct type *
6598 ada_tag_type (struct value *val)
6599 {
6600 return ada_lookup_struct_elt_type (value_type (val), "_tag", 1, 0);
6601 }
6602
6603 /* Return 1 if TAG follows the old scheme for Ada tags (used for Ada 95,
6604 retired at Ada 05). */
6605
6606 static int
6607 is_ada95_tag (struct value *tag)
6608 {
6609 return ada_value_struct_elt (tag, "tsd", 1) != NULL;
6610 }
6611
6612 /* The value of the tag on VAL. */
6613
6614 static struct value *
6615 ada_value_tag (struct value *val)
6616 {
6617 return ada_value_struct_elt (val, "_tag", 0);
6618 }
6619
6620 /* The value of the tag on the object of type TYPE whose contents are
6621 saved at VALADDR, if it is non-null, or is at memory address
6622 ADDRESS. */
6623
6624 static struct value *
6625 value_tag_from_contents_and_address (struct type *type,
6626 const gdb_byte *valaddr,
6627 CORE_ADDR address)
6628 {
6629 int tag_byte_offset;
6630 struct type *tag_type;
6631
6632 if (find_struct_field ("_tag", type, 0, &tag_type, &tag_byte_offset,
6633 NULL, NULL, NULL))
6634 {
6635 const gdb_byte *valaddr1 = ((valaddr == NULL)
6636 ? NULL
6637 : valaddr + tag_byte_offset);
6638 CORE_ADDR address1 = (address == 0) ? 0 : address + tag_byte_offset;
6639
6640 return value_from_contents_and_address (tag_type, valaddr1, address1);
6641 }
6642 return NULL;
6643 }
6644
6645 static struct type *
6646 type_from_tag (struct value *tag)
6647 {
6648 const char *type_name = ada_tag_name (tag);
6649
6650 if (type_name != NULL)
6651 return ada_find_any_type (ada_encode (type_name));
6652 return NULL;
6653 }
6654
6655 /* Given a value OBJ of a tagged type, return a value of this
6656 type at the base address of the object. The base address, as
6657 defined in Ada.Tags, it is the address of the primary tag of
6658 the object, and therefore where the field values of its full
6659 view can be fetched. */
6660
6661 struct value *
6662 ada_tag_value_at_base_address (struct value *obj)
6663 {
6664 struct value *val;
6665 LONGEST offset_to_top = 0;
6666 struct type *ptr_type, *obj_type;
6667 struct value *tag;
6668 CORE_ADDR base_address;
6669
6670 obj_type = value_type (obj);
6671
6672 /* It is the responsability of the caller to deref pointers. */
6673
6674 if (obj_type->code () == TYPE_CODE_PTR || obj_type->code () == TYPE_CODE_REF)
6675 return obj;
6676
6677 tag = ada_value_tag (obj);
6678 if (!tag)
6679 return obj;
6680
6681 /* Base addresses only appeared with Ada 05 and multiple inheritance. */
6682
6683 if (is_ada95_tag (tag))
6684 return obj;
6685
6686 ptr_type = language_lookup_primitive_type
6687 (language_def (language_ada), target_gdbarch(), "storage_offset");
6688 ptr_type = lookup_pointer_type (ptr_type);
6689 val = value_cast (ptr_type, tag);
6690 if (!val)
6691 return obj;
6692
6693 /* It is perfectly possible that an exception be raised while
6694 trying to determine the base address, just like for the tag;
6695 see ada_tag_name for more details. We do not print the error
6696 message for the same reason. */
6697
6698 try
6699 {
6700 offset_to_top = value_as_long (value_ind (value_ptradd (val, -2)));
6701 }
6702
6703 catch (const gdb_exception_error &e)
6704 {
6705 return obj;
6706 }
6707
6708 /* If offset is null, nothing to do. */
6709
6710 if (offset_to_top == 0)
6711 return obj;
6712
6713 /* -1 is a special case in Ada.Tags; however, what should be done
6714 is not quite clear from the documentation. So do nothing for
6715 now. */
6716
6717 if (offset_to_top == -1)
6718 return obj;
6719
6720 /* OFFSET_TO_TOP used to be a positive value to be subtracted
6721 from the base address. This was however incompatible with
6722 C++ dispatch table: C++ uses a *negative* value to *add*
6723 to the base address. Ada's convention has therefore been
6724 changed in GNAT 19.0w 20171023: since then, C++ and Ada
6725 use the same convention. Here, we support both cases by
6726 checking the sign of OFFSET_TO_TOP. */
6727
6728 if (offset_to_top > 0)
6729 offset_to_top = -offset_to_top;
6730
6731 base_address = value_address (obj) + offset_to_top;
6732 tag = value_tag_from_contents_and_address (obj_type, NULL, base_address);
6733
6734 /* Make sure that we have a proper tag at the new address.
6735 Otherwise, offset_to_top is bogus (which can happen when
6736 the object is not initialized yet). */
6737
6738 if (!tag)
6739 return obj;
6740
6741 obj_type = type_from_tag (tag);
6742
6743 if (!obj_type)
6744 return obj;
6745
6746 return value_from_contents_and_address (obj_type, NULL, base_address);
6747 }
6748
6749 /* Return the "ada__tags__type_specific_data" type. */
6750
6751 static struct type *
6752 ada_get_tsd_type (struct inferior *inf)
6753 {
6754 struct ada_inferior_data *data = get_ada_inferior_data (inf);
6755
6756 if (data->tsd_type == 0)
6757 data->tsd_type = ada_find_any_type ("ada__tags__type_specific_data");
6758 return data->tsd_type;
6759 }
6760
6761 /* Return the TSD (type-specific data) associated to the given TAG.
6762 TAG is assumed to be the tag of a tagged-type entity.
6763
6764 May return NULL if we are unable to get the TSD. */
6765
6766 static struct value *
6767 ada_get_tsd_from_tag (struct value *tag)
6768 {
6769 struct value *val;
6770 struct type *type;
6771
6772 /* First option: The TSD is simply stored as a field of our TAG.
6773 Only older versions of GNAT would use this format, but we have
6774 to test it first, because there are no visible markers for
6775 the current approach except the absence of that field. */
6776
6777 val = ada_value_struct_elt (tag, "tsd", 1);
6778 if (val)
6779 return val;
6780
6781 /* Try the second representation for the dispatch table (in which
6782 there is no explicit 'tsd' field in the referent of the tag pointer,
6783 and instead the tsd pointer is stored just before the dispatch
6784 table. */
6785
6786 type = ada_get_tsd_type (current_inferior());
6787 if (type == NULL)
6788 return NULL;
6789 type = lookup_pointer_type (lookup_pointer_type (type));
6790 val = value_cast (type, tag);
6791 if (val == NULL)
6792 return NULL;
6793 return value_ind (value_ptradd (val, -1));
6794 }
6795
6796 /* Given the TSD of a tag (type-specific data), return a string
6797 containing the name of the associated type.
6798
6799 The returned value is good until the next call. May return NULL
6800 if we are unable to determine the tag name. */
6801
6802 static char *
6803 ada_tag_name_from_tsd (struct value *tsd)
6804 {
6805 static char name[1024];
6806 char *p;
6807 struct value *val;
6808
6809 val = ada_value_struct_elt (tsd, "expanded_name", 1);
6810 if (val == NULL)
6811 return NULL;
6812 read_memory_string (value_as_address (val), name, sizeof (name) - 1);
6813 for (p = name; *p != '\0'; p += 1)
6814 if (isalpha (*p))
6815 *p = tolower (*p);
6816 return name;
6817 }
6818
6819 /* The type name of the dynamic type denoted by the 'tag value TAG, as
6820 a C string.
6821
6822 Return NULL if the TAG is not an Ada tag, or if we were unable to
6823 determine the name of that tag. The result is good until the next
6824 call. */
6825
6826 const char *
6827 ada_tag_name (struct value *tag)
6828 {
6829 char *name = NULL;
6830
6831 if (!ada_is_tag_type (value_type (tag)))
6832 return NULL;
6833
6834 /* It is perfectly possible that an exception be raised while trying
6835 to determine the TAG's name, even under normal circumstances:
6836 The associated variable may be uninitialized or corrupted, for
6837 instance. We do not let any exception propagate past this point.
6838 instead we return NULL.
6839
6840 We also do not print the error message either (which often is very
6841 low-level (Eg: "Cannot read memory at 0x[...]"), but instead let
6842 the caller print a more meaningful message if necessary. */
6843 try
6844 {
6845 struct value *tsd = ada_get_tsd_from_tag (tag);
6846
6847 if (tsd != NULL)
6848 name = ada_tag_name_from_tsd (tsd);
6849 }
6850 catch (const gdb_exception_error &e)
6851 {
6852 }
6853
6854 return name;
6855 }
6856
6857 /* The parent type of TYPE, or NULL if none. */
6858
6859 struct type *
6860 ada_parent_type (struct type *type)
6861 {
6862 int i;
6863
6864 type = ada_check_typedef (type);
6865
6866 if (type == NULL || type->code () != TYPE_CODE_STRUCT)
6867 return NULL;
6868
6869 for (i = 0; i < type->num_fields (); i += 1)
6870 if (ada_is_parent_field (type, i))
6871 {
6872 struct type *parent_type = TYPE_FIELD_TYPE (type, i);
6873
6874 /* If the _parent field is a pointer, then dereference it. */
6875 if (parent_type->code () == TYPE_CODE_PTR)
6876 parent_type = TYPE_TARGET_TYPE (parent_type);
6877 /* If there is a parallel XVS type, get the actual base type. */
6878 parent_type = ada_get_base_type (parent_type);
6879
6880 return ada_check_typedef (parent_type);
6881 }
6882
6883 return NULL;
6884 }
6885
6886 /* True iff field number FIELD_NUM of structure type TYPE contains the
6887 parent-type (inherited) fields of a derived type. Assumes TYPE is
6888 a structure type with at least FIELD_NUM+1 fields. */
6889
6890 int
6891 ada_is_parent_field (struct type *type, int field_num)
6892 {
6893 const char *name = TYPE_FIELD_NAME (ada_check_typedef (type), field_num);
6894
6895 return (name != NULL
6896 && (startswith (name, "PARENT")
6897 || startswith (name, "_parent")));
6898 }
6899
6900 /* True iff field number FIELD_NUM of structure type TYPE is a
6901 transparent wrapper field (which should be silently traversed when doing
6902 field selection and flattened when printing). Assumes TYPE is a
6903 structure type with at least FIELD_NUM+1 fields. Such fields are always
6904 structures. */
6905
6906 int
6907 ada_is_wrapper_field (struct type *type, int field_num)
6908 {
6909 const char *name = TYPE_FIELD_NAME (type, field_num);
6910
6911 if (name != NULL && strcmp (name, "RETVAL") == 0)
6912 {
6913 /* This happens in functions with "out" or "in out" parameters
6914 which are passed by copy. For such functions, GNAT describes
6915 the function's return type as being a struct where the return
6916 value is in a field called RETVAL, and where the other "out"
6917 or "in out" parameters are fields of that struct. This is not
6918 a wrapper. */
6919 return 0;
6920 }
6921
6922 return (name != NULL
6923 && (startswith (name, "PARENT")
6924 || strcmp (name, "REP") == 0
6925 || startswith (name, "_parent")
6926 || name[0] == 'S' || name[0] == 'R' || name[0] == 'O'));
6927 }
6928
6929 /* True iff field number FIELD_NUM of structure or union type TYPE
6930 is a variant wrapper. Assumes TYPE is a structure type with at least
6931 FIELD_NUM+1 fields. */
6932
6933 int
6934 ada_is_variant_part (struct type *type, int field_num)
6935 {
6936 /* Only Ada types are eligible. */
6937 if (!ADA_TYPE_P (type))
6938 return 0;
6939
6940 struct type *field_type = TYPE_FIELD_TYPE (type, field_num);
6941
6942 return (field_type->code () == TYPE_CODE_UNION
6943 || (is_dynamic_field (type, field_num)
6944 && (TYPE_TARGET_TYPE (field_type)->code ()
6945 == TYPE_CODE_UNION)));
6946 }
6947
6948 /* Assuming that VAR_TYPE is a variant wrapper (type of the variant part)
6949 whose discriminants are contained in the record type OUTER_TYPE,
6950 returns the type of the controlling discriminant for the variant.
6951 May return NULL if the type could not be found. */
6952
6953 struct type *
6954 ada_variant_discrim_type (struct type *var_type, struct type *outer_type)
6955 {
6956 const char *name = ada_variant_discrim_name (var_type);
6957
6958 return ada_lookup_struct_elt_type (outer_type, name, 1, 1);
6959 }
6960
6961 /* Assuming that TYPE is the type of a variant wrapper, and FIELD_NUM is a
6962 valid field number within it, returns 1 iff field FIELD_NUM of TYPE
6963 represents a 'when others' clause; otherwise 0. */
6964
6965 static int
6966 ada_is_others_clause (struct type *type, int field_num)
6967 {
6968 const char *name = TYPE_FIELD_NAME (type, field_num);
6969
6970 return (name != NULL && name[0] == 'O');
6971 }
6972
6973 /* Assuming that TYPE0 is the type of the variant part of a record,
6974 returns the name of the discriminant controlling the variant.
6975 The value is valid until the next call to ada_variant_discrim_name. */
6976
6977 const char *
6978 ada_variant_discrim_name (struct type *type0)
6979 {
6980 static char *result = NULL;
6981 static size_t result_len = 0;
6982 struct type *type;
6983 const char *name;
6984 const char *discrim_end;
6985 const char *discrim_start;
6986
6987 if (type0->code () == TYPE_CODE_PTR)
6988 type = TYPE_TARGET_TYPE (type0);
6989 else
6990 type = type0;
6991
6992 name = ada_type_name (type);
6993
6994 if (name == NULL || name[0] == '\000')
6995 return "";
6996
6997 for (discrim_end = name + strlen (name) - 6; discrim_end != name;
6998 discrim_end -= 1)
6999 {
7000 if (startswith (discrim_end, "___XVN"))
7001 break;
7002 }
7003 if (discrim_end == name)
7004 return "";
7005
7006 for (discrim_start = discrim_end; discrim_start != name + 3;
7007 discrim_start -= 1)
7008 {
7009 if (discrim_start == name + 1)
7010 return "";
7011 if ((discrim_start > name + 3
7012 && startswith (discrim_start - 3, "___"))
7013 || discrim_start[-1] == '.')
7014 break;
7015 }
7016
7017 GROW_VECT (result, result_len, discrim_end - discrim_start + 1);
7018 strncpy (result, discrim_start, discrim_end - discrim_start);
7019 result[discrim_end - discrim_start] = '\0';
7020 return result;
7021 }
7022
7023 /* Scan STR for a subtype-encoded number, beginning at position K.
7024 Put the position of the character just past the number scanned in
7025 *NEW_K, if NEW_K!=NULL. Put the scanned number in *R, if R!=NULL.
7026 Return 1 if there was a valid number at the given position, and 0
7027 otherwise. A "subtype-encoded" number consists of the absolute value
7028 in decimal, followed by the letter 'm' to indicate a negative number.
7029 Assumes 0m does not occur. */
7030
7031 int
7032 ada_scan_number (const char str[], int k, LONGEST * R, int *new_k)
7033 {
7034 ULONGEST RU;
7035
7036 if (!isdigit (str[k]))
7037 return 0;
7038
7039 /* Do it the hard way so as not to make any assumption about
7040 the relationship of unsigned long (%lu scan format code) and
7041 LONGEST. */
7042 RU = 0;
7043 while (isdigit (str[k]))
7044 {
7045 RU = RU * 10 + (str[k] - '0');
7046 k += 1;
7047 }
7048
7049 if (str[k] == 'm')
7050 {
7051 if (R != NULL)
7052 *R = (-(LONGEST) (RU - 1)) - 1;
7053 k += 1;
7054 }
7055 else if (R != NULL)
7056 *R = (LONGEST) RU;
7057
7058 /* NOTE on the above: Technically, C does not say what the results of
7059 - (LONGEST) RU or (LONGEST) -RU are for RU == largest positive
7060 number representable as a LONGEST (although either would probably work
7061 in most implementations). When RU>0, the locution in the then branch
7062 above is always equivalent to the negative of RU. */
7063
7064 if (new_k != NULL)
7065 *new_k = k;
7066 return 1;
7067 }
7068
7069 /* Assuming that TYPE is a variant part wrapper type (a VARIANTS field),
7070 and FIELD_NUM is a valid field number within it, returns 1 iff VAL is
7071 in the range encoded by field FIELD_NUM of TYPE; otherwise 0. */
7072
7073 static int
7074 ada_in_variant (LONGEST val, struct type *type, int field_num)
7075 {
7076 const char *name = TYPE_FIELD_NAME (type, field_num);
7077 int p;
7078
7079 p = 0;
7080 while (1)
7081 {
7082 switch (name[p])
7083 {
7084 case '\0':
7085 return 0;
7086 case 'S':
7087 {
7088 LONGEST W;
7089
7090 if (!ada_scan_number (name, p + 1, &W, &p))
7091 return 0;
7092 if (val == W)
7093 return 1;
7094 break;
7095 }
7096 case 'R':
7097 {
7098 LONGEST L, U;
7099
7100 if (!ada_scan_number (name, p + 1, &L, &p)
7101 || name[p] != 'T' || !ada_scan_number (name, p + 1, &U, &p))
7102 return 0;
7103 if (val >= L && val <= U)
7104 return 1;
7105 break;
7106 }
7107 case 'O':
7108 return 1;
7109 default:
7110 return 0;
7111 }
7112 }
7113 }
7114
7115 /* FIXME: Lots of redundancy below. Try to consolidate. */
7116
7117 /* Given a value ARG1 (offset by OFFSET bytes) of a struct or union type
7118 ARG_TYPE, extract and return the value of one of its (non-static)
7119 fields. FIELDNO says which field. Differs from value_primitive_field
7120 only in that it can handle packed values of arbitrary type. */
7121
7122 struct value *
7123 ada_value_primitive_field (struct value *arg1, int offset, int fieldno,
7124 struct type *arg_type)
7125 {
7126 struct type *type;
7127
7128 arg_type = ada_check_typedef (arg_type);
7129 type = TYPE_FIELD_TYPE (arg_type, fieldno);
7130
7131 /* Handle packed fields. It might be that the field is not packed
7132 relative to its containing structure, but the structure itself is
7133 packed; in this case we must take the bit-field path. */
7134 if (TYPE_FIELD_BITSIZE (arg_type, fieldno) != 0 || value_bitpos (arg1) != 0)
7135 {
7136 int bit_pos = TYPE_FIELD_BITPOS (arg_type, fieldno);
7137 int bit_size = TYPE_FIELD_BITSIZE (arg_type, fieldno);
7138
7139 return ada_value_primitive_packed_val (arg1, value_contents (arg1),
7140 offset + bit_pos / 8,
7141 bit_pos % 8, bit_size, type);
7142 }
7143 else
7144 return value_primitive_field (arg1, offset, fieldno, arg_type);
7145 }
7146
7147 /* Find field with name NAME in object of type TYPE. If found,
7148 set the following for each argument that is non-null:
7149 - *FIELD_TYPE_P to the field's type;
7150 - *BYTE_OFFSET_P to OFFSET + the byte offset of the field within
7151 an object of that type;
7152 - *BIT_OFFSET_P to the bit offset modulo byte size of the field;
7153 - *BIT_SIZE_P to its size in bits if the field is packed, and
7154 0 otherwise;
7155 If INDEX_P is non-null, increment *INDEX_P by the number of source-visible
7156 fields up to but not including the desired field, or by the total
7157 number of fields if not found. A NULL value of NAME never
7158 matches; the function just counts visible fields in this case.
7159
7160 Notice that we need to handle when a tagged record hierarchy
7161 has some components with the same name, like in this scenario:
7162
7163 type Top_T is tagged record
7164 N : Integer := 1;
7165 U : Integer := 974;
7166 A : Integer := 48;
7167 end record;
7168
7169 type Middle_T is new Top.Top_T with record
7170 N : Character := 'a';
7171 C : Integer := 3;
7172 end record;
7173
7174 type Bottom_T is new Middle.Middle_T with record
7175 N : Float := 4.0;
7176 C : Character := '5';
7177 X : Integer := 6;
7178 A : Character := 'J';
7179 end record;
7180
7181 Let's say we now have a variable declared and initialized as follow:
7182
7183 TC : Top_A := new Bottom_T;
7184
7185 And then we use this variable to call this function
7186
7187 procedure Assign (Obj: in out Top_T; TV : Integer);
7188
7189 as follow:
7190
7191 Assign (Top_T (B), 12);
7192
7193 Now, we're in the debugger, and we're inside that procedure
7194 then and we want to print the value of obj.c:
7195
7196 Usually, the tagged record or one of the parent type owns the
7197 component to print and there's no issue but in this particular
7198 case, what does it mean to ask for Obj.C? Since the actual
7199 type for object is type Bottom_T, it could mean two things: type
7200 component C from the Middle_T view, but also component C from
7201 Bottom_T. So in that "undefined" case, when the component is
7202 not found in the non-resolved type (which includes all the
7203 components of the parent type), then resolve it and see if we
7204 get better luck once expanded.
7205
7206 In the case of homonyms in the derived tagged type, we don't
7207 guaranty anything, and pick the one that's easiest for us
7208 to program.
7209
7210 Returns 1 if found, 0 otherwise. */
7211
7212 static int
7213 find_struct_field (const char *name, struct type *type, int offset,
7214 struct type **field_type_p,
7215 int *byte_offset_p, int *bit_offset_p, int *bit_size_p,
7216 int *index_p)
7217 {
7218 int i;
7219 int parent_offset = -1;
7220
7221 type = ada_check_typedef (type);
7222
7223 if (field_type_p != NULL)
7224 *field_type_p = NULL;
7225 if (byte_offset_p != NULL)
7226 *byte_offset_p = 0;
7227 if (bit_offset_p != NULL)
7228 *bit_offset_p = 0;
7229 if (bit_size_p != NULL)
7230 *bit_size_p = 0;
7231
7232 for (i = 0; i < type->num_fields (); i += 1)
7233 {
7234 int bit_pos = TYPE_FIELD_BITPOS (type, i);
7235 int fld_offset = offset + bit_pos / 8;
7236 const char *t_field_name = TYPE_FIELD_NAME (type, i);
7237
7238 if (t_field_name == NULL)
7239 continue;
7240
7241 else if (ada_is_parent_field (type, i))
7242 {
7243 /* This is a field pointing us to the parent type of a tagged
7244 type. As hinted in this function's documentation, we give
7245 preference to fields in the current record first, so what
7246 we do here is just record the index of this field before
7247 we skip it. If it turns out we couldn't find our field
7248 in the current record, then we'll get back to it and search
7249 inside it whether the field might exist in the parent. */
7250
7251 parent_offset = i;
7252 continue;
7253 }
7254
7255 else if (name != NULL && field_name_match (t_field_name, name))
7256 {
7257 int bit_size = TYPE_FIELD_BITSIZE (type, i);
7258
7259 if (field_type_p != NULL)
7260 *field_type_p = TYPE_FIELD_TYPE (type, i);
7261 if (byte_offset_p != NULL)
7262 *byte_offset_p = fld_offset;
7263 if (bit_offset_p != NULL)
7264 *bit_offset_p = bit_pos % 8;
7265 if (bit_size_p != NULL)
7266 *bit_size_p = bit_size;
7267 return 1;
7268 }
7269 else if (ada_is_wrapper_field (type, i))
7270 {
7271 if (find_struct_field (name, TYPE_FIELD_TYPE (type, i), fld_offset,
7272 field_type_p, byte_offset_p, bit_offset_p,
7273 bit_size_p, index_p))
7274 return 1;
7275 }
7276 else if (ada_is_variant_part (type, i))
7277 {
7278 /* PNH: Wait. Do we ever execute this section, or is ARG always of
7279 fixed type?? */
7280 int j;
7281 struct type *field_type
7282 = ada_check_typedef (TYPE_FIELD_TYPE (type, i));
7283
7284 for (j = 0; j < field_type->num_fields (); j += 1)
7285 {
7286 if (find_struct_field (name, TYPE_FIELD_TYPE (field_type, j),
7287 fld_offset
7288 + TYPE_FIELD_BITPOS (field_type, j) / 8,
7289 field_type_p, byte_offset_p,
7290 bit_offset_p, bit_size_p, index_p))
7291 return 1;
7292 }
7293 }
7294 else if (index_p != NULL)
7295 *index_p += 1;
7296 }
7297
7298 /* Field not found so far. If this is a tagged type which
7299 has a parent, try finding that field in the parent now. */
7300
7301 if (parent_offset != -1)
7302 {
7303 int bit_pos = TYPE_FIELD_BITPOS (type, parent_offset);
7304 int fld_offset = offset + bit_pos / 8;
7305
7306 if (find_struct_field (name, TYPE_FIELD_TYPE (type, parent_offset),
7307 fld_offset, field_type_p, byte_offset_p,
7308 bit_offset_p, bit_size_p, index_p))
7309 return 1;
7310 }
7311
7312 return 0;
7313 }
7314
7315 /* Number of user-visible fields in record type TYPE. */
7316
7317 static int
7318 num_visible_fields (struct type *type)
7319 {
7320 int n;
7321
7322 n = 0;
7323 find_struct_field (NULL, type, 0, NULL, NULL, NULL, NULL, &n);
7324 return n;
7325 }
7326
7327 /* Look for a field NAME in ARG. Adjust the address of ARG by OFFSET bytes,
7328 and search in it assuming it has (class) type TYPE.
7329 If found, return value, else return NULL.
7330
7331 Searches recursively through wrapper fields (e.g., '_parent').
7332
7333 In the case of homonyms in the tagged types, please refer to the
7334 long explanation in find_struct_field's function documentation. */
7335
7336 static struct value *
7337 ada_search_struct_field (const char *name, struct value *arg, int offset,
7338 struct type *type)
7339 {
7340 int i;
7341 int parent_offset = -1;
7342
7343 type = ada_check_typedef (type);
7344 for (i = 0; i < type->num_fields (); i += 1)
7345 {
7346 const char *t_field_name = TYPE_FIELD_NAME (type, i);
7347
7348 if (t_field_name == NULL)
7349 continue;
7350
7351 else if (ada_is_parent_field (type, i))
7352 {
7353 /* This is a field pointing us to the parent type of a tagged
7354 type. As hinted in this function's documentation, we give
7355 preference to fields in the current record first, so what
7356 we do here is just record the index of this field before
7357 we skip it. If it turns out we couldn't find our field
7358 in the current record, then we'll get back to it and search
7359 inside it whether the field might exist in the parent. */
7360
7361 parent_offset = i;
7362 continue;
7363 }
7364
7365 else if (field_name_match (t_field_name, name))
7366 return ada_value_primitive_field (arg, offset, i, type);
7367
7368 else if (ada_is_wrapper_field (type, i))
7369 {
7370 struct value *v = /* Do not let indent join lines here. */
7371 ada_search_struct_field (name, arg,
7372 offset + TYPE_FIELD_BITPOS (type, i) / 8,
7373 TYPE_FIELD_TYPE (type, i));
7374
7375 if (v != NULL)
7376 return v;
7377 }
7378
7379 else if (ada_is_variant_part (type, i))
7380 {
7381 /* PNH: Do we ever get here? See find_struct_field. */
7382 int j;
7383 struct type *field_type = ada_check_typedef (TYPE_FIELD_TYPE (type,
7384 i));
7385 int var_offset = offset + TYPE_FIELD_BITPOS (type, i) / 8;
7386
7387 for (j = 0; j < field_type->num_fields (); j += 1)
7388 {
7389 struct value *v = ada_search_struct_field /* Force line
7390 break. */
7391 (name, arg,
7392 var_offset + TYPE_FIELD_BITPOS (field_type, j) / 8,
7393 TYPE_FIELD_TYPE (field_type, j));
7394
7395 if (v != NULL)
7396 return v;
7397 }
7398 }
7399 }
7400
7401 /* Field not found so far. If this is a tagged type which
7402 has a parent, try finding that field in the parent now. */
7403
7404 if (parent_offset != -1)
7405 {
7406 struct value *v = ada_search_struct_field (
7407 name, arg, offset + TYPE_FIELD_BITPOS (type, parent_offset) / 8,
7408 TYPE_FIELD_TYPE (type, parent_offset));
7409
7410 if (v != NULL)
7411 return v;
7412 }
7413
7414 return NULL;
7415 }
7416
7417 static struct value *ada_index_struct_field_1 (int *, struct value *,
7418 int, struct type *);
7419
7420
7421 /* Return field #INDEX in ARG, where the index is that returned by
7422 * find_struct_field through its INDEX_P argument. Adjust the address
7423 * of ARG by OFFSET bytes, and search in it assuming it has (class) type TYPE.
7424 * If found, return value, else return NULL. */
7425
7426 static struct value *
7427 ada_index_struct_field (int index, struct value *arg, int offset,
7428 struct type *type)
7429 {
7430 return ada_index_struct_field_1 (&index, arg, offset, type);
7431 }
7432
7433
7434 /* Auxiliary function for ada_index_struct_field. Like
7435 * ada_index_struct_field, but takes index from *INDEX_P and modifies
7436 * *INDEX_P. */
7437
7438 static struct value *
7439 ada_index_struct_field_1 (int *index_p, struct value *arg, int offset,
7440 struct type *type)
7441 {
7442 int i;
7443 type = ada_check_typedef (type);
7444
7445 for (i = 0; i < type->num_fields (); i += 1)
7446 {
7447 if (TYPE_FIELD_NAME (type, i) == NULL)
7448 continue;
7449 else if (ada_is_wrapper_field (type, i))
7450 {
7451 struct value *v = /* Do not let indent join lines here. */
7452 ada_index_struct_field_1 (index_p, arg,
7453 offset + TYPE_FIELD_BITPOS (type, i) / 8,
7454 TYPE_FIELD_TYPE (type, i));
7455
7456 if (v != NULL)
7457 return v;
7458 }
7459
7460 else if (ada_is_variant_part (type, i))
7461 {
7462 /* PNH: Do we ever get here? See ada_search_struct_field,
7463 find_struct_field. */
7464 error (_("Cannot assign this kind of variant record"));
7465 }
7466 else if (*index_p == 0)
7467 return ada_value_primitive_field (arg, offset, i, type);
7468 else
7469 *index_p -= 1;
7470 }
7471 return NULL;
7472 }
7473
7474 /* Return a string representation of type TYPE. */
7475
7476 static std::string
7477 type_as_string (struct type *type)
7478 {
7479 string_file tmp_stream;
7480
7481 type_print (type, "", &tmp_stream, -1);
7482
7483 return std::move (tmp_stream.string ());
7484 }
7485
7486 /* Given a type TYPE, look up the type of the component of type named NAME.
7487 If DISPP is non-null, add its byte displacement from the beginning of a
7488 structure (pointed to by a value) of type TYPE to *DISPP (does not
7489 work for packed fields).
7490
7491 Matches any field whose name has NAME as a prefix, possibly
7492 followed by "___".
7493
7494 TYPE can be either a struct or union. If REFOK, TYPE may also
7495 be a (pointer or reference)+ to a struct or union, and the
7496 ultimate target type will be searched.
7497
7498 Looks recursively into variant clauses and parent types.
7499
7500 In the case of homonyms in the tagged types, please refer to the
7501 long explanation in find_struct_field's function documentation.
7502
7503 If NOERR is nonzero, return NULL if NAME is not suitably defined or
7504 TYPE is not a type of the right kind. */
7505
7506 static struct type *
7507 ada_lookup_struct_elt_type (struct type *type, const char *name, int refok,
7508 int noerr)
7509 {
7510 int i;
7511 int parent_offset = -1;
7512
7513 if (name == NULL)
7514 goto BadName;
7515
7516 if (refok && type != NULL)
7517 while (1)
7518 {
7519 type = ada_check_typedef (type);
7520 if (type->code () != TYPE_CODE_PTR && type->code () != TYPE_CODE_REF)
7521 break;
7522 type = TYPE_TARGET_TYPE (type);
7523 }
7524
7525 if (type == NULL
7526 || (type->code () != TYPE_CODE_STRUCT
7527 && type->code () != TYPE_CODE_UNION))
7528 {
7529 if (noerr)
7530 return NULL;
7531
7532 error (_("Type %s is not a structure or union type"),
7533 type != NULL ? type_as_string (type).c_str () : _("(null)"));
7534 }
7535
7536 type = to_static_fixed_type (type);
7537
7538 for (i = 0; i < type->num_fields (); i += 1)
7539 {
7540 const char *t_field_name = TYPE_FIELD_NAME (type, i);
7541 struct type *t;
7542
7543 if (t_field_name == NULL)
7544 continue;
7545
7546 else if (ada_is_parent_field (type, i))
7547 {
7548 /* This is a field pointing us to the parent type of a tagged
7549 type. As hinted in this function's documentation, we give
7550 preference to fields in the current record first, so what
7551 we do here is just record the index of this field before
7552 we skip it. If it turns out we couldn't find our field
7553 in the current record, then we'll get back to it and search
7554 inside it whether the field might exist in the parent. */
7555
7556 parent_offset = i;
7557 continue;
7558 }
7559
7560 else if (field_name_match (t_field_name, name))
7561 return TYPE_FIELD_TYPE (type, i);
7562
7563 else if (ada_is_wrapper_field (type, i))
7564 {
7565 t = ada_lookup_struct_elt_type (TYPE_FIELD_TYPE (type, i), name,
7566 0, 1);
7567 if (t != NULL)
7568 return t;
7569 }
7570
7571 else if (ada_is_variant_part (type, i))
7572 {
7573 int j;
7574 struct type *field_type = ada_check_typedef (TYPE_FIELD_TYPE (type,
7575 i));
7576
7577 for (j = field_type->num_fields () - 1; j >= 0; j -= 1)
7578 {
7579 /* FIXME pnh 2008/01/26: We check for a field that is
7580 NOT wrapped in a struct, since the compiler sometimes
7581 generates these for unchecked variant types. Revisit
7582 if the compiler changes this practice. */
7583 const char *v_field_name = TYPE_FIELD_NAME (field_type, j);
7584
7585 if (v_field_name != NULL
7586 && field_name_match (v_field_name, name))
7587 t = TYPE_FIELD_TYPE (field_type, j);
7588 else
7589 t = ada_lookup_struct_elt_type (TYPE_FIELD_TYPE (field_type,
7590 j),
7591 name, 0, 1);
7592
7593 if (t != NULL)
7594 return t;
7595 }
7596 }
7597
7598 }
7599
7600 /* Field not found so far. If this is a tagged type which
7601 has a parent, try finding that field in the parent now. */
7602
7603 if (parent_offset != -1)
7604 {
7605 struct type *t;
7606
7607 t = ada_lookup_struct_elt_type (TYPE_FIELD_TYPE (type, parent_offset),
7608 name, 0, 1);
7609 if (t != NULL)
7610 return t;
7611 }
7612
7613 BadName:
7614 if (!noerr)
7615 {
7616 const char *name_str = name != NULL ? name : _("<null>");
7617
7618 error (_("Type %s has no component named %s"),
7619 type_as_string (type).c_str (), name_str);
7620 }
7621
7622 return NULL;
7623 }
7624
7625 /* Assuming that VAR_TYPE is the type of a variant part of a record (a union),
7626 within a value of type OUTER_TYPE, return true iff VAR_TYPE
7627 represents an unchecked union (that is, the variant part of a
7628 record that is named in an Unchecked_Union pragma). */
7629
7630 static int
7631 is_unchecked_variant (struct type *var_type, struct type *outer_type)
7632 {
7633 const char *discrim_name = ada_variant_discrim_name (var_type);
7634
7635 return (ada_lookup_struct_elt_type (outer_type, discrim_name, 0, 1) == NULL);
7636 }
7637
7638
7639 /* Assuming that VAR_TYPE is the type of a variant part of a record (a union),
7640 within OUTER, determine which variant clause (field number in VAR_TYPE,
7641 numbering from 0) is applicable. Returns -1 if none are. */
7642
7643 int
7644 ada_which_variant_applies (struct type *var_type, struct value *outer)
7645 {
7646 int others_clause;
7647 int i;
7648 const char *discrim_name = ada_variant_discrim_name (var_type);
7649 struct value *discrim;
7650 LONGEST discrim_val;
7651
7652 /* Using plain value_from_contents_and_address here causes problems
7653 because we will end up trying to resolve a type that is currently
7654 being constructed. */
7655 discrim = ada_value_struct_elt (outer, discrim_name, 1);
7656 if (discrim == NULL)
7657 return -1;
7658 discrim_val = value_as_long (discrim);
7659
7660 others_clause = -1;
7661 for (i = 0; i < var_type->num_fields (); i += 1)
7662 {
7663 if (ada_is_others_clause (var_type, i))
7664 others_clause = i;
7665 else if (ada_in_variant (discrim_val, var_type, i))
7666 return i;
7667 }
7668
7669 return others_clause;
7670 }
7671 \f
7672
7673
7674 /* Dynamic-Sized Records */
7675
7676 /* Strategy: The type ostensibly attached to a value with dynamic size
7677 (i.e., a size that is not statically recorded in the debugging
7678 data) does not accurately reflect the size or layout of the value.
7679 Our strategy is to convert these values to values with accurate,
7680 conventional types that are constructed on the fly. */
7681
7682 /* There is a subtle and tricky problem here. In general, we cannot
7683 determine the size of dynamic records without its data. However,
7684 the 'struct value' data structure, which GDB uses to represent
7685 quantities in the inferior process (the target), requires the size
7686 of the type at the time of its allocation in order to reserve space
7687 for GDB's internal copy of the data. That's why the
7688 'to_fixed_xxx_type' routines take (target) addresses as parameters,
7689 rather than struct value*s.
7690
7691 However, GDB's internal history variables ($1, $2, etc.) are
7692 struct value*s containing internal copies of the data that are not, in
7693 general, the same as the data at their corresponding addresses in
7694 the target. Fortunately, the types we give to these values are all
7695 conventional, fixed-size types (as per the strategy described
7696 above), so that we don't usually have to perform the
7697 'to_fixed_xxx_type' conversions to look at their values.
7698 Unfortunately, there is one exception: if one of the internal
7699 history variables is an array whose elements are unconstrained
7700 records, then we will need to create distinct fixed types for each
7701 element selected. */
7702
7703 /* The upshot of all of this is that many routines take a (type, host
7704 address, target address) triple as arguments to represent a value.
7705 The host address, if non-null, is supposed to contain an internal
7706 copy of the relevant data; otherwise, the program is to consult the
7707 target at the target address. */
7708
7709 /* Assuming that VAL0 represents a pointer value, the result of
7710 dereferencing it. Differs from value_ind in its treatment of
7711 dynamic-sized types. */
7712
7713 struct value *
7714 ada_value_ind (struct value *val0)
7715 {
7716 struct value *val = value_ind (val0);
7717
7718 if (ada_is_tagged_type (value_type (val), 0))
7719 val = ada_tag_value_at_base_address (val);
7720
7721 return ada_to_fixed_value (val);
7722 }
7723
7724 /* The value resulting from dereferencing any "reference to"
7725 qualifiers on VAL0. */
7726
7727 static struct value *
7728 ada_coerce_ref (struct value *val0)
7729 {
7730 if (value_type (val0)->code () == TYPE_CODE_REF)
7731 {
7732 struct value *val = val0;
7733
7734 val = coerce_ref (val);
7735
7736 if (ada_is_tagged_type (value_type (val), 0))
7737 val = ada_tag_value_at_base_address (val);
7738
7739 return ada_to_fixed_value (val);
7740 }
7741 else
7742 return val0;
7743 }
7744
7745 /* Return the bit alignment required for field #F of template type TYPE. */
7746
7747 static unsigned int
7748 field_alignment (struct type *type, int f)
7749 {
7750 const char *name = TYPE_FIELD_NAME (type, f);
7751 int len;
7752 int align_offset;
7753
7754 /* The field name should never be null, unless the debugging information
7755 is somehow malformed. In this case, we assume the field does not
7756 require any alignment. */
7757 if (name == NULL)
7758 return 1;
7759
7760 len = strlen (name);
7761
7762 if (!isdigit (name[len - 1]))
7763 return 1;
7764
7765 if (isdigit (name[len - 2]))
7766 align_offset = len - 2;
7767 else
7768 align_offset = len - 1;
7769
7770 if (align_offset < 7 || !startswith (name + align_offset - 6, "___XV"))
7771 return TARGET_CHAR_BIT;
7772
7773 return atoi (name + align_offset) * TARGET_CHAR_BIT;
7774 }
7775
7776 /* Find a typedef or tag symbol named NAME. Ignores ambiguity. */
7777
7778 static struct symbol *
7779 ada_find_any_type_symbol (const char *name)
7780 {
7781 struct symbol *sym;
7782
7783 sym = standard_lookup (name, get_selected_block (NULL), VAR_DOMAIN);
7784 if (sym != NULL && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
7785 return sym;
7786
7787 sym = standard_lookup (name, NULL, STRUCT_DOMAIN);
7788 return sym;
7789 }
7790
7791 /* Find a type named NAME. Ignores ambiguity. This routine will look
7792 solely for types defined by debug info, it will not search the GDB
7793 primitive types. */
7794
7795 static struct type *
7796 ada_find_any_type (const char *name)
7797 {
7798 struct symbol *sym = ada_find_any_type_symbol (name);
7799
7800 if (sym != NULL)
7801 return SYMBOL_TYPE (sym);
7802
7803 return NULL;
7804 }
7805
7806 /* Given NAME_SYM and an associated BLOCK, find a "renaming" symbol
7807 associated with NAME_SYM's name. NAME_SYM may itself be a renaming
7808 symbol, in which case it is returned. Otherwise, this looks for
7809 symbols whose name is that of NAME_SYM suffixed with "___XR".
7810 Return symbol if found, and NULL otherwise. */
7811
7812 static bool
7813 ada_is_renaming_symbol (struct symbol *name_sym)
7814 {
7815 const char *name = name_sym->linkage_name ();
7816 return strstr (name, "___XR") != NULL;
7817 }
7818
7819 /* Because of GNAT encoding conventions, several GDB symbols may match a
7820 given type name. If the type denoted by TYPE0 is to be preferred to
7821 that of TYPE1 for purposes of type printing, return non-zero;
7822 otherwise return 0. */
7823
7824 int
7825 ada_prefer_type (struct type *type0, struct type *type1)
7826 {
7827 if (type1 == NULL)
7828 return 1;
7829 else if (type0 == NULL)
7830 return 0;
7831 else if (type1->code () == TYPE_CODE_VOID)
7832 return 1;
7833 else if (type0->code () == TYPE_CODE_VOID)
7834 return 0;
7835 else if (type1->name () == NULL && type0->name () != NULL)
7836 return 1;
7837 else if (ada_is_constrained_packed_array_type (type0))
7838 return 1;
7839 else if (ada_is_array_descriptor_type (type0)
7840 && !ada_is_array_descriptor_type (type1))
7841 return 1;
7842 else
7843 {
7844 const char *type0_name = type0->name ();
7845 const char *type1_name = type1->name ();
7846
7847 if (type0_name != NULL && strstr (type0_name, "___XR") != NULL
7848 && (type1_name == NULL || strstr (type1_name, "___XR") == NULL))
7849 return 1;
7850 }
7851 return 0;
7852 }
7853
7854 /* The name of TYPE, which is its TYPE_NAME. Null if TYPE is
7855 null. */
7856
7857 const char *
7858 ada_type_name (struct type *type)
7859 {
7860 if (type == NULL)
7861 return NULL;
7862 return type->name ();
7863 }
7864
7865 /* Search the list of "descriptive" types associated to TYPE for a type
7866 whose name is NAME. */
7867
7868 static struct type *
7869 find_parallel_type_by_descriptive_type (struct type *type, const char *name)
7870 {
7871 struct type *result, *tmp;
7872
7873 if (ada_ignore_descriptive_types_p)
7874 return NULL;
7875
7876 /* If there no descriptive-type info, then there is no parallel type
7877 to be found. */
7878 if (!HAVE_GNAT_AUX_INFO (type))
7879 return NULL;
7880
7881 result = TYPE_DESCRIPTIVE_TYPE (type);
7882 while (result != NULL)
7883 {
7884 const char *result_name = ada_type_name (result);
7885
7886 if (result_name == NULL)
7887 {
7888 warning (_("unexpected null name on descriptive type"));
7889 return NULL;
7890 }
7891
7892 /* If the names match, stop. */
7893 if (strcmp (result_name, name) == 0)
7894 break;
7895
7896 /* Otherwise, look at the next item on the list, if any. */
7897 if (HAVE_GNAT_AUX_INFO (result))
7898 tmp = TYPE_DESCRIPTIVE_TYPE (result);
7899 else
7900 tmp = NULL;
7901
7902 /* If not found either, try after having resolved the typedef. */
7903 if (tmp != NULL)
7904 result = tmp;
7905 else
7906 {
7907 result = check_typedef (result);
7908 if (HAVE_GNAT_AUX_INFO (result))
7909 result = TYPE_DESCRIPTIVE_TYPE (result);
7910 else
7911 result = NULL;
7912 }
7913 }
7914
7915 /* If we didn't find a match, see whether this is a packed array. With
7916 older compilers, the descriptive type information is either absent or
7917 irrelevant when it comes to packed arrays so the above lookup fails.
7918 Fall back to using a parallel lookup by name in this case. */
7919 if (result == NULL && ada_is_constrained_packed_array_type (type))
7920 return ada_find_any_type (name);
7921
7922 return result;
7923 }
7924
7925 /* Find a parallel type to TYPE with the specified NAME, using the
7926 descriptive type taken from the debugging information, if available,
7927 and otherwise using the (slower) name-based method. */
7928
7929 static struct type *
7930 ada_find_parallel_type_with_name (struct type *type, const char *name)
7931 {
7932 struct type *result = NULL;
7933
7934 if (HAVE_GNAT_AUX_INFO (type))
7935 result = find_parallel_type_by_descriptive_type (type, name);
7936 else
7937 result = ada_find_any_type (name);
7938
7939 return result;
7940 }
7941
7942 /* Same as above, but specify the name of the parallel type by appending
7943 SUFFIX to the name of TYPE. */
7944
7945 struct type *
7946 ada_find_parallel_type (struct type *type, const char *suffix)
7947 {
7948 char *name;
7949 const char *type_name = ada_type_name (type);
7950 int len;
7951
7952 if (type_name == NULL)
7953 return NULL;
7954
7955 len = strlen (type_name);
7956
7957 name = (char *) alloca (len + strlen (suffix) + 1);
7958
7959 strcpy (name, type_name);
7960 strcpy (name + len, suffix);
7961
7962 return ada_find_parallel_type_with_name (type, name);
7963 }
7964
7965 /* If TYPE is a variable-size record type, return the corresponding template
7966 type describing its fields. Otherwise, return NULL. */
7967
7968 static struct type *
7969 dynamic_template_type (struct type *type)
7970 {
7971 type = ada_check_typedef (type);
7972
7973 if (type == NULL || type->code () != TYPE_CODE_STRUCT
7974 || ada_type_name (type) == NULL)
7975 return NULL;
7976 else
7977 {
7978 int len = strlen (ada_type_name (type));
7979
7980 if (len > 6 && strcmp (ada_type_name (type) + len - 6, "___XVE") == 0)
7981 return type;
7982 else
7983 return ada_find_parallel_type (type, "___XVE");
7984 }
7985 }
7986
7987 /* Assuming that TEMPL_TYPE is a union or struct type, returns
7988 non-zero iff field FIELD_NUM of TEMPL_TYPE has dynamic size. */
7989
7990 static int
7991 is_dynamic_field (struct type *templ_type, int field_num)
7992 {
7993 const char *name = TYPE_FIELD_NAME (templ_type, field_num);
7994
7995 return name != NULL
7996 && TYPE_FIELD_TYPE (templ_type, field_num)->code () == TYPE_CODE_PTR
7997 && strstr (name, "___XVL") != NULL;
7998 }
7999
8000 /* The index of the variant field of TYPE, or -1 if TYPE does not
8001 represent a variant record type. */
8002
8003 static int
8004 variant_field_index (struct type *type)
8005 {
8006 int f;
8007
8008 if (type == NULL || type->code () != TYPE_CODE_STRUCT)
8009 return -1;
8010
8011 for (f = 0; f < type->num_fields (); f += 1)
8012 {
8013 if (ada_is_variant_part (type, f))
8014 return f;
8015 }
8016 return -1;
8017 }
8018
8019 /* A record type with no fields. */
8020
8021 static struct type *
8022 empty_record (struct type *templ)
8023 {
8024 struct type *type = alloc_type_copy (templ);
8025
8026 type->set_code (TYPE_CODE_STRUCT);
8027 INIT_NONE_SPECIFIC (type);
8028 type->set_name ("<empty>");
8029 TYPE_LENGTH (type) = 0;
8030 return type;
8031 }
8032
8033 /* An ordinary record type (with fixed-length fields) that describes
8034 the value of type TYPE at VALADDR or ADDRESS (see comments at
8035 the beginning of this section) VAL according to GNAT conventions.
8036 DVAL0 should describe the (portion of a) record that contains any
8037 necessary discriminants. It should be NULL if value_type (VAL) is
8038 an outer-level type (i.e., as opposed to a branch of a variant.) A
8039 variant field (unless unchecked) is replaced by a particular branch
8040 of the variant.
8041
8042 If not KEEP_DYNAMIC_FIELDS, then all fields whose position or
8043 length are not statically known are discarded. As a consequence,
8044 VALADDR, ADDRESS and DVAL0 are ignored.
8045
8046 NOTE: Limitations: For now, we assume that dynamic fields and
8047 variants occupy whole numbers of bytes. However, they need not be
8048 byte-aligned. */
8049
8050 struct type *
8051 ada_template_to_fixed_record_type_1 (struct type *type,
8052 const gdb_byte *valaddr,
8053 CORE_ADDR address, struct value *dval0,
8054 int keep_dynamic_fields)
8055 {
8056 struct value *mark = value_mark ();
8057 struct value *dval;
8058 struct type *rtype;
8059 int nfields, bit_len;
8060 int variant_field;
8061 long off;
8062 int fld_bit_len;
8063 int f;
8064
8065 /* Compute the number of fields in this record type that are going
8066 to be processed: unless keep_dynamic_fields, this includes only
8067 fields whose position and length are static will be processed. */
8068 if (keep_dynamic_fields)
8069 nfields = type->num_fields ();
8070 else
8071 {
8072 nfields = 0;
8073 while (nfields < type->num_fields ()
8074 && !ada_is_variant_part (type, nfields)
8075 && !is_dynamic_field (type, nfields))
8076 nfields++;
8077 }
8078
8079 rtype = alloc_type_copy (type);
8080 rtype->set_code (TYPE_CODE_STRUCT);
8081 INIT_NONE_SPECIFIC (rtype);
8082 rtype->set_num_fields (nfields);
8083 rtype->set_fields
8084 ((struct field *) TYPE_ZALLOC (rtype, nfields * sizeof (struct field)));
8085 rtype->set_name (ada_type_name (type));
8086 TYPE_FIXED_INSTANCE (rtype) = 1;
8087
8088 off = 0;
8089 bit_len = 0;
8090 variant_field = -1;
8091
8092 for (f = 0; f < nfields; f += 1)
8093 {
8094 off = align_up (off, field_alignment (type, f))
8095 + TYPE_FIELD_BITPOS (type, f);
8096 SET_FIELD_BITPOS (rtype->field (f), off);
8097 TYPE_FIELD_BITSIZE (rtype, f) = 0;
8098
8099 if (ada_is_variant_part (type, f))
8100 {
8101 variant_field = f;
8102 fld_bit_len = 0;
8103 }
8104 else if (is_dynamic_field (type, f))
8105 {
8106 const gdb_byte *field_valaddr = valaddr;
8107 CORE_ADDR field_address = address;
8108 struct type *field_type =
8109 TYPE_TARGET_TYPE (TYPE_FIELD_TYPE (type, f));
8110
8111 if (dval0 == NULL)
8112 {
8113 /* rtype's length is computed based on the run-time
8114 value of discriminants. If the discriminants are not
8115 initialized, the type size may be completely bogus and
8116 GDB may fail to allocate a value for it. So check the
8117 size first before creating the value. */
8118 ada_ensure_varsize_limit (rtype);
8119 /* Using plain value_from_contents_and_address here
8120 causes problems because we will end up trying to
8121 resolve a type that is currently being
8122 constructed. */
8123 dval = value_from_contents_and_address_unresolved (rtype,
8124 valaddr,
8125 address);
8126 rtype = value_type (dval);
8127 }
8128 else
8129 dval = dval0;
8130
8131 /* If the type referenced by this field is an aligner type, we need
8132 to unwrap that aligner type, because its size might not be set.
8133 Keeping the aligner type would cause us to compute the wrong
8134 size for this field, impacting the offset of the all the fields
8135 that follow this one. */
8136 if (ada_is_aligner_type (field_type))
8137 {
8138 long field_offset = TYPE_FIELD_BITPOS (field_type, f);
8139
8140 field_valaddr = cond_offset_host (field_valaddr, field_offset);
8141 field_address = cond_offset_target (field_address, field_offset);
8142 field_type = ada_aligned_type (field_type);
8143 }
8144
8145 field_valaddr = cond_offset_host (field_valaddr,
8146 off / TARGET_CHAR_BIT);
8147 field_address = cond_offset_target (field_address,
8148 off / TARGET_CHAR_BIT);
8149
8150 /* Get the fixed type of the field. Note that, in this case,
8151 we do not want to get the real type out of the tag: if
8152 the current field is the parent part of a tagged record,
8153 we will get the tag of the object. Clearly wrong: the real
8154 type of the parent is not the real type of the child. We
8155 would end up in an infinite loop. */
8156 field_type = ada_get_base_type (field_type);
8157 field_type = ada_to_fixed_type (field_type, field_valaddr,
8158 field_address, dval, 0);
8159 /* If the field size is already larger than the maximum
8160 object size, then the record itself will necessarily
8161 be larger than the maximum object size. We need to make
8162 this check now, because the size might be so ridiculously
8163 large (due to an uninitialized variable in the inferior)
8164 that it would cause an overflow when adding it to the
8165 record size. */
8166 ada_ensure_varsize_limit (field_type);
8167
8168 TYPE_FIELD_TYPE (rtype, f) = field_type;
8169 TYPE_FIELD_NAME (rtype, f) = TYPE_FIELD_NAME (type, f);
8170 /* The multiplication can potentially overflow. But because
8171 the field length has been size-checked just above, and
8172 assuming that the maximum size is a reasonable value,
8173 an overflow should not happen in practice. So rather than
8174 adding overflow recovery code to this already complex code,
8175 we just assume that it's not going to happen. */
8176 fld_bit_len =
8177 TYPE_LENGTH (TYPE_FIELD_TYPE (rtype, f)) * TARGET_CHAR_BIT;
8178 }
8179 else
8180 {
8181 /* Note: If this field's type is a typedef, it is important
8182 to preserve the typedef layer.
8183
8184 Otherwise, we might be transforming a typedef to a fat
8185 pointer (encoding a pointer to an unconstrained array),
8186 into a basic fat pointer (encoding an unconstrained
8187 array). As both types are implemented using the same
8188 structure, the typedef is the only clue which allows us
8189 to distinguish between the two options. Stripping it
8190 would prevent us from printing this field appropriately. */
8191 TYPE_FIELD_TYPE (rtype, f) = TYPE_FIELD_TYPE (type, f);
8192 TYPE_FIELD_NAME (rtype, f) = TYPE_FIELD_NAME (type, f);
8193 if (TYPE_FIELD_BITSIZE (type, f) > 0)
8194 fld_bit_len =
8195 TYPE_FIELD_BITSIZE (rtype, f) = TYPE_FIELD_BITSIZE (type, f);
8196 else
8197 {
8198 struct type *field_type = TYPE_FIELD_TYPE (type, f);
8199
8200 /* We need to be careful of typedefs when computing
8201 the length of our field. If this is a typedef,
8202 get the length of the target type, not the length
8203 of the typedef. */
8204 if (field_type->code () == TYPE_CODE_TYPEDEF)
8205 field_type = ada_typedef_target_type (field_type);
8206
8207 fld_bit_len =
8208 TYPE_LENGTH (ada_check_typedef (field_type)) * TARGET_CHAR_BIT;
8209 }
8210 }
8211 if (off + fld_bit_len > bit_len)
8212 bit_len = off + fld_bit_len;
8213 off += fld_bit_len;
8214 TYPE_LENGTH (rtype) =
8215 align_up (bit_len, TARGET_CHAR_BIT) / TARGET_CHAR_BIT;
8216 }
8217
8218 /* We handle the variant part, if any, at the end because of certain
8219 odd cases in which it is re-ordered so as NOT to be the last field of
8220 the record. This can happen in the presence of representation
8221 clauses. */
8222 if (variant_field >= 0)
8223 {
8224 struct type *branch_type;
8225
8226 off = TYPE_FIELD_BITPOS (rtype, variant_field);
8227
8228 if (dval0 == NULL)
8229 {
8230 /* Using plain value_from_contents_and_address here causes
8231 problems because we will end up trying to resolve a type
8232 that is currently being constructed. */
8233 dval = value_from_contents_and_address_unresolved (rtype, valaddr,
8234 address);
8235 rtype = value_type (dval);
8236 }
8237 else
8238 dval = dval0;
8239
8240 branch_type =
8241 to_fixed_variant_branch_type
8242 (TYPE_FIELD_TYPE (type, variant_field),
8243 cond_offset_host (valaddr, off / TARGET_CHAR_BIT),
8244 cond_offset_target (address, off / TARGET_CHAR_BIT), dval);
8245 if (branch_type == NULL)
8246 {
8247 for (f = variant_field + 1; f < rtype->num_fields (); f += 1)
8248 rtype->field (f - 1) = rtype->field (f);
8249 rtype->set_num_fields (rtype->num_fields () - 1);
8250 }
8251 else
8252 {
8253 TYPE_FIELD_TYPE (rtype, variant_field) = branch_type;
8254 TYPE_FIELD_NAME (rtype, variant_field) = "S";
8255 fld_bit_len =
8256 TYPE_LENGTH (TYPE_FIELD_TYPE (rtype, variant_field)) *
8257 TARGET_CHAR_BIT;
8258 if (off + fld_bit_len > bit_len)
8259 bit_len = off + fld_bit_len;
8260 TYPE_LENGTH (rtype) =
8261 align_up (bit_len, TARGET_CHAR_BIT) / TARGET_CHAR_BIT;
8262 }
8263 }
8264
8265 /* According to exp_dbug.ads, the size of TYPE for variable-size records
8266 should contain the alignment of that record, which should be a strictly
8267 positive value. If null or negative, then something is wrong, most
8268 probably in the debug info. In that case, we don't round up the size
8269 of the resulting type. If this record is not part of another structure,
8270 the current RTYPE length might be good enough for our purposes. */
8271 if (TYPE_LENGTH (type) <= 0)
8272 {
8273 if (rtype->name ())
8274 warning (_("Invalid type size for `%s' detected: %s."),
8275 rtype->name (), pulongest (TYPE_LENGTH (type)));
8276 else
8277 warning (_("Invalid type size for <unnamed> detected: %s."),
8278 pulongest (TYPE_LENGTH (type)));
8279 }
8280 else
8281 {
8282 TYPE_LENGTH (rtype) = align_up (TYPE_LENGTH (rtype),
8283 TYPE_LENGTH (type));
8284 }
8285
8286 value_free_to_mark (mark);
8287 if (TYPE_LENGTH (rtype) > varsize_limit)
8288 error (_("record type with dynamic size is larger than varsize-limit"));
8289 return rtype;
8290 }
8291
8292 /* As for ada_template_to_fixed_record_type_1 with KEEP_DYNAMIC_FIELDS
8293 of 1. */
8294
8295 static struct type *
8296 template_to_fixed_record_type (struct type *type, const gdb_byte *valaddr,
8297 CORE_ADDR address, struct value *dval0)
8298 {
8299 return ada_template_to_fixed_record_type_1 (type, valaddr,
8300 address, dval0, 1);
8301 }
8302
8303 /* An ordinary record type in which ___XVL-convention fields and
8304 ___XVU- and ___XVN-convention field types in TYPE0 are replaced with
8305 static approximations, containing all possible fields. Uses
8306 no runtime values. Useless for use in values, but that's OK,
8307 since the results are used only for type determinations. Works on both
8308 structs and unions. Representation note: to save space, we memorize
8309 the result of this function in the TYPE_TARGET_TYPE of the
8310 template type. */
8311
8312 static struct type *
8313 template_to_static_fixed_type (struct type *type0)
8314 {
8315 struct type *type;
8316 int nfields;
8317 int f;
8318
8319 /* No need no do anything if the input type is already fixed. */
8320 if (TYPE_FIXED_INSTANCE (type0))
8321 return type0;
8322
8323 /* Likewise if we already have computed the static approximation. */
8324 if (TYPE_TARGET_TYPE (type0) != NULL)
8325 return TYPE_TARGET_TYPE (type0);
8326
8327 /* Don't clone TYPE0 until we are sure we are going to need a copy. */
8328 type = type0;
8329 nfields = type0->num_fields ();
8330
8331 /* Whether or not we cloned TYPE0, cache the result so that we don't do
8332 recompute all over next time. */
8333 TYPE_TARGET_TYPE (type0) = type;
8334
8335 for (f = 0; f < nfields; f += 1)
8336 {
8337 struct type *field_type = TYPE_FIELD_TYPE (type0, f);
8338 struct type *new_type;
8339
8340 if (is_dynamic_field (type0, f))
8341 {
8342 field_type = ada_check_typedef (field_type);
8343 new_type = to_static_fixed_type (TYPE_TARGET_TYPE (field_type));
8344 }
8345 else
8346 new_type = static_unwrap_type (field_type);
8347
8348 if (new_type != field_type)
8349 {
8350 /* Clone TYPE0 only the first time we get a new field type. */
8351 if (type == type0)
8352 {
8353 TYPE_TARGET_TYPE (type0) = type = alloc_type_copy (type0);
8354 type->set_code (type0->code ());
8355 INIT_NONE_SPECIFIC (type);
8356 type->set_num_fields (nfields);
8357
8358 field *fields =
8359 ((struct field *)
8360 TYPE_ALLOC (type, nfields * sizeof (struct field)));
8361 memcpy (fields, type0->fields (),
8362 sizeof (struct field) * nfields);
8363 type->set_fields (fields);
8364
8365 type->set_name (ada_type_name (type0));
8366 TYPE_FIXED_INSTANCE (type) = 1;
8367 TYPE_LENGTH (type) = 0;
8368 }
8369 TYPE_FIELD_TYPE (type, f) = new_type;
8370 TYPE_FIELD_NAME (type, f) = TYPE_FIELD_NAME (type0, f);
8371 }
8372 }
8373
8374 return type;
8375 }
8376
8377 /* Given an object of type TYPE whose contents are at VALADDR and
8378 whose address in memory is ADDRESS, returns a revision of TYPE,
8379 which should be a non-dynamic-sized record, in which the variant
8380 part, if any, is replaced with the appropriate branch. Looks
8381 for discriminant values in DVAL0, which can be NULL if the record
8382 contains the necessary discriminant values. */
8383
8384 static struct type *
8385 to_record_with_fixed_variant_part (struct type *type, const gdb_byte *valaddr,
8386 CORE_ADDR address, struct value *dval0)
8387 {
8388 struct value *mark = value_mark ();
8389 struct value *dval;
8390 struct type *rtype;
8391 struct type *branch_type;
8392 int nfields = type->num_fields ();
8393 int variant_field = variant_field_index (type);
8394
8395 if (variant_field == -1)
8396 return type;
8397
8398 if (dval0 == NULL)
8399 {
8400 dval = value_from_contents_and_address (type, valaddr, address);
8401 type = value_type (dval);
8402 }
8403 else
8404 dval = dval0;
8405
8406 rtype = alloc_type_copy (type);
8407 rtype->set_code (TYPE_CODE_STRUCT);
8408 INIT_NONE_SPECIFIC (rtype);
8409 rtype->set_num_fields (nfields);
8410
8411 field *fields =
8412 (struct field *) TYPE_ALLOC (rtype, nfields * sizeof (struct field));
8413 memcpy (fields, type->fields (), sizeof (struct field) * nfields);
8414 rtype->set_fields (fields);
8415
8416 rtype->set_name (ada_type_name (type));
8417 TYPE_FIXED_INSTANCE (rtype) = 1;
8418 TYPE_LENGTH (rtype) = TYPE_LENGTH (type);
8419
8420 branch_type = to_fixed_variant_branch_type
8421 (TYPE_FIELD_TYPE (type, variant_field),
8422 cond_offset_host (valaddr,
8423 TYPE_FIELD_BITPOS (type, variant_field)
8424 / TARGET_CHAR_BIT),
8425 cond_offset_target (address,
8426 TYPE_FIELD_BITPOS (type, variant_field)
8427 / TARGET_CHAR_BIT), dval);
8428 if (branch_type == NULL)
8429 {
8430 int f;
8431
8432 for (f = variant_field + 1; f < nfields; f += 1)
8433 rtype->field (f - 1) = rtype->field (f);
8434 rtype->set_num_fields (rtype->num_fields () - 1);
8435 }
8436 else
8437 {
8438 TYPE_FIELD_TYPE (rtype, variant_field) = branch_type;
8439 TYPE_FIELD_NAME (rtype, variant_field) = "S";
8440 TYPE_FIELD_BITSIZE (rtype, variant_field) = 0;
8441 TYPE_LENGTH (rtype) += TYPE_LENGTH (branch_type);
8442 }
8443 TYPE_LENGTH (rtype) -= TYPE_LENGTH (TYPE_FIELD_TYPE (type, variant_field));
8444
8445 value_free_to_mark (mark);
8446 return rtype;
8447 }
8448
8449 /* An ordinary record type (with fixed-length fields) that describes
8450 the value at (TYPE0, VALADDR, ADDRESS) [see explanation at
8451 beginning of this section]. Any necessary discriminants' values
8452 should be in DVAL, a record value; it may be NULL if the object
8453 at ADDR itself contains any necessary discriminant values.
8454 Additionally, VALADDR and ADDRESS may also be NULL if no discriminant
8455 values from the record are needed. Except in the case that DVAL,
8456 VALADDR, and ADDRESS are all 0 or NULL, a variant field (unless
8457 unchecked) is replaced by a particular branch of the variant.
8458
8459 NOTE: the case in which DVAL and VALADDR are NULL and ADDRESS is 0
8460 is questionable and may be removed. It can arise during the
8461 processing of an unconstrained-array-of-record type where all the
8462 variant branches have exactly the same size. This is because in
8463 such cases, the compiler does not bother to use the XVS convention
8464 when encoding the record. I am currently dubious of this
8465 shortcut and suspect the compiler should be altered. FIXME. */
8466
8467 static struct type *
8468 to_fixed_record_type (struct type *type0, const gdb_byte *valaddr,
8469 CORE_ADDR address, struct value *dval)
8470 {
8471 struct type *templ_type;
8472
8473 if (TYPE_FIXED_INSTANCE (type0))
8474 return type0;
8475
8476 templ_type = dynamic_template_type (type0);
8477
8478 if (templ_type != NULL)
8479 return template_to_fixed_record_type (templ_type, valaddr, address, dval);
8480 else if (variant_field_index (type0) >= 0)
8481 {
8482 if (dval == NULL && valaddr == NULL && address == 0)
8483 return type0;
8484 return to_record_with_fixed_variant_part (type0, valaddr, address,
8485 dval);
8486 }
8487 else
8488 {
8489 TYPE_FIXED_INSTANCE (type0) = 1;
8490 return type0;
8491 }
8492
8493 }
8494
8495 /* An ordinary record type (with fixed-length fields) that describes
8496 the value at (VAR_TYPE0, VALADDR, ADDRESS), where VAR_TYPE0 is a
8497 union type. Any necessary discriminants' values should be in DVAL,
8498 a record value. That is, this routine selects the appropriate
8499 branch of the union at ADDR according to the discriminant value
8500 indicated in the union's type name. Returns VAR_TYPE0 itself if
8501 it represents a variant subject to a pragma Unchecked_Union. */
8502
8503 static struct type *
8504 to_fixed_variant_branch_type (struct type *var_type0, const gdb_byte *valaddr,
8505 CORE_ADDR address, struct value *dval)
8506 {
8507 int which;
8508 struct type *templ_type;
8509 struct type *var_type;
8510
8511 if (var_type0->code () == TYPE_CODE_PTR)
8512 var_type = TYPE_TARGET_TYPE (var_type0);
8513 else
8514 var_type = var_type0;
8515
8516 templ_type = ada_find_parallel_type (var_type, "___XVU");
8517
8518 if (templ_type != NULL)
8519 var_type = templ_type;
8520
8521 if (is_unchecked_variant (var_type, value_type (dval)))
8522 return var_type0;
8523 which = ada_which_variant_applies (var_type, dval);
8524
8525 if (which < 0)
8526 return empty_record (var_type);
8527 else if (is_dynamic_field (var_type, which))
8528 return to_fixed_record_type
8529 (TYPE_TARGET_TYPE (TYPE_FIELD_TYPE (var_type, which)),
8530 valaddr, address, dval);
8531 else if (variant_field_index (TYPE_FIELD_TYPE (var_type, which)) >= 0)
8532 return
8533 to_fixed_record_type
8534 (TYPE_FIELD_TYPE (var_type, which), valaddr, address, dval);
8535 else
8536 return TYPE_FIELD_TYPE (var_type, which);
8537 }
8538
8539 /* Assuming RANGE_TYPE is a TYPE_CODE_RANGE, return nonzero if
8540 ENCODING_TYPE, a type following the GNAT conventions for discrete
8541 type encodings, only carries redundant information. */
8542
8543 static int
8544 ada_is_redundant_range_encoding (struct type *range_type,
8545 struct type *encoding_type)
8546 {
8547 const char *bounds_str;
8548 int n;
8549 LONGEST lo, hi;
8550
8551 gdb_assert (range_type->code () == TYPE_CODE_RANGE);
8552
8553 if (get_base_type (range_type)->code ()
8554 != get_base_type (encoding_type)->code ())
8555 {
8556 /* The compiler probably used a simple base type to describe
8557 the range type instead of the range's actual base type,
8558 expecting us to get the real base type from the encoding
8559 anyway. In this situation, the encoding cannot be ignored
8560 as redundant. */
8561 return 0;
8562 }
8563
8564 if (is_dynamic_type (range_type))
8565 return 0;
8566
8567 if (encoding_type->name () == NULL)
8568 return 0;
8569
8570 bounds_str = strstr (encoding_type->name (), "___XDLU_");
8571 if (bounds_str == NULL)
8572 return 0;
8573
8574 n = 8; /* Skip "___XDLU_". */
8575 if (!ada_scan_number (bounds_str, n, &lo, &n))
8576 return 0;
8577 if (TYPE_LOW_BOUND (range_type) != lo)
8578 return 0;
8579
8580 n += 2; /* Skip the "__" separator between the two bounds. */
8581 if (!ada_scan_number (bounds_str, n, &hi, &n))
8582 return 0;
8583 if (TYPE_HIGH_BOUND (range_type) != hi)
8584 return 0;
8585
8586 return 1;
8587 }
8588
8589 /* Given the array type ARRAY_TYPE, return nonzero if DESC_TYPE,
8590 a type following the GNAT encoding for describing array type
8591 indices, only carries redundant information. */
8592
8593 static int
8594 ada_is_redundant_index_type_desc (struct type *array_type,
8595 struct type *desc_type)
8596 {
8597 struct type *this_layer = check_typedef (array_type);
8598 int i;
8599
8600 for (i = 0; i < desc_type->num_fields (); i++)
8601 {
8602 if (!ada_is_redundant_range_encoding (TYPE_INDEX_TYPE (this_layer),
8603 TYPE_FIELD_TYPE (desc_type, i)))
8604 return 0;
8605 this_layer = check_typedef (TYPE_TARGET_TYPE (this_layer));
8606 }
8607
8608 return 1;
8609 }
8610
8611 /* Assuming that TYPE0 is an array type describing the type of a value
8612 at ADDR, and that DVAL describes a record containing any
8613 discriminants used in TYPE0, returns a type for the value that
8614 contains no dynamic components (that is, no components whose sizes
8615 are determined by run-time quantities). Unless IGNORE_TOO_BIG is
8616 true, gives an error message if the resulting type's size is over
8617 varsize_limit. */
8618
8619 static struct type *
8620 to_fixed_array_type (struct type *type0, struct value *dval,
8621 int ignore_too_big)
8622 {
8623 struct type *index_type_desc;
8624 struct type *result;
8625 int constrained_packed_array_p;
8626 static const char *xa_suffix = "___XA";
8627
8628 type0 = ada_check_typedef (type0);
8629 if (TYPE_FIXED_INSTANCE (type0))
8630 return type0;
8631
8632 constrained_packed_array_p = ada_is_constrained_packed_array_type (type0);
8633 if (constrained_packed_array_p)
8634 type0 = decode_constrained_packed_array_type (type0);
8635
8636 index_type_desc = ada_find_parallel_type (type0, xa_suffix);
8637
8638 /* As mentioned in exp_dbug.ads, for non bit-packed arrays an
8639 encoding suffixed with 'P' may still be generated. If so,
8640 it should be used to find the XA type. */
8641
8642 if (index_type_desc == NULL)
8643 {
8644 const char *type_name = ada_type_name (type0);
8645
8646 if (type_name != NULL)
8647 {
8648 const int len = strlen (type_name);
8649 char *name = (char *) alloca (len + strlen (xa_suffix));
8650
8651 if (type_name[len - 1] == 'P')
8652 {
8653 strcpy (name, type_name);
8654 strcpy (name + len - 1, xa_suffix);
8655 index_type_desc = ada_find_parallel_type_with_name (type0, name);
8656 }
8657 }
8658 }
8659
8660 ada_fixup_array_indexes_type (index_type_desc);
8661 if (index_type_desc != NULL
8662 && ada_is_redundant_index_type_desc (type0, index_type_desc))
8663 {
8664 /* Ignore this ___XA parallel type, as it does not bring any
8665 useful information. This allows us to avoid creating fixed
8666 versions of the array's index types, which would be identical
8667 to the original ones. This, in turn, can also help avoid
8668 the creation of fixed versions of the array itself. */
8669 index_type_desc = NULL;
8670 }
8671
8672 if (index_type_desc == NULL)
8673 {
8674 struct type *elt_type0 = ada_check_typedef (TYPE_TARGET_TYPE (type0));
8675
8676 /* NOTE: elt_type---the fixed version of elt_type0---should never
8677 depend on the contents of the array in properly constructed
8678 debugging data. */
8679 /* Create a fixed version of the array element type.
8680 We're not providing the address of an element here,
8681 and thus the actual object value cannot be inspected to do
8682 the conversion. This should not be a problem, since arrays of
8683 unconstrained objects are not allowed. In particular, all
8684 the elements of an array of a tagged type should all be of
8685 the same type specified in the debugging info. No need to
8686 consult the object tag. */
8687 struct type *elt_type = ada_to_fixed_type (elt_type0, 0, 0, dval, 1);
8688
8689 /* Make sure we always create a new array type when dealing with
8690 packed array types, since we're going to fix-up the array
8691 type length and element bitsize a little further down. */
8692 if (elt_type0 == elt_type && !constrained_packed_array_p)
8693 result = type0;
8694 else
8695 result = create_array_type (alloc_type_copy (type0),
8696 elt_type, TYPE_INDEX_TYPE (type0));
8697 }
8698 else
8699 {
8700 int i;
8701 struct type *elt_type0;
8702
8703 elt_type0 = type0;
8704 for (i = index_type_desc->num_fields (); i > 0; i -= 1)
8705 elt_type0 = TYPE_TARGET_TYPE (elt_type0);
8706
8707 /* NOTE: result---the fixed version of elt_type0---should never
8708 depend on the contents of the array in properly constructed
8709 debugging data. */
8710 /* Create a fixed version of the array element type.
8711 We're not providing the address of an element here,
8712 and thus the actual object value cannot be inspected to do
8713 the conversion. This should not be a problem, since arrays of
8714 unconstrained objects are not allowed. In particular, all
8715 the elements of an array of a tagged type should all be of
8716 the same type specified in the debugging info. No need to
8717 consult the object tag. */
8718 result =
8719 ada_to_fixed_type (ada_check_typedef (elt_type0), 0, 0, dval, 1);
8720
8721 elt_type0 = type0;
8722 for (i = index_type_desc->num_fields () - 1; i >= 0; i -= 1)
8723 {
8724 struct type *range_type =
8725 to_fixed_range_type (TYPE_FIELD_TYPE (index_type_desc, i), dval);
8726
8727 result = create_array_type (alloc_type_copy (elt_type0),
8728 result, range_type);
8729 elt_type0 = TYPE_TARGET_TYPE (elt_type0);
8730 }
8731 if (!ignore_too_big && TYPE_LENGTH (result) > varsize_limit)
8732 error (_("array type with dynamic size is larger than varsize-limit"));
8733 }
8734
8735 /* We want to preserve the type name. This can be useful when
8736 trying to get the type name of a value that has already been
8737 printed (for instance, if the user did "print VAR; whatis $". */
8738 result->set_name (type0->name ());
8739
8740 if (constrained_packed_array_p)
8741 {
8742 /* So far, the resulting type has been created as if the original
8743 type was a regular (non-packed) array type. As a result, the
8744 bitsize of the array elements needs to be set again, and the array
8745 length needs to be recomputed based on that bitsize. */
8746 int len = TYPE_LENGTH (result) / TYPE_LENGTH (TYPE_TARGET_TYPE (result));
8747 int elt_bitsize = TYPE_FIELD_BITSIZE (type0, 0);
8748
8749 TYPE_FIELD_BITSIZE (result, 0) = TYPE_FIELD_BITSIZE (type0, 0);
8750 TYPE_LENGTH (result) = len * elt_bitsize / HOST_CHAR_BIT;
8751 if (TYPE_LENGTH (result) * HOST_CHAR_BIT < len * elt_bitsize)
8752 TYPE_LENGTH (result)++;
8753 }
8754
8755 TYPE_FIXED_INSTANCE (result) = 1;
8756 return result;
8757 }
8758
8759
8760 /* A standard type (containing no dynamically sized components)
8761 corresponding to TYPE for the value (TYPE, VALADDR, ADDRESS)
8762 DVAL describes a record containing any discriminants used in TYPE0,
8763 and may be NULL if there are none, or if the object of type TYPE at
8764 ADDRESS or in VALADDR contains these discriminants.
8765
8766 If CHECK_TAG is not null, in the case of tagged types, this function
8767 attempts to locate the object's tag and use it to compute the actual
8768 type. However, when ADDRESS is null, we cannot use it to determine the
8769 location of the tag, and therefore compute the tagged type's actual type.
8770 So we return the tagged type without consulting the tag. */
8771
8772 static struct type *
8773 ada_to_fixed_type_1 (struct type *type, const gdb_byte *valaddr,
8774 CORE_ADDR address, struct value *dval, int check_tag)
8775 {
8776 type = ada_check_typedef (type);
8777
8778 /* Only un-fixed types need to be handled here. */
8779 if (!HAVE_GNAT_AUX_INFO (type))
8780 return type;
8781
8782 switch (type->code ())
8783 {
8784 default:
8785 return type;
8786 case TYPE_CODE_STRUCT:
8787 {
8788 struct type *static_type = to_static_fixed_type (type);
8789 struct type *fixed_record_type =
8790 to_fixed_record_type (type, valaddr, address, NULL);
8791
8792 /* If STATIC_TYPE is a tagged type and we know the object's address,
8793 then we can determine its tag, and compute the object's actual
8794 type from there. Note that we have to use the fixed record
8795 type (the parent part of the record may have dynamic fields
8796 and the way the location of _tag is expressed may depend on
8797 them). */
8798
8799 if (check_tag && address != 0 && ada_is_tagged_type (static_type, 0))
8800 {
8801 struct value *tag =
8802 value_tag_from_contents_and_address
8803 (fixed_record_type,
8804 valaddr,
8805 address);
8806 struct type *real_type = type_from_tag (tag);
8807 struct value *obj =
8808 value_from_contents_and_address (fixed_record_type,
8809 valaddr,
8810 address);
8811 fixed_record_type = value_type (obj);
8812 if (real_type != NULL)
8813 return to_fixed_record_type
8814 (real_type, NULL,
8815 value_address (ada_tag_value_at_base_address (obj)), NULL);
8816 }
8817
8818 /* Check to see if there is a parallel ___XVZ variable.
8819 If there is, then it provides the actual size of our type. */
8820 else if (ada_type_name (fixed_record_type) != NULL)
8821 {
8822 const char *name = ada_type_name (fixed_record_type);
8823 char *xvz_name
8824 = (char *) alloca (strlen (name) + 7 /* "___XVZ\0" */);
8825 bool xvz_found = false;
8826 LONGEST size;
8827
8828 xsnprintf (xvz_name, strlen (name) + 7, "%s___XVZ", name);
8829 try
8830 {
8831 xvz_found = get_int_var_value (xvz_name, size);
8832 }
8833 catch (const gdb_exception_error &except)
8834 {
8835 /* We found the variable, but somehow failed to read
8836 its value. Rethrow the same error, but with a little
8837 bit more information, to help the user understand
8838 what went wrong (Eg: the variable might have been
8839 optimized out). */
8840 throw_error (except.error,
8841 _("unable to read value of %s (%s)"),
8842 xvz_name, except.what ());
8843 }
8844
8845 if (xvz_found && TYPE_LENGTH (fixed_record_type) != size)
8846 {
8847 fixed_record_type = copy_type (fixed_record_type);
8848 TYPE_LENGTH (fixed_record_type) = size;
8849
8850 /* The FIXED_RECORD_TYPE may have be a stub. We have
8851 observed this when the debugging info is STABS, and
8852 apparently it is something that is hard to fix.
8853
8854 In practice, we don't need the actual type definition
8855 at all, because the presence of the XVZ variable allows us
8856 to assume that there must be a XVS type as well, which we
8857 should be able to use later, when we need the actual type
8858 definition.
8859
8860 In the meantime, pretend that the "fixed" type we are
8861 returning is NOT a stub, because this can cause trouble
8862 when using this type to create new types targeting it.
8863 Indeed, the associated creation routines often check
8864 whether the target type is a stub and will try to replace
8865 it, thus using a type with the wrong size. This, in turn,
8866 might cause the new type to have the wrong size too.
8867 Consider the case of an array, for instance, where the size
8868 of the array is computed from the number of elements in
8869 our array multiplied by the size of its element. */
8870 TYPE_STUB (fixed_record_type) = 0;
8871 }
8872 }
8873 return fixed_record_type;
8874 }
8875 case TYPE_CODE_ARRAY:
8876 return to_fixed_array_type (type, dval, 1);
8877 case TYPE_CODE_UNION:
8878 if (dval == NULL)
8879 return type;
8880 else
8881 return to_fixed_variant_branch_type (type, valaddr, address, dval);
8882 }
8883 }
8884
8885 /* The same as ada_to_fixed_type_1, except that it preserves the type
8886 if it is a TYPE_CODE_TYPEDEF of a type that is already fixed.
8887
8888 The typedef layer needs be preserved in order to differentiate between
8889 arrays and array pointers when both types are implemented using the same
8890 fat pointer. In the array pointer case, the pointer is encoded as
8891 a typedef of the pointer type. For instance, considering:
8892
8893 type String_Access is access String;
8894 S1 : String_Access := null;
8895
8896 To the debugger, S1 is defined as a typedef of type String. But
8897 to the user, it is a pointer. So if the user tries to print S1,
8898 we should not dereference the array, but print the array address
8899 instead.
8900
8901 If we didn't preserve the typedef layer, we would lose the fact that
8902 the type is to be presented as a pointer (needs de-reference before
8903 being printed). And we would also use the source-level type name. */
8904
8905 struct type *
8906 ada_to_fixed_type (struct type *type, const gdb_byte *valaddr,
8907 CORE_ADDR address, struct value *dval, int check_tag)
8908
8909 {
8910 struct type *fixed_type =
8911 ada_to_fixed_type_1 (type, valaddr, address, dval, check_tag);
8912
8913 /* If TYPE is a typedef and its target type is the same as the FIXED_TYPE,
8914 then preserve the typedef layer.
8915
8916 Implementation note: We can only check the main-type portion of
8917 the TYPE and FIXED_TYPE, because eliminating the typedef layer
8918 from TYPE now returns a type that has the same instance flags
8919 as TYPE. For instance, if TYPE is a "typedef const", and its
8920 target type is a "struct", then the typedef elimination will return
8921 a "const" version of the target type. See check_typedef for more
8922 details about how the typedef layer elimination is done.
8923
8924 brobecker/2010-11-19: It seems to me that the only case where it is
8925 useful to preserve the typedef layer is when dealing with fat pointers.
8926 Perhaps, we could add a check for that and preserve the typedef layer
8927 only in that situation. But this seems unnecessary so far, probably
8928 because we call check_typedef/ada_check_typedef pretty much everywhere.
8929 */
8930 if (type->code () == TYPE_CODE_TYPEDEF
8931 && (TYPE_MAIN_TYPE (ada_typedef_target_type (type))
8932 == TYPE_MAIN_TYPE (fixed_type)))
8933 return type;
8934
8935 return fixed_type;
8936 }
8937
8938 /* A standard (static-sized) type corresponding as well as possible to
8939 TYPE0, but based on no runtime data. */
8940
8941 static struct type *
8942 to_static_fixed_type (struct type *type0)
8943 {
8944 struct type *type;
8945
8946 if (type0 == NULL)
8947 return NULL;
8948
8949 if (TYPE_FIXED_INSTANCE (type0))
8950 return type0;
8951
8952 type0 = ada_check_typedef (type0);
8953
8954 switch (type0->code ())
8955 {
8956 default:
8957 return type0;
8958 case TYPE_CODE_STRUCT:
8959 type = dynamic_template_type (type0);
8960 if (type != NULL)
8961 return template_to_static_fixed_type (type);
8962 else
8963 return template_to_static_fixed_type (type0);
8964 case TYPE_CODE_UNION:
8965 type = ada_find_parallel_type (type0, "___XVU");
8966 if (type != NULL)
8967 return template_to_static_fixed_type (type);
8968 else
8969 return template_to_static_fixed_type (type0);
8970 }
8971 }
8972
8973 /* A static approximation of TYPE with all type wrappers removed. */
8974
8975 static struct type *
8976 static_unwrap_type (struct type *type)
8977 {
8978 if (ada_is_aligner_type (type))
8979 {
8980 struct type *type1 = TYPE_FIELD_TYPE (ada_check_typedef (type), 0);
8981 if (ada_type_name (type1) == NULL)
8982 type1->set_name (ada_type_name (type));
8983
8984 return static_unwrap_type (type1);
8985 }
8986 else
8987 {
8988 struct type *raw_real_type = ada_get_base_type (type);
8989
8990 if (raw_real_type == type)
8991 return type;
8992 else
8993 return to_static_fixed_type (raw_real_type);
8994 }
8995 }
8996
8997 /* In some cases, incomplete and private types require
8998 cross-references that are not resolved as records (for example,
8999 type Foo;
9000 type FooP is access Foo;
9001 V: FooP;
9002 type Foo is array ...;
9003 ). In these cases, since there is no mechanism for producing
9004 cross-references to such types, we instead substitute for FooP a
9005 stub enumeration type that is nowhere resolved, and whose tag is
9006 the name of the actual type. Call these types "non-record stubs". */
9007
9008 /* A type equivalent to TYPE that is not a non-record stub, if one
9009 exists, otherwise TYPE. */
9010
9011 struct type *
9012 ada_check_typedef (struct type *type)
9013 {
9014 if (type == NULL)
9015 return NULL;
9016
9017 /* If our type is an access to an unconstrained array, which is encoded
9018 as a TYPE_CODE_TYPEDEF of a fat pointer, then we're done.
9019 We don't want to strip the TYPE_CODE_TYPDEF layer, because this is
9020 what allows us to distinguish between fat pointers that represent
9021 array types, and fat pointers that represent array access types
9022 (in both cases, the compiler implements them as fat pointers). */
9023 if (ada_is_access_to_unconstrained_array (type))
9024 return type;
9025
9026 type = check_typedef (type);
9027 if (type == NULL || type->code () != TYPE_CODE_ENUM
9028 || !TYPE_STUB (type)
9029 || type->name () == NULL)
9030 return type;
9031 else
9032 {
9033 const char *name = type->name ();
9034 struct type *type1 = ada_find_any_type (name);
9035
9036 if (type1 == NULL)
9037 return type;
9038
9039 /* TYPE1 might itself be a TYPE_CODE_TYPEDEF (this can happen with
9040 stubs pointing to arrays, as we don't create symbols for array
9041 types, only for the typedef-to-array types). If that's the case,
9042 strip the typedef layer. */
9043 if (type1->code () == TYPE_CODE_TYPEDEF)
9044 type1 = ada_check_typedef (type1);
9045
9046 return type1;
9047 }
9048 }
9049
9050 /* A value representing the data at VALADDR/ADDRESS as described by
9051 type TYPE0, but with a standard (static-sized) type that correctly
9052 describes it. If VAL0 is not NULL and TYPE0 already is a standard
9053 type, then return VAL0 [this feature is simply to avoid redundant
9054 creation of struct values]. */
9055
9056 static struct value *
9057 ada_to_fixed_value_create (struct type *type0, CORE_ADDR address,
9058 struct value *val0)
9059 {
9060 struct type *type = ada_to_fixed_type (type0, 0, address, NULL, 1);
9061
9062 if (type == type0 && val0 != NULL)
9063 return val0;
9064
9065 if (VALUE_LVAL (val0) != lval_memory)
9066 {
9067 /* Our value does not live in memory; it could be a convenience
9068 variable, for instance. Create a not_lval value using val0's
9069 contents. */
9070 return value_from_contents (type, value_contents (val0));
9071 }
9072
9073 return value_from_contents_and_address (type, 0, address);
9074 }
9075
9076 /* A value representing VAL, but with a standard (static-sized) type
9077 that correctly describes it. Does not necessarily create a new
9078 value. */
9079
9080 struct value *
9081 ada_to_fixed_value (struct value *val)
9082 {
9083 val = unwrap_value (val);
9084 val = ada_to_fixed_value_create (value_type (val), value_address (val), val);
9085 return val;
9086 }
9087 \f
9088
9089 /* Attributes */
9090
9091 /* Table mapping attribute numbers to names.
9092 NOTE: Keep up to date with enum ada_attribute definition in ada-lang.h. */
9093
9094 static const char *attribute_names[] = {
9095 "<?>",
9096
9097 "first",
9098 "last",
9099 "length",
9100 "image",
9101 "max",
9102 "min",
9103 "modulus",
9104 "pos",
9105 "size",
9106 "tag",
9107 "val",
9108 0
9109 };
9110
9111 static const char *
9112 ada_attribute_name (enum exp_opcode n)
9113 {
9114 if (n >= OP_ATR_FIRST && n <= (int) OP_ATR_VAL)
9115 return attribute_names[n - OP_ATR_FIRST + 1];
9116 else
9117 return attribute_names[0];
9118 }
9119
9120 /* Evaluate the 'POS attribute applied to ARG. */
9121
9122 static LONGEST
9123 pos_atr (struct value *arg)
9124 {
9125 struct value *val = coerce_ref (arg);
9126 struct type *type = value_type (val);
9127 LONGEST result;
9128
9129 if (!discrete_type_p (type))
9130 error (_("'POS only defined on discrete types"));
9131
9132 if (!discrete_position (type, value_as_long (val), &result))
9133 error (_("enumeration value is invalid: can't find 'POS"));
9134
9135 return result;
9136 }
9137
9138 static struct value *
9139 value_pos_atr (struct type *type, struct value *arg)
9140 {
9141 return value_from_longest (type, pos_atr (arg));
9142 }
9143
9144 /* Evaluate the TYPE'VAL attribute applied to ARG. */
9145
9146 static struct value *
9147 val_atr (struct type *type, LONGEST val)
9148 {
9149 gdb_assert (discrete_type_p (type));
9150 if (type->code () == TYPE_CODE_RANGE)
9151 type = TYPE_TARGET_TYPE (type);
9152 if (type->code () == TYPE_CODE_ENUM)
9153 {
9154 if (val < 0 || val >= type->num_fields ())
9155 error (_("argument to 'VAL out of range"));
9156 val = TYPE_FIELD_ENUMVAL (type, val);
9157 }
9158 return value_from_longest (type, val);
9159 }
9160
9161 static struct value *
9162 value_val_atr (struct type *type, struct value *arg)
9163 {
9164 if (!discrete_type_p (type))
9165 error (_("'VAL only defined on discrete types"));
9166 if (!integer_type_p (value_type (arg)))
9167 error (_("'VAL requires integral argument"));
9168
9169 return val_atr (type, value_as_long (arg));
9170 }
9171 \f
9172
9173 /* Evaluation */
9174
9175 /* True if TYPE appears to be an Ada character type.
9176 [At the moment, this is true only for Character and Wide_Character;
9177 It is a heuristic test that could stand improvement]. */
9178
9179 bool
9180 ada_is_character_type (struct type *type)
9181 {
9182 const char *name;
9183
9184 /* If the type code says it's a character, then assume it really is,
9185 and don't check any further. */
9186 if (type->code () == TYPE_CODE_CHAR)
9187 return true;
9188
9189 /* Otherwise, assume it's a character type iff it is a discrete type
9190 with a known character type name. */
9191 name = ada_type_name (type);
9192 return (name != NULL
9193 && (type->code () == TYPE_CODE_INT
9194 || type->code () == TYPE_CODE_RANGE)
9195 && (strcmp (name, "character") == 0
9196 || strcmp (name, "wide_character") == 0
9197 || strcmp (name, "wide_wide_character") == 0
9198 || strcmp (name, "unsigned char") == 0));
9199 }
9200
9201 /* True if TYPE appears to be an Ada string type. */
9202
9203 bool
9204 ada_is_string_type (struct type *type)
9205 {
9206 type = ada_check_typedef (type);
9207 if (type != NULL
9208 && type->code () != TYPE_CODE_PTR
9209 && (ada_is_simple_array_type (type)
9210 || ada_is_array_descriptor_type (type))
9211 && ada_array_arity (type) == 1)
9212 {
9213 struct type *elttype = ada_array_element_type (type, 1);
9214
9215 return ada_is_character_type (elttype);
9216 }
9217 else
9218 return false;
9219 }
9220
9221 /* The compiler sometimes provides a parallel XVS type for a given
9222 PAD type. Normally, it is safe to follow the PAD type directly,
9223 but older versions of the compiler have a bug that causes the offset
9224 of its "F" field to be wrong. Following that field in that case
9225 would lead to incorrect results, but this can be worked around
9226 by ignoring the PAD type and using the associated XVS type instead.
9227
9228 Set to True if the debugger should trust the contents of PAD types.
9229 Otherwise, ignore the PAD type if there is a parallel XVS type. */
9230 static bool trust_pad_over_xvs = true;
9231
9232 /* True if TYPE is a struct type introduced by the compiler to force the
9233 alignment of a value. Such types have a single field with a
9234 distinctive name. */
9235
9236 int
9237 ada_is_aligner_type (struct type *type)
9238 {
9239 type = ada_check_typedef (type);
9240
9241 if (!trust_pad_over_xvs && ada_find_parallel_type (type, "___XVS") != NULL)
9242 return 0;
9243
9244 return (type->code () == TYPE_CODE_STRUCT
9245 && type->num_fields () == 1
9246 && strcmp (TYPE_FIELD_NAME (type, 0), "F") == 0);
9247 }
9248
9249 /* If there is an ___XVS-convention type parallel to SUBTYPE, return
9250 the parallel type. */
9251
9252 struct type *
9253 ada_get_base_type (struct type *raw_type)
9254 {
9255 struct type *real_type_namer;
9256 struct type *raw_real_type;
9257
9258 if (raw_type == NULL || raw_type->code () != TYPE_CODE_STRUCT)
9259 return raw_type;
9260
9261 if (ada_is_aligner_type (raw_type))
9262 /* The encoding specifies that we should always use the aligner type.
9263 So, even if this aligner type has an associated XVS type, we should
9264 simply ignore it.
9265
9266 According to the compiler gurus, an XVS type parallel to an aligner
9267 type may exist because of a stabs limitation. In stabs, aligner
9268 types are empty because the field has a variable-sized type, and
9269 thus cannot actually be used as an aligner type. As a result,
9270 we need the associated parallel XVS type to decode the type.
9271 Since the policy in the compiler is to not change the internal
9272 representation based on the debugging info format, we sometimes
9273 end up having a redundant XVS type parallel to the aligner type. */
9274 return raw_type;
9275
9276 real_type_namer = ada_find_parallel_type (raw_type, "___XVS");
9277 if (real_type_namer == NULL
9278 || real_type_namer->code () != TYPE_CODE_STRUCT
9279 || real_type_namer->num_fields () != 1)
9280 return raw_type;
9281
9282 if (TYPE_FIELD_TYPE (real_type_namer, 0)->code () != TYPE_CODE_REF)
9283 {
9284 /* This is an older encoding form where the base type needs to be
9285 looked up by name. We prefer the newer encoding because it is
9286 more efficient. */
9287 raw_real_type = ada_find_any_type (TYPE_FIELD_NAME (real_type_namer, 0));
9288 if (raw_real_type == NULL)
9289 return raw_type;
9290 else
9291 return raw_real_type;
9292 }
9293
9294 /* The field in our XVS type is a reference to the base type. */
9295 return TYPE_TARGET_TYPE (TYPE_FIELD_TYPE (real_type_namer, 0));
9296 }
9297
9298 /* The type of value designated by TYPE, with all aligners removed. */
9299
9300 struct type *
9301 ada_aligned_type (struct type *type)
9302 {
9303 if (ada_is_aligner_type (type))
9304 return ada_aligned_type (TYPE_FIELD_TYPE (type, 0));
9305 else
9306 return ada_get_base_type (type);
9307 }
9308
9309
9310 /* The address of the aligned value in an object at address VALADDR
9311 having type TYPE. Assumes ada_is_aligner_type (TYPE). */
9312
9313 const gdb_byte *
9314 ada_aligned_value_addr (struct type *type, const gdb_byte *valaddr)
9315 {
9316 if (ada_is_aligner_type (type))
9317 return ada_aligned_value_addr (TYPE_FIELD_TYPE (type, 0),
9318 valaddr +
9319 TYPE_FIELD_BITPOS (type,
9320 0) / TARGET_CHAR_BIT);
9321 else
9322 return valaddr;
9323 }
9324
9325
9326
9327 /* The printed representation of an enumeration literal with encoded
9328 name NAME. The value is good to the next call of ada_enum_name. */
9329 const char *
9330 ada_enum_name (const char *name)
9331 {
9332 static char *result;
9333 static size_t result_len = 0;
9334 const char *tmp;
9335
9336 /* First, unqualify the enumeration name:
9337 1. Search for the last '.' character. If we find one, then skip
9338 all the preceding characters, the unqualified name starts
9339 right after that dot.
9340 2. Otherwise, we may be debugging on a target where the compiler
9341 translates dots into "__". Search forward for double underscores,
9342 but stop searching when we hit an overloading suffix, which is
9343 of the form "__" followed by digits. */
9344
9345 tmp = strrchr (name, '.');
9346 if (tmp != NULL)
9347 name = tmp + 1;
9348 else
9349 {
9350 while ((tmp = strstr (name, "__")) != NULL)
9351 {
9352 if (isdigit (tmp[2]))
9353 break;
9354 else
9355 name = tmp + 2;
9356 }
9357 }
9358
9359 if (name[0] == 'Q')
9360 {
9361 int v;
9362
9363 if (name[1] == 'U' || name[1] == 'W')
9364 {
9365 if (sscanf (name + 2, "%x", &v) != 1)
9366 return name;
9367 }
9368 else if (((name[1] >= '0' && name[1] <= '9')
9369 || (name[1] >= 'a' && name[1] <= 'z'))
9370 && name[2] == '\0')
9371 {
9372 GROW_VECT (result, result_len, 4);
9373 xsnprintf (result, result_len, "'%c'", name[1]);
9374 return result;
9375 }
9376 else
9377 return name;
9378
9379 GROW_VECT (result, result_len, 16);
9380 if (isascii (v) && isprint (v))
9381 xsnprintf (result, result_len, "'%c'", v);
9382 else if (name[1] == 'U')
9383 xsnprintf (result, result_len, "[\"%02x\"]", v);
9384 else
9385 xsnprintf (result, result_len, "[\"%04x\"]", v);
9386
9387 return result;
9388 }
9389 else
9390 {
9391 tmp = strstr (name, "__");
9392 if (tmp == NULL)
9393 tmp = strstr (name, "$");
9394 if (tmp != NULL)
9395 {
9396 GROW_VECT (result, result_len, tmp - name + 1);
9397 strncpy (result, name, tmp - name);
9398 result[tmp - name] = '\0';
9399 return result;
9400 }
9401
9402 return name;
9403 }
9404 }
9405
9406 /* Evaluate the subexpression of EXP starting at *POS as for
9407 evaluate_type, updating *POS to point just past the evaluated
9408 expression. */
9409
9410 static struct value *
9411 evaluate_subexp_type (struct expression *exp, int *pos)
9412 {
9413 return evaluate_subexp (NULL_TYPE, exp, pos, EVAL_AVOID_SIDE_EFFECTS);
9414 }
9415
9416 /* If VAL is wrapped in an aligner or subtype wrapper, return the
9417 value it wraps. */
9418
9419 static struct value *
9420 unwrap_value (struct value *val)
9421 {
9422 struct type *type = ada_check_typedef (value_type (val));
9423
9424 if (ada_is_aligner_type (type))
9425 {
9426 struct value *v = ada_value_struct_elt (val, "F", 0);
9427 struct type *val_type = ada_check_typedef (value_type (v));
9428
9429 if (ada_type_name (val_type) == NULL)
9430 val_type->set_name (ada_type_name (type));
9431
9432 return unwrap_value (v);
9433 }
9434 else
9435 {
9436 struct type *raw_real_type =
9437 ada_check_typedef (ada_get_base_type (type));
9438
9439 /* If there is no parallel XVS or XVE type, then the value is
9440 already unwrapped. Return it without further modification. */
9441 if ((type == raw_real_type)
9442 && ada_find_parallel_type (type, "___XVE") == NULL)
9443 return val;
9444
9445 return
9446 coerce_unspec_val_to_type
9447 (val, ada_to_fixed_type (raw_real_type, 0,
9448 value_address (val),
9449 NULL, 1));
9450 }
9451 }
9452
9453 static struct value *
9454 cast_from_fixed (struct type *type, struct value *arg)
9455 {
9456 struct value *scale = ada_scaling_factor (value_type (arg));
9457 arg = value_cast (value_type (scale), arg);
9458
9459 arg = value_binop (arg, scale, BINOP_MUL);
9460 return value_cast (type, arg);
9461 }
9462
9463 static struct value *
9464 cast_to_fixed (struct type *type, struct value *arg)
9465 {
9466 if (type == value_type (arg))
9467 return arg;
9468
9469 struct value *scale = ada_scaling_factor (type);
9470 if (ada_is_gnat_encoded_fixed_point_type (value_type (arg)))
9471 arg = cast_from_fixed (value_type (scale), arg);
9472 else
9473 arg = value_cast (value_type (scale), arg);
9474
9475 arg = value_binop (arg, scale, BINOP_DIV);
9476 return value_cast (type, arg);
9477 }
9478
9479 /* Given two array types T1 and T2, return nonzero iff both arrays
9480 contain the same number of elements. */
9481
9482 static int
9483 ada_same_array_size_p (struct type *t1, struct type *t2)
9484 {
9485 LONGEST lo1, hi1, lo2, hi2;
9486
9487 /* Get the array bounds in order to verify that the size of
9488 the two arrays match. */
9489 if (!get_array_bounds (t1, &lo1, &hi1)
9490 || !get_array_bounds (t2, &lo2, &hi2))
9491 error (_("unable to determine array bounds"));
9492
9493 /* To make things easier for size comparison, normalize a bit
9494 the case of empty arrays by making sure that the difference
9495 between upper bound and lower bound is always -1. */
9496 if (lo1 > hi1)
9497 hi1 = lo1 - 1;
9498 if (lo2 > hi2)
9499 hi2 = lo2 - 1;
9500
9501 return (hi1 - lo1 == hi2 - lo2);
9502 }
9503
9504 /* Assuming that VAL is an array of integrals, and TYPE represents
9505 an array with the same number of elements, but with wider integral
9506 elements, return an array "casted" to TYPE. In practice, this
9507 means that the returned array is built by casting each element
9508 of the original array into TYPE's (wider) element type. */
9509
9510 static struct value *
9511 ada_promote_array_of_integrals (struct type *type, struct value *val)
9512 {
9513 struct type *elt_type = TYPE_TARGET_TYPE (type);
9514 LONGEST lo, hi;
9515 struct value *res;
9516 LONGEST i;
9517
9518 /* Verify that both val and type are arrays of scalars, and
9519 that the size of val's elements is smaller than the size
9520 of type's element. */
9521 gdb_assert (type->code () == TYPE_CODE_ARRAY);
9522 gdb_assert (is_integral_type (TYPE_TARGET_TYPE (type)));
9523 gdb_assert (value_type (val)->code () == TYPE_CODE_ARRAY);
9524 gdb_assert (is_integral_type (TYPE_TARGET_TYPE (value_type (val))));
9525 gdb_assert (TYPE_LENGTH (TYPE_TARGET_TYPE (type))
9526 > TYPE_LENGTH (TYPE_TARGET_TYPE (value_type (val))));
9527
9528 if (!get_array_bounds (type, &lo, &hi))
9529 error (_("unable to determine array bounds"));
9530
9531 res = allocate_value (type);
9532
9533 /* Promote each array element. */
9534 for (i = 0; i < hi - lo + 1; i++)
9535 {
9536 struct value *elt = value_cast (elt_type, value_subscript (val, lo + i));
9537
9538 memcpy (value_contents_writeable (res) + (i * TYPE_LENGTH (elt_type)),
9539 value_contents_all (elt), TYPE_LENGTH (elt_type));
9540 }
9541
9542 return res;
9543 }
9544
9545 /* Coerce VAL as necessary for assignment to an lval of type TYPE, and
9546 return the converted value. */
9547
9548 static struct value *
9549 coerce_for_assign (struct type *type, struct value *val)
9550 {
9551 struct type *type2 = value_type (val);
9552
9553 if (type == type2)
9554 return val;
9555
9556 type2 = ada_check_typedef (type2);
9557 type = ada_check_typedef (type);
9558
9559 if (type2->code () == TYPE_CODE_PTR
9560 && type->code () == TYPE_CODE_ARRAY)
9561 {
9562 val = ada_value_ind (val);
9563 type2 = value_type (val);
9564 }
9565
9566 if (type2->code () == TYPE_CODE_ARRAY
9567 && type->code () == TYPE_CODE_ARRAY)
9568 {
9569 if (!ada_same_array_size_p (type, type2))
9570 error (_("cannot assign arrays of different length"));
9571
9572 if (is_integral_type (TYPE_TARGET_TYPE (type))
9573 && is_integral_type (TYPE_TARGET_TYPE (type2))
9574 && TYPE_LENGTH (TYPE_TARGET_TYPE (type2))
9575 < TYPE_LENGTH (TYPE_TARGET_TYPE (type)))
9576 {
9577 /* Allow implicit promotion of the array elements to
9578 a wider type. */
9579 return ada_promote_array_of_integrals (type, val);
9580 }
9581
9582 if (TYPE_LENGTH (TYPE_TARGET_TYPE (type2))
9583 != TYPE_LENGTH (TYPE_TARGET_TYPE (type)))
9584 error (_("Incompatible types in assignment"));
9585 deprecated_set_value_type (val, type);
9586 }
9587 return val;
9588 }
9589
9590 static struct value *
9591 ada_value_binop (struct value *arg1, struct value *arg2, enum exp_opcode op)
9592 {
9593 struct value *val;
9594 struct type *type1, *type2;
9595 LONGEST v, v1, v2;
9596
9597 arg1 = coerce_ref (arg1);
9598 arg2 = coerce_ref (arg2);
9599 type1 = get_base_type (ada_check_typedef (value_type (arg1)));
9600 type2 = get_base_type (ada_check_typedef (value_type (arg2)));
9601
9602 if (type1->code () != TYPE_CODE_INT
9603 || type2->code () != TYPE_CODE_INT)
9604 return value_binop (arg1, arg2, op);
9605
9606 switch (op)
9607 {
9608 case BINOP_MOD:
9609 case BINOP_DIV:
9610 case BINOP_REM:
9611 break;
9612 default:
9613 return value_binop (arg1, arg2, op);
9614 }
9615
9616 v2 = value_as_long (arg2);
9617 if (v2 == 0)
9618 error (_("second operand of %s must not be zero."), op_string (op));
9619
9620 if (TYPE_UNSIGNED (type1) || op == BINOP_MOD)
9621 return value_binop (arg1, arg2, op);
9622
9623 v1 = value_as_long (arg1);
9624 switch (op)
9625 {
9626 case BINOP_DIV:
9627 v = v1 / v2;
9628 if (!TRUNCATION_TOWARDS_ZERO && v1 * (v1 % v2) < 0)
9629 v += v > 0 ? -1 : 1;
9630 break;
9631 case BINOP_REM:
9632 v = v1 % v2;
9633 if (v * v1 < 0)
9634 v -= v2;
9635 break;
9636 default:
9637 /* Should not reach this point. */
9638 v = 0;
9639 }
9640
9641 val = allocate_value (type1);
9642 store_unsigned_integer (value_contents_raw (val),
9643 TYPE_LENGTH (value_type (val)),
9644 type_byte_order (type1), v);
9645 return val;
9646 }
9647
9648 static int
9649 ada_value_equal (struct value *arg1, struct value *arg2)
9650 {
9651 if (ada_is_direct_array_type (value_type (arg1))
9652 || ada_is_direct_array_type (value_type (arg2)))
9653 {
9654 struct type *arg1_type, *arg2_type;
9655
9656 /* Automatically dereference any array reference before
9657 we attempt to perform the comparison. */
9658 arg1 = ada_coerce_ref (arg1);
9659 arg2 = ada_coerce_ref (arg2);
9660
9661 arg1 = ada_coerce_to_simple_array (arg1);
9662 arg2 = ada_coerce_to_simple_array (arg2);
9663
9664 arg1_type = ada_check_typedef (value_type (arg1));
9665 arg2_type = ada_check_typedef (value_type (arg2));
9666
9667 if (arg1_type->code () != TYPE_CODE_ARRAY
9668 || arg2_type->code () != TYPE_CODE_ARRAY)
9669 error (_("Attempt to compare array with non-array"));
9670 /* FIXME: The following works only for types whose
9671 representations use all bits (no padding or undefined bits)
9672 and do not have user-defined equality. */
9673 return (TYPE_LENGTH (arg1_type) == TYPE_LENGTH (arg2_type)
9674 && memcmp (value_contents (arg1), value_contents (arg2),
9675 TYPE_LENGTH (arg1_type)) == 0);
9676 }
9677 return value_equal (arg1, arg2);
9678 }
9679
9680 /* Total number of component associations in the aggregate starting at
9681 index PC in EXP. Assumes that index PC is the start of an
9682 OP_AGGREGATE. */
9683
9684 static int
9685 num_component_specs (struct expression *exp, int pc)
9686 {
9687 int n, m, i;
9688
9689 m = exp->elts[pc + 1].longconst;
9690 pc += 3;
9691 n = 0;
9692 for (i = 0; i < m; i += 1)
9693 {
9694 switch (exp->elts[pc].opcode)
9695 {
9696 default:
9697 n += 1;
9698 break;
9699 case OP_CHOICES:
9700 n += exp->elts[pc + 1].longconst;
9701 break;
9702 }
9703 ada_evaluate_subexp (NULL, exp, &pc, EVAL_SKIP);
9704 }
9705 return n;
9706 }
9707
9708 /* Assign the result of evaluating EXP starting at *POS to the INDEXth
9709 component of LHS (a simple array or a record), updating *POS past
9710 the expression, assuming that LHS is contained in CONTAINER. Does
9711 not modify the inferior's memory, nor does it modify LHS (unless
9712 LHS == CONTAINER). */
9713
9714 static void
9715 assign_component (struct value *container, struct value *lhs, LONGEST index,
9716 struct expression *exp, int *pos)
9717 {
9718 struct value *mark = value_mark ();
9719 struct value *elt;
9720 struct type *lhs_type = check_typedef (value_type (lhs));
9721
9722 if (lhs_type->code () == TYPE_CODE_ARRAY)
9723 {
9724 struct type *index_type = builtin_type (exp->gdbarch)->builtin_int;
9725 struct value *index_val = value_from_longest (index_type, index);
9726
9727 elt = unwrap_value (ada_value_subscript (lhs, 1, &index_val));
9728 }
9729 else
9730 {
9731 elt = ada_index_struct_field (index, lhs, 0, value_type (lhs));
9732 elt = ada_to_fixed_value (elt);
9733 }
9734
9735 if (exp->elts[*pos].opcode == OP_AGGREGATE)
9736 assign_aggregate (container, elt, exp, pos, EVAL_NORMAL);
9737 else
9738 value_assign_to_component (container, elt,
9739 ada_evaluate_subexp (NULL, exp, pos,
9740 EVAL_NORMAL));
9741
9742 value_free_to_mark (mark);
9743 }
9744
9745 /* Assuming that LHS represents an lvalue having a record or array
9746 type, and EXP->ELTS[*POS] is an OP_AGGREGATE, evaluate an assignment
9747 of that aggregate's value to LHS, advancing *POS past the
9748 aggregate. NOSIDE is as for evaluate_subexp. CONTAINER is an
9749 lvalue containing LHS (possibly LHS itself). Does not modify
9750 the inferior's memory, nor does it modify the contents of
9751 LHS (unless == CONTAINER). Returns the modified CONTAINER. */
9752
9753 static struct value *
9754 assign_aggregate (struct value *container,
9755 struct value *lhs, struct expression *exp,
9756 int *pos, enum noside noside)
9757 {
9758 struct type *lhs_type;
9759 int n = exp->elts[*pos+1].longconst;
9760 LONGEST low_index, high_index;
9761 int num_specs;
9762 LONGEST *indices;
9763 int max_indices, num_indices;
9764 int i;
9765
9766 *pos += 3;
9767 if (noside != EVAL_NORMAL)
9768 {
9769 for (i = 0; i < n; i += 1)
9770 ada_evaluate_subexp (NULL, exp, pos, noside);
9771 return container;
9772 }
9773
9774 container = ada_coerce_ref (container);
9775 if (ada_is_direct_array_type (value_type (container)))
9776 container = ada_coerce_to_simple_array (container);
9777 lhs = ada_coerce_ref (lhs);
9778 if (!deprecated_value_modifiable (lhs))
9779 error (_("Left operand of assignment is not a modifiable lvalue."));
9780
9781 lhs_type = check_typedef (value_type (lhs));
9782 if (ada_is_direct_array_type (lhs_type))
9783 {
9784 lhs = ada_coerce_to_simple_array (lhs);
9785 lhs_type = check_typedef (value_type (lhs));
9786 low_index = TYPE_ARRAY_LOWER_BOUND_VALUE (lhs_type);
9787 high_index = TYPE_ARRAY_UPPER_BOUND_VALUE (lhs_type);
9788 }
9789 else if (lhs_type->code () == TYPE_CODE_STRUCT)
9790 {
9791 low_index = 0;
9792 high_index = num_visible_fields (lhs_type) - 1;
9793 }
9794 else
9795 error (_("Left-hand side must be array or record."));
9796
9797 num_specs = num_component_specs (exp, *pos - 3);
9798 max_indices = 4 * num_specs + 4;
9799 indices = XALLOCAVEC (LONGEST, max_indices);
9800 indices[0] = indices[1] = low_index - 1;
9801 indices[2] = indices[3] = high_index + 1;
9802 num_indices = 4;
9803
9804 for (i = 0; i < n; i += 1)
9805 {
9806 switch (exp->elts[*pos].opcode)
9807 {
9808 case OP_CHOICES:
9809 aggregate_assign_from_choices (container, lhs, exp, pos, indices,
9810 &num_indices, max_indices,
9811 low_index, high_index);
9812 break;
9813 case OP_POSITIONAL:
9814 aggregate_assign_positional (container, lhs, exp, pos, indices,
9815 &num_indices, max_indices,
9816 low_index, high_index);
9817 break;
9818 case OP_OTHERS:
9819 if (i != n-1)
9820 error (_("Misplaced 'others' clause"));
9821 aggregate_assign_others (container, lhs, exp, pos, indices,
9822 num_indices, low_index, high_index);
9823 break;
9824 default:
9825 error (_("Internal error: bad aggregate clause"));
9826 }
9827 }
9828
9829 return container;
9830 }
9831
9832 /* Assign into the component of LHS indexed by the OP_POSITIONAL
9833 construct at *POS, updating *POS past the construct, given that
9834 the positions are relative to lower bound LOW, where HIGH is the
9835 upper bound. Record the position in INDICES[0 .. MAX_INDICES-1]
9836 updating *NUM_INDICES as needed. CONTAINER is as for
9837 assign_aggregate. */
9838 static void
9839 aggregate_assign_positional (struct value *container,
9840 struct value *lhs, struct expression *exp,
9841 int *pos, LONGEST *indices, int *num_indices,
9842 int max_indices, LONGEST low, LONGEST high)
9843 {
9844 LONGEST ind = longest_to_int (exp->elts[*pos + 1].longconst) + low;
9845
9846 if (ind - 1 == high)
9847 warning (_("Extra components in aggregate ignored."));
9848 if (ind <= high)
9849 {
9850 add_component_interval (ind, ind, indices, num_indices, max_indices);
9851 *pos += 3;
9852 assign_component (container, lhs, ind, exp, pos);
9853 }
9854 else
9855 ada_evaluate_subexp (NULL, exp, pos, EVAL_SKIP);
9856 }
9857
9858 /* Assign into the components of LHS indexed by the OP_CHOICES
9859 construct at *POS, updating *POS past the construct, given that
9860 the allowable indices are LOW..HIGH. Record the indices assigned
9861 to in INDICES[0 .. MAX_INDICES-1], updating *NUM_INDICES as
9862 needed. CONTAINER is as for assign_aggregate. */
9863 static void
9864 aggregate_assign_from_choices (struct value *container,
9865 struct value *lhs, struct expression *exp,
9866 int *pos, LONGEST *indices, int *num_indices,
9867 int max_indices, LONGEST low, LONGEST high)
9868 {
9869 int j;
9870 int n_choices = longest_to_int (exp->elts[*pos+1].longconst);
9871 int choice_pos, expr_pc;
9872 int is_array = ada_is_direct_array_type (value_type (lhs));
9873
9874 choice_pos = *pos += 3;
9875
9876 for (j = 0; j < n_choices; j += 1)
9877 ada_evaluate_subexp (NULL, exp, pos, EVAL_SKIP);
9878 expr_pc = *pos;
9879 ada_evaluate_subexp (NULL, exp, pos, EVAL_SKIP);
9880
9881 for (j = 0; j < n_choices; j += 1)
9882 {
9883 LONGEST lower, upper;
9884 enum exp_opcode op = exp->elts[choice_pos].opcode;
9885
9886 if (op == OP_DISCRETE_RANGE)
9887 {
9888 choice_pos += 1;
9889 lower = value_as_long (ada_evaluate_subexp (NULL, exp, pos,
9890 EVAL_NORMAL));
9891 upper = value_as_long (ada_evaluate_subexp (NULL, exp, pos,
9892 EVAL_NORMAL));
9893 }
9894 else if (is_array)
9895 {
9896 lower = value_as_long (ada_evaluate_subexp (NULL, exp, &choice_pos,
9897 EVAL_NORMAL));
9898 upper = lower;
9899 }
9900 else
9901 {
9902 int ind;
9903 const char *name;
9904
9905 switch (op)
9906 {
9907 case OP_NAME:
9908 name = &exp->elts[choice_pos + 2].string;
9909 break;
9910 case OP_VAR_VALUE:
9911 name = exp->elts[choice_pos + 2].symbol->natural_name ();
9912 break;
9913 default:
9914 error (_("Invalid record component association."));
9915 }
9916 ada_evaluate_subexp (NULL, exp, &choice_pos, EVAL_SKIP);
9917 ind = 0;
9918 if (! find_struct_field (name, value_type (lhs), 0,
9919 NULL, NULL, NULL, NULL, &ind))
9920 error (_("Unknown component name: %s."), name);
9921 lower = upper = ind;
9922 }
9923
9924 if (lower <= upper && (lower < low || upper > high))
9925 error (_("Index in component association out of bounds."));
9926
9927 add_component_interval (lower, upper, indices, num_indices,
9928 max_indices);
9929 while (lower <= upper)
9930 {
9931 int pos1;
9932
9933 pos1 = expr_pc;
9934 assign_component (container, lhs, lower, exp, &pos1);
9935 lower += 1;
9936 }
9937 }
9938 }
9939
9940 /* Assign the value of the expression in the OP_OTHERS construct in
9941 EXP at *POS into the components of LHS indexed from LOW .. HIGH that
9942 have not been previously assigned. The index intervals already assigned
9943 are in INDICES[0 .. NUM_INDICES-1]. Updates *POS to after the
9944 OP_OTHERS clause. CONTAINER is as for assign_aggregate. */
9945 static void
9946 aggregate_assign_others (struct value *container,
9947 struct value *lhs, struct expression *exp,
9948 int *pos, LONGEST *indices, int num_indices,
9949 LONGEST low, LONGEST high)
9950 {
9951 int i;
9952 int expr_pc = *pos + 1;
9953
9954 for (i = 0; i < num_indices - 2; i += 2)
9955 {
9956 LONGEST ind;
9957
9958 for (ind = indices[i + 1] + 1; ind < indices[i + 2]; ind += 1)
9959 {
9960 int localpos;
9961
9962 localpos = expr_pc;
9963 assign_component (container, lhs, ind, exp, &localpos);
9964 }
9965 }
9966 ada_evaluate_subexp (NULL, exp, pos, EVAL_SKIP);
9967 }
9968
9969 /* Add the interval [LOW .. HIGH] to the sorted set of intervals
9970 [ INDICES[0] .. INDICES[1] ],..., [ INDICES[*SIZE-2] .. INDICES[*SIZE-1] ],
9971 modifying *SIZE as needed. It is an error if *SIZE exceeds
9972 MAX_SIZE. The resulting intervals do not overlap. */
9973 static void
9974 add_component_interval (LONGEST low, LONGEST high,
9975 LONGEST* indices, int *size, int max_size)
9976 {
9977 int i, j;
9978
9979 for (i = 0; i < *size; i += 2) {
9980 if (high >= indices[i] && low <= indices[i + 1])
9981 {
9982 int kh;
9983
9984 for (kh = i + 2; kh < *size; kh += 2)
9985 if (high < indices[kh])
9986 break;
9987 if (low < indices[i])
9988 indices[i] = low;
9989 indices[i + 1] = indices[kh - 1];
9990 if (high > indices[i + 1])
9991 indices[i + 1] = high;
9992 memcpy (indices + i + 2, indices + kh, *size - kh);
9993 *size -= kh - i - 2;
9994 return;
9995 }
9996 else if (high < indices[i])
9997 break;
9998 }
9999
10000 if (*size == max_size)
10001 error (_("Internal error: miscounted aggregate components."));
10002 *size += 2;
10003 for (j = *size-1; j >= i+2; j -= 1)
10004 indices[j] = indices[j - 2];
10005 indices[i] = low;
10006 indices[i + 1] = high;
10007 }
10008
10009 /* Perform and Ada cast of ARG2 to type TYPE if the type of ARG2
10010 is different. */
10011
10012 static struct value *
10013 ada_value_cast (struct type *type, struct value *arg2)
10014 {
10015 if (type == ada_check_typedef (value_type (arg2)))
10016 return arg2;
10017
10018 if (ada_is_gnat_encoded_fixed_point_type (type))
10019 return cast_to_fixed (type, arg2);
10020
10021 if (ada_is_gnat_encoded_fixed_point_type (value_type (arg2)))
10022 return cast_from_fixed (type, arg2);
10023
10024 return value_cast (type, arg2);
10025 }
10026
10027 /* Evaluating Ada expressions, and printing their result.
10028 ------------------------------------------------------
10029
10030 1. Introduction:
10031 ----------------
10032
10033 We usually evaluate an Ada expression in order to print its value.
10034 We also evaluate an expression in order to print its type, which
10035 happens during the EVAL_AVOID_SIDE_EFFECTS phase of the evaluation,
10036 but we'll focus mostly on the EVAL_NORMAL phase. In practice, the
10037 EVAL_AVOID_SIDE_EFFECTS phase allows us to simplify certain aspects of
10038 the evaluation compared to the EVAL_NORMAL, but is otherwise very
10039 similar.
10040
10041 Evaluating expressions is a little more complicated for Ada entities
10042 than it is for entities in languages such as C. The main reason for
10043 this is that Ada provides types whose definition might be dynamic.
10044 One example of such types is variant records. Or another example
10045 would be an array whose bounds can only be known at run time.
10046
10047 The following description is a general guide as to what should be
10048 done (and what should NOT be done) in order to evaluate an expression
10049 involving such types, and when. This does not cover how the semantic
10050 information is encoded by GNAT as this is covered separatly. For the
10051 document used as the reference for the GNAT encoding, see exp_dbug.ads
10052 in the GNAT sources.
10053
10054 Ideally, we should embed each part of this description next to its
10055 associated code. Unfortunately, the amount of code is so vast right
10056 now that it's hard to see whether the code handling a particular
10057 situation might be duplicated or not. One day, when the code is
10058 cleaned up, this guide might become redundant with the comments
10059 inserted in the code, and we might want to remove it.
10060
10061 2. ``Fixing'' an Entity, the Simple Case:
10062 -----------------------------------------
10063
10064 When evaluating Ada expressions, the tricky issue is that they may
10065 reference entities whose type contents and size are not statically
10066 known. Consider for instance a variant record:
10067
10068 type Rec (Empty : Boolean := True) is record
10069 case Empty is
10070 when True => null;
10071 when False => Value : Integer;
10072 end case;
10073 end record;
10074 Yes : Rec := (Empty => False, Value => 1);
10075 No : Rec := (empty => True);
10076
10077 The size and contents of that record depends on the value of the
10078 descriminant (Rec.Empty). At this point, neither the debugging
10079 information nor the associated type structure in GDB are able to
10080 express such dynamic types. So what the debugger does is to create
10081 "fixed" versions of the type that applies to the specific object.
10082 We also informally refer to this operation as "fixing" an object,
10083 which means creating its associated fixed type.
10084
10085 Example: when printing the value of variable "Yes" above, its fixed
10086 type would look like this:
10087
10088 type Rec is record
10089 Empty : Boolean;
10090 Value : Integer;
10091 end record;
10092
10093 On the other hand, if we printed the value of "No", its fixed type
10094 would become:
10095
10096 type Rec is record
10097 Empty : Boolean;
10098 end record;
10099
10100 Things become a little more complicated when trying to fix an entity
10101 with a dynamic type that directly contains another dynamic type,
10102 such as an array of variant records, for instance. There are
10103 two possible cases: Arrays, and records.
10104
10105 3. ``Fixing'' Arrays:
10106 ---------------------
10107
10108 The type structure in GDB describes an array in terms of its bounds,
10109 and the type of its elements. By design, all elements in the array
10110 have the same type and we cannot represent an array of variant elements
10111 using the current type structure in GDB. When fixing an array,
10112 we cannot fix the array element, as we would potentially need one
10113 fixed type per element of the array. As a result, the best we can do
10114 when fixing an array is to produce an array whose bounds and size
10115 are correct (allowing us to read it from memory), but without having
10116 touched its element type. Fixing each element will be done later,
10117 when (if) necessary.
10118
10119 Arrays are a little simpler to handle than records, because the same
10120 amount of memory is allocated for each element of the array, even if
10121 the amount of space actually used by each element differs from element
10122 to element. Consider for instance the following array of type Rec:
10123
10124 type Rec_Array is array (1 .. 2) of Rec;
10125
10126 The actual amount of memory occupied by each element might be different
10127 from element to element, depending on the value of their discriminant.
10128 But the amount of space reserved for each element in the array remains
10129 fixed regardless. So we simply need to compute that size using
10130 the debugging information available, from which we can then determine
10131 the array size (we multiply the number of elements of the array by
10132 the size of each element).
10133
10134 The simplest case is when we have an array of a constrained element
10135 type. For instance, consider the following type declarations:
10136
10137 type Bounded_String (Max_Size : Integer) is
10138 Length : Integer;
10139 Buffer : String (1 .. Max_Size);
10140 end record;
10141 type Bounded_String_Array is array (1 ..2) of Bounded_String (80);
10142
10143 In this case, the compiler describes the array as an array of
10144 variable-size elements (identified by its XVS suffix) for which
10145 the size can be read in the parallel XVZ variable.
10146
10147 In the case of an array of an unconstrained element type, the compiler
10148 wraps the array element inside a private PAD type. This type should not
10149 be shown to the user, and must be "unwrap"'ed before printing. Note
10150 that we also use the adjective "aligner" in our code to designate
10151 these wrapper types.
10152
10153 In some cases, the size allocated for each element is statically
10154 known. In that case, the PAD type already has the correct size,
10155 and the array element should remain unfixed.
10156
10157 But there are cases when this size is not statically known.
10158 For instance, assuming that "Five" is an integer variable:
10159
10160 type Dynamic is array (1 .. Five) of Integer;
10161 type Wrapper (Has_Length : Boolean := False) is record
10162 Data : Dynamic;
10163 case Has_Length is
10164 when True => Length : Integer;
10165 when False => null;
10166 end case;
10167 end record;
10168 type Wrapper_Array is array (1 .. 2) of Wrapper;
10169
10170 Hello : Wrapper_Array := (others => (Has_Length => True,
10171 Data => (others => 17),
10172 Length => 1));
10173
10174
10175 The debugging info would describe variable Hello as being an
10176 array of a PAD type. The size of that PAD type is not statically
10177 known, but can be determined using a parallel XVZ variable.
10178 In that case, a copy of the PAD type with the correct size should
10179 be used for the fixed array.
10180
10181 3. ``Fixing'' record type objects:
10182 ----------------------------------
10183
10184 Things are slightly different from arrays in the case of dynamic
10185 record types. In this case, in order to compute the associated
10186 fixed type, we need to determine the size and offset of each of
10187 its components. This, in turn, requires us to compute the fixed
10188 type of each of these components.
10189
10190 Consider for instance the example:
10191
10192 type Bounded_String (Max_Size : Natural) is record
10193 Str : String (1 .. Max_Size);
10194 Length : Natural;
10195 end record;
10196 My_String : Bounded_String (Max_Size => 10);
10197
10198 In that case, the position of field "Length" depends on the size
10199 of field Str, which itself depends on the value of the Max_Size
10200 discriminant. In order to fix the type of variable My_String,
10201 we need to fix the type of field Str. Therefore, fixing a variant
10202 record requires us to fix each of its components.
10203
10204 However, if a component does not have a dynamic size, the component
10205 should not be fixed. In particular, fields that use a PAD type
10206 should not fixed. Here is an example where this might happen
10207 (assuming type Rec above):
10208
10209 type Container (Big : Boolean) is record
10210 First : Rec;
10211 After : Integer;
10212 case Big is
10213 when True => Another : Integer;
10214 when False => null;
10215 end case;
10216 end record;
10217 My_Container : Container := (Big => False,
10218 First => (Empty => True),
10219 After => 42);
10220
10221 In that example, the compiler creates a PAD type for component First,
10222 whose size is constant, and then positions the component After just
10223 right after it. The offset of component After is therefore constant
10224 in this case.
10225
10226 The debugger computes the position of each field based on an algorithm
10227 that uses, among other things, the actual position and size of the field
10228 preceding it. Let's now imagine that the user is trying to print
10229 the value of My_Container. If the type fixing was recursive, we would
10230 end up computing the offset of field After based on the size of the
10231 fixed version of field First. And since in our example First has
10232 only one actual field, the size of the fixed type is actually smaller
10233 than the amount of space allocated to that field, and thus we would
10234 compute the wrong offset of field After.
10235
10236 To make things more complicated, we need to watch out for dynamic
10237 components of variant records (identified by the ___XVL suffix in
10238 the component name). Even if the target type is a PAD type, the size
10239 of that type might not be statically known. So the PAD type needs
10240 to be unwrapped and the resulting type needs to be fixed. Otherwise,
10241 we might end up with the wrong size for our component. This can be
10242 observed with the following type declarations:
10243
10244 type Octal is new Integer range 0 .. 7;
10245 type Octal_Array is array (Positive range <>) of Octal;
10246 pragma Pack (Octal_Array);
10247
10248 type Octal_Buffer (Size : Positive) is record
10249 Buffer : Octal_Array (1 .. Size);
10250 Length : Integer;
10251 end record;
10252
10253 In that case, Buffer is a PAD type whose size is unset and needs
10254 to be computed by fixing the unwrapped type.
10255
10256 4. When to ``Fix'' un-``Fixed'' sub-elements of an entity:
10257 ----------------------------------------------------------
10258
10259 Lastly, when should the sub-elements of an entity that remained unfixed
10260 thus far, be actually fixed?
10261
10262 The answer is: Only when referencing that element. For instance
10263 when selecting one component of a record, this specific component
10264 should be fixed at that point in time. Or when printing the value
10265 of a record, each component should be fixed before its value gets
10266 printed. Similarly for arrays, the element of the array should be
10267 fixed when printing each element of the array, or when extracting
10268 one element out of that array. On the other hand, fixing should
10269 not be performed on the elements when taking a slice of an array!
10270
10271 Note that one of the side effects of miscomputing the offset and
10272 size of each field is that we end up also miscomputing the size
10273 of the containing type. This can have adverse results when computing
10274 the value of an entity. GDB fetches the value of an entity based
10275 on the size of its type, and thus a wrong size causes GDB to fetch
10276 the wrong amount of memory. In the case where the computed size is
10277 too small, GDB fetches too little data to print the value of our
10278 entity. Results in this case are unpredictable, as we usually read
10279 past the buffer containing the data =:-o. */
10280
10281 /* Evaluate a subexpression of EXP, at index *POS, and return a value
10282 for that subexpression cast to TO_TYPE. Advance *POS over the
10283 subexpression. */
10284
10285 static value *
10286 ada_evaluate_subexp_for_cast (expression *exp, int *pos,
10287 enum noside noside, struct type *to_type)
10288 {
10289 int pc = *pos;
10290
10291 if (exp->elts[pc].opcode == OP_VAR_MSYM_VALUE
10292 || exp->elts[pc].opcode == OP_VAR_VALUE)
10293 {
10294 (*pos) += 4;
10295
10296 value *val;
10297 if (exp->elts[pc].opcode == OP_VAR_MSYM_VALUE)
10298 {
10299 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10300 return value_zero (to_type, not_lval);
10301
10302 val = evaluate_var_msym_value (noside,
10303 exp->elts[pc + 1].objfile,
10304 exp->elts[pc + 2].msymbol);
10305 }
10306 else
10307 val = evaluate_var_value (noside,
10308 exp->elts[pc + 1].block,
10309 exp->elts[pc + 2].symbol);
10310
10311 if (noside == EVAL_SKIP)
10312 return eval_skip_value (exp);
10313
10314 val = ada_value_cast (to_type, val);
10315
10316 /* Follow the Ada language semantics that do not allow taking
10317 an address of the result of a cast (view conversion in Ada). */
10318 if (VALUE_LVAL (val) == lval_memory)
10319 {
10320 if (value_lazy (val))
10321 value_fetch_lazy (val);
10322 VALUE_LVAL (val) = not_lval;
10323 }
10324 return val;
10325 }
10326
10327 value *val = evaluate_subexp (to_type, exp, pos, noside);
10328 if (noside == EVAL_SKIP)
10329 return eval_skip_value (exp);
10330 return ada_value_cast (to_type, val);
10331 }
10332
10333 /* Implement the evaluate_exp routine in the exp_descriptor structure
10334 for the Ada language. */
10335
10336 static struct value *
10337 ada_evaluate_subexp (struct type *expect_type, struct expression *exp,
10338 int *pos, enum noside noside)
10339 {
10340 enum exp_opcode op;
10341 int tem;
10342 int pc;
10343 int preeval_pos;
10344 struct value *arg1 = NULL, *arg2 = NULL, *arg3;
10345 struct type *type;
10346 int nargs, oplen;
10347 struct value **argvec;
10348
10349 pc = *pos;
10350 *pos += 1;
10351 op = exp->elts[pc].opcode;
10352
10353 switch (op)
10354 {
10355 default:
10356 *pos -= 1;
10357 arg1 = evaluate_subexp_standard (expect_type, exp, pos, noside);
10358
10359 if (noside == EVAL_NORMAL)
10360 arg1 = unwrap_value (arg1);
10361
10362 /* If evaluating an OP_FLOAT and an EXPECT_TYPE was provided,
10363 then we need to perform the conversion manually, because
10364 evaluate_subexp_standard doesn't do it. This conversion is
10365 necessary in Ada because the different kinds of float/fixed
10366 types in Ada have different representations.
10367
10368 Similarly, we need to perform the conversion from OP_LONG
10369 ourselves. */
10370 if ((op == OP_FLOAT || op == OP_LONG) && expect_type != NULL)
10371 arg1 = ada_value_cast (expect_type, arg1);
10372
10373 return arg1;
10374
10375 case OP_STRING:
10376 {
10377 struct value *result;
10378
10379 *pos -= 1;
10380 result = evaluate_subexp_standard (expect_type, exp, pos, noside);
10381 /* The result type will have code OP_STRING, bashed there from
10382 OP_ARRAY. Bash it back. */
10383 if (value_type (result)->code () == TYPE_CODE_STRING)
10384 value_type (result)->set_code (TYPE_CODE_ARRAY);
10385 return result;
10386 }
10387
10388 case UNOP_CAST:
10389 (*pos) += 2;
10390 type = exp->elts[pc + 1].type;
10391 return ada_evaluate_subexp_for_cast (exp, pos, noside, type);
10392
10393 case UNOP_QUAL:
10394 (*pos) += 2;
10395 type = exp->elts[pc + 1].type;
10396 return ada_evaluate_subexp (type, exp, pos, noside);
10397
10398 case BINOP_ASSIGN:
10399 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10400 if (exp->elts[*pos].opcode == OP_AGGREGATE)
10401 {
10402 arg1 = assign_aggregate (arg1, arg1, exp, pos, noside);
10403 if (noside == EVAL_SKIP || noside == EVAL_AVOID_SIDE_EFFECTS)
10404 return arg1;
10405 return ada_value_assign (arg1, arg1);
10406 }
10407 /* Force the evaluation of the rhs ARG2 to the type of the lhs ARG1,
10408 except if the lhs of our assignment is a convenience variable.
10409 In the case of assigning to a convenience variable, the lhs
10410 should be exactly the result of the evaluation of the rhs. */
10411 type = value_type (arg1);
10412 if (VALUE_LVAL (arg1) == lval_internalvar)
10413 type = NULL;
10414 arg2 = evaluate_subexp (type, exp, pos, noside);
10415 if (noside == EVAL_SKIP || noside == EVAL_AVOID_SIDE_EFFECTS)
10416 return arg1;
10417 if (VALUE_LVAL (arg1) == lval_internalvar)
10418 {
10419 /* Nothing. */
10420 }
10421 else if (ada_is_gnat_encoded_fixed_point_type (value_type (arg1)))
10422 arg2 = cast_to_fixed (value_type (arg1), arg2);
10423 else if (ada_is_gnat_encoded_fixed_point_type (value_type (arg2)))
10424 error
10425 (_("Fixed-point values must be assigned to fixed-point variables"));
10426 else
10427 arg2 = coerce_for_assign (value_type (arg1), arg2);
10428 return ada_value_assign (arg1, arg2);
10429
10430 case BINOP_ADD:
10431 arg1 = evaluate_subexp_with_coercion (exp, pos, noside);
10432 arg2 = evaluate_subexp_with_coercion (exp, pos, noside);
10433 if (noside == EVAL_SKIP)
10434 goto nosideret;
10435 if (value_type (arg1)->code () == TYPE_CODE_PTR)
10436 return (value_from_longest
10437 (value_type (arg1),
10438 value_as_long (arg1) + value_as_long (arg2)));
10439 if (value_type (arg2)->code () == TYPE_CODE_PTR)
10440 return (value_from_longest
10441 (value_type (arg2),
10442 value_as_long (arg1) + value_as_long (arg2)));
10443 if ((ada_is_gnat_encoded_fixed_point_type (value_type (arg1))
10444 || ada_is_gnat_encoded_fixed_point_type (value_type (arg2)))
10445 && value_type (arg1) != value_type (arg2))
10446 error (_("Operands of fixed-point addition must have the same type"));
10447 /* Do the addition, and cast the result to the type of the first
10448 argument. We cannot cast the result to a reference type, so if
10449 ARG1 is a reference type, find its underlying type. */
10450 type = value_type (arg1);
10451 while (type->code () == TYPE_CODE_REF)
10452 type = TYPE_TARGET_TYPE (type);
10453 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10454 return value_cast (type, value_binop (arg1, arg2, BINOP_ADD));
10455
10456 case BINOP_SUB:
10457 arg1 = evaluate_subexp_with_coercion (exp, pos, noside);
10458 arg2 = evaluate_subexp_with_coercion (exp, pos, noside);
10459 if (noside == EVAL_SKIP)
10460 goto nosideret;
10461 if (value_type (arg1)->code () == TYPE_CODE_PTR)
10462 return (value_from_longest
10463 (value_type (arg1),
10464 value_as_long (arg1) - value_as_long (arg2)));
10465 if (value_type (arg2)->code () == TYPE_CODE_PTR)
10466 return (value_from_longest
10467 (value_type (arg2),
10468 value_as_long (arg1) - value_as_long (arg2)));
10469 if ((ada_is_gnat_encoded_fixed_point_type (value_type (arg1))
10470 || ada_is_gnat_encoded_fixed_point_type (value_type (arg2)))
10471 && value_type (arg1) != value_type (arg2))
10472 error (_("Operands of fixed-point subtraction "
10473 "must have the same type"));
10474 /* Do the substraction, and cast the result to the type of the first
10475 argument. We cannot cast the result to a reference type, so if
10476 ARG1 is a reference type, find its underlying type. */
10477 type = value_type (arg1);
10478 while (type->code () == TYPE_CODE_REF)
10479 type = TYPE_TARGET_TYPE (type);
10480 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10481 return value_cast (type, value_binop (arg1, arg2, BINOP_SUB));
10482
10483 case BINOP_MUL:
10484 case BINOP_DIV:
10485 case BINOP_REM:
10486 case BINOP_MOD:
10487 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10488 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10489 if (noside == EVAL_SKIP)
10490 goto nosideret;
10491 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
10492 {
10493 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10494 return value_zero (value_type (arg1), not_lval);
10495 }
10496 else
10497 {
10498 type = builtin_type (exp->gdbarch)->builtin_double;
10499 if (ada_is_gnat_encoded_fixed_point_type (value_type (arg1)))
10500 arg1 = cast_from_fixed (type, arg1);
10501 if (ada_is_gnat_encoded_fixed_point_type (value_type (arg2)))
10502 arg2 = cast_from_fixed (type, arg2);
10503 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10504 return ada_value_binop (arg1, arg2, op);
10505 }
10506
10507 case BINOP_EQUAL:
10508 case BINOP_NOTEQUAL:
10509 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10510 arg2 = evaluate_subexp (value_type (arg1), exp, pos, noside);
10511 if (noside == EVAL_SKIP)
10512 goto nosideret;
10513 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10514 tem = 0;
10515 else
10516 {
10517 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10518 tem = ada_value_equal (arg1, arg2);
10519 }
10520 if (op == BINOP_NOTEQUAL)
10521 tem = !tem;
10522 type = language_bool_type (exp->language_defn, exp->gdbarch);
10523 return value_from_longest (type, (LONGEST) tem);
10524
10525 case UNOP_NEG:
10526 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10527 if (noside == EVAL_SKIP)
10528 goto nosideret;
10529 else if (ada_is_gnat_encoded_fixed_point_type (value_type (arg1)))
10530 return value_cast (value_type (arg1), value_neg (arg1));
10531 else
10532 {
10533 unop_promote (exp->language_defn, exp->gdbarch, &arg1);
10534 return value_neg (arg1);
10535 }
10536
10537 case BINOP_LOGICAL_AND:
10538 case BINOP_LOGICAL_OR:
10539 case UNOP_LOGICAL_NOT:
10540 {
10541 struct value *val;
10542
10543 *pos -= 1;
10544 val = evaluate_subexp_standard (expect_type, exp, pos, noside);
10545 type = language_bool_type (exp->language_defn, exp->gdbarch);
10546 return value_cast (type, val);
10547 }
10548
10549 case BINOP_BITWISE_AND:
10550 case BINOP_BITWISE_IOR:
10551 case BINOP_BITWISE_XOR:
10552 {
10553 struct value *val;
10554
10555 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, EVAL_AVOID_SIDE_EFFECTS);
10556 *pos = pc;
10557 val = evaluate_subexp_standard (expect_type, exp, pos, noside);
10558
10559 return value_cast (value_type (arg1), val);
10560 }
10561
10562 case OP_VAR_VALUE:
10563 *pos -= 1;
10564
10565 if (noside == EVAL_SKIP)
10566 {
10567 *pos += 4;
10568 goto nosideret;
10569 }
10570
10571 if (SYMBOL_DOMAIN (exp->elts[pc + 2].symbol) == UNDEF_DOMAIN)
10572 /* Only encountered when an unresolved symbol occurs in a
10573 context other than a function call, in which case, it is
10574 invalid. */
10575 error (_("Unexpected unresolved symbol, %s, during evaluation"),
10576 exp->elts[pc + 2].symbol->print_name ());
10577
10578 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10579 {
10580 type = static_unwrap_type (SYMBOL_TYPE (exp->elts[pc + 2].symbol));
10581 /* Check to see if this is a tagged type. We also need to handle
10582 the case where the type is a reference to a tagged type, but
10583 we have to be careful to exclude pointers to tagged types.
10584 The latter should be shown as usual (as a pointer), whereas
10585 a reference should mostly be transparent to the user. */
10586 if (ada_is_tagged_type (type, 0)
10587 || (type->code () == TYPE_CODE_REF
10588 && ada_is_tagged_type (TYPE_TARGET_TYPE (type), 0)))
10589 {
10590 /* Tagged types are a little special in the fact that the real
10591 type is dynamic and can only be determined by inspecting the
10592 object's tag. This means that we need to get the object's
10593 value first (EVAL_NORMAL) and then extract the actual object
10594 type from its tag.
10595
10596 Note that we cannot skip the final step where we extract
10597 the object type from its tag, because the EVAL_NORMAL phase
10598 results in dynamic components being resolved into fixed ones.
10599 This can cause problems when trying to print the type
10600 description of tagged types whose parent has a dynamic size:
10601 We use the type name of the "_parent" component in order
10602 to print the name of the ancestor type in the type description.
10603 If that component had a dynamic size, the resolution into
10604 a fixed type would result in the loss of that type name,
10605 thus preventing us from printing the name of the ancestor
10606 type in the type description. */
10607 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, EVAL_NORMAL);
10608
10609 if (type->code () != TYPE_CODE_REF)
10610 {
10611 struct type *actual_type;
10612
10613 actual_type = type_from_tag (ada_value_tag (arg1));
10614 if (actual_type == NULL)
10615 /* If, for some reason, we were unable to determine
10616 the actual type from the tag, then use the static
10617 approximation that we just computed as a fallback.
10618 This can happen if the debugging information is
10619 incomplete, for instance. */
10620 actual_type = type;
10621 return value_zero (actual_type, not_lval);
10622 }
10623 else
10624 {
10625 /* In the case of a ref, ada_coerce_ref takes care
10626 of determining the actual type. But the evaluation
10627 should return a ref as it should be valid to ask
10628 for its address; so rebuild a ref after coerce. */
10629 arg1 = ada_coerce_ref (arg1);
10630 return value_ref (arg1, TYPE_CODE_REF);
10631 }
10632 }
10633
10634 /* Records and unions for which GNAT encodings have been
10635 generated need to be statically fixed as well.
10636 Otherwise, non-static fixing produces a type where
10637 all dynamic properties are removed, which prevents "ptype"
10638 from being able to completely describe the type.
10639 For instance, a case statement in a variant record would be
10640 replaced by the relevant components based on the actual
10641 value of the discriminants. */
10642 if ((type->code () == TYPE_CODE_STRUCT
10643 && dynamic_template_type (type) != NULL)
10644 || (type->code () == TYPE_CODE_UNION
10645 && ada_find_parallel_type (type, "___XVU") != NULL))
10646 {
10647 *pos += 4;
10648 return value_zero (to_static_fixed_type (type), not_lval);
10649 }
10650 }
10651
10652 arg1 = evaluate_subexp_standard (expect_type, exp, pos, noside);
10653 return ada_to_fixed_value (arg1);
10654
10655 case OP_FUNCALL:
10656 (*pos) += 2;
10657
10658 /* Allocate arg vector, including space for the function to be
10659 called in argvec[0] and a terminating NULL. */
10660 nargs = longest_to_int (exp->elts[pc + 1].longconst);
10661 argvec = XALLOCAVEC (struct value *, nargs + 2);
10662
10663 if (exp->elts[*pos].opcode == OP_VAR_VALUE
10664 && SYMBOL_DOMAIN (exp->elts[pc + 5].symbol) == UNDEF_DOMAIN)
10665 error (_("Unexpected unresolved symbol, %s, during evaluation"),
10666 exp->elts[pc + 5].symbol->print_name ());
10667 else
10668 {
10669 for (tem = 0; tem <= nargs; tem += 1)
10670 argvec[tem] = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10671 argvec[tem] = 0;
10672
10673 if (noside == EVAL_SKIP)
10674 goto nosideret;
10675 }
10676
10677 if (ada_is_constrained_packed_array_type
10678 (desc_base_type (value_type (argvec[0]))))
10679 argvec[0] = ada_coerce_to_simple_array (argvec[0]);
10680 else if (value_type (argvec[0])->code () == TYPE_CODE_ARRAY
10681 && TYPE_FIELD_BITSIZE (value_type (argvec[0]), 0) != 0)
10682 /* This is a packed array that has already been fixed, and
10683 therefore already coerced to a simple array. Nothing further
10684 to do. */
10685 ;
10686 else if (value_type (argvec[0])->code () == TYPE_CODE_REF)
10687 {
10688 /* Make sure we dereference references so that all the code below
10689 feels like it's really handling the referenced value. Wrapping
10690 types (for alignment) may be there, so make sure we strip them as
10691 well. */
10692 argvec[0] = ada_to_fixed_value (coerce_ref (argvec[0]));
10693 }
10694 else if (value_type (argvec[0])->code () == TYPE_CODE_ARRAY
10695 && VALUE_LVAL (argvec[0]) == lval_memory)
10696 argvec[0] = value_addr (argvec[0]);
10697
10698 type = ada_check_typedef (value_type (argvec[0]));
10699
10700 /* Ada allows us to implicitly dereference arrays when subscripting
10701 them. So, if this is an array typedef (encoding use for array
10702 access types encoded as fat pointers), strip it now. */
10703 if (type->code () == TYPE_CODE_TYPEDEF)
10704 type = ada_typedef_target_type (type);
10705
10706 if (type->code () == TYPE_CODE_PTR)
10707 {
10708 switch (ada_check_typedef (TYPE_TARGET_TYPE (type))->code ())
10709 {
10710 case TYPE_CODE_FUNC:
10711 type = ada_check_typedef (TYPE_TARGET_TYPE (type));
10712 break;
10713 case TYPE_CODE_ARRAY:
10714 break;
10715 case TYPE_CODE_STRUCT:
10716 if (noside != EVAL_AVOID_SIDE_EFFECTS)
10717 argvec[0] = ada_value_ind (argvec[0]);
10718 type = ada_check_typedef (TYPE_TARGET_TYPE (type));
10719 break;
10720 default:
10721 error (_("cannot subscript or call something of type `%s'"),
10722 ada_type_name (value_type (argvec[0])));
10723 break;
10724 }
10725 }
10726
10727 switch (type->code ())
10728 {
10729 case TYPE_CODE_FUNC:
10730 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10731 {
10732 if (TYPE_TARGET_TYPE (type) == NULL)
10733 error_call_unknown_return_type (NULL);
10734 return allocate_value (TYPE_TARGET_TYPE (type));
10735 }
10736 return call_function_by_hand (argvec[0], NULL,
10737 gdb::make_array_view (argvec + 1,
10738 nargs));
10739 case TYPE_CODE_INTERNAL_FUNCTION:
10740 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10741 /* We don't know anything about what the internal
10742 function might return, but we have to return
10743 something. */
10744 return value_zero (builtin_type (exp->gdbarch)->builtin_int,
10745 not_lval);
10746 else
10747 return call_internal_function (exp->gdbarch, exp->language_defn,
10748 argvec[0], nargs, argvec + 1);
10749
10750 case TYPE_CODE_STRUCT:
10751 {
10752 int arity;
10753
10754 arity = ada_array_arity (type);
10755 type = ada_array_element_type (type, nargs);
10756 if (type == NULL)
10757 error (_("cannot subscript or call a record"));
10758 if (arity != nargs)
10759 error (_("wrong number of subscripts; expecting %d"), arity);
10760 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10761 return value_zero (ada_aligned_type (type), lval_memory);
10762 return
10763 unwrap_value (ada_value_subscript
10764 (argvec[0], nargs, argvec + 1));
10765 }
10766 case TYPE_CODE_ARRAY:
10767 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10768 {
10769 type = ada_array_element_type (type, nargs);
10770 if (type == NULL)
10771 error (_("element type of array unknown"));
10772 else
10773 return value_zero (ada_aligned_type (type), lval_memory);
10774 }
10775 return
10776 unwrap_value (ada_value_subscript
10777 (ada_coerce_to_simple_array (argvec[0]),
10778 nargs, argvec + 1));
10779 case TYPE_CODE_PTR: /* Pointer to array */
10780 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10781 {
10782 type = to_fixed_array_type (TYPE_TARGET_TYPE (type), NULL, 1);
10783 type = ada_array_element_type (type, nargs);
10784 if (type == NULL)
10785 error (_("element type of array unknown"));
10786 else
10787 return value_zero (ada_aligned_type (type), lval_memory);
10788 }
10789 return
10790 unwrap_value (ada_value_ptr_subscript (argvec[0],
10791 nargs, argvec + 1));
10792
10793 default:
10794 error (_("Attempt to index or call something other than an "
10795 "array or function"));
10796 }
10797
10798 case TERNOP_SLICE:
10799 {
10800 struct value *array = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10801 struct value *low_bound_val =
10802 evaluate_subexp (NULL_TYPE, exp, pos, noside);
10803 struct value *high_bound_val =
10804 evaluate_subexp (NULL_TYPE, exp, pos, noside);
10805 LONGEST low_bound;
10806 LONGEST high_bound;
10807
10808 low_bound_val = coerce_ref (low_bound_val);
10809 high_bound_val = coerce_ref (high_bound_val);
10810 low_bound = value_as_long (low_bound_val);
10811 high_bound = value_as_long (high_bound_val);
10812
10813 if (noside == EVAL_SKIP)
10814 goto nosideret;
10815
10816 /* If this is a reference to an aligner type, then remove all
10817 the aligners. */
10818 if (value_type (array)->code () == TYPE_CODE_REF
10819 && ada_is_aligner_type (TYPE_TARGET_TYPE (value_type (array))))
10820 TYPE_TARGET_TYPE (value_type (array)) =
10821 ada_aligned_type (TYPE_TARGET_TYPE (value_type (array)));
10822
10823 if (ada_is_constrained_packed_array_type (value_type (array)))
10824 error (_("cannot slice a packed array"));
10825
10826 /* If this is a reference to an array or an array lvalue,
10827 convert to a pointer. */
10828 if (value_type (array)->code () == TYPE_CODE_REF
10829 || (value_type (array)->code () == TYPE_CODE_ARRAY
10830 && VALUE_LVAL (array) == lval_memory))
10831 array = value_addr (array);
10832
10833 if (noside == EVAL_AVOID_SIDE_EFFECTS
10834 && ada_is_array_descriptor_type (ada_check_typedef
10835 (value_type (array))))
10836 return empty_array (ada_type_of_array (array, 0), low_bound,
10837 high_bound);
10838
10839 array = ada_coerce_to_simple_array_ptr (array);
10840
10841 /* If we have more than one level of pointer indirection,
10842 dereference the value until we get only one level. */
10843 while (value_type (array)->code () == TYPE_CODE_PTR
10844 && (TYPE_TARGET_TYPE (value_type (array))->code ()
10845 == TYPE_CODE_PTR))
10846 array = value_ind (array);
10847
10848 /* Make sure we really do have an array type before going further,
10849 to avoid a SEGV when trying to get the index type or the target
10850 type later down the road if the debug info generated by
10851 the compiler is incorrect or incomplete. */
10852 if (!ada_is_simple_array_type (value_type (array)))
10853 error (_("cannot take slice of non-array"));
10854
10855 if (ada_check_typedef (value_type (array))->code ()
10856 == TYPE_CODE_PTR)
10857 {
10858 struct type *type0 = ada_check_typedef (value_type (array));
10859
10860 if (high_bound < low_bound || noside == EVAL_AVOID_SIDE_EFFECTS)
10861 return empty_array (TYPE_TARGET_TYPE (type0), low_bound, high_bound);
10862 else
10863 {
10864 struct type *arr_type0 =
10865 to_fixed_array_type (TYPE_TARGET_TYPE (type0), NULL, 1);
10866
10867 return ada_value_slice_from_ptr (array, arr_type0,
10868 longest_to_int (low_bound),
10869 longest_to_int (high_bound));
10870 }
10871 }
10872 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
10873 return array;
10874 else if (high_bound < low_bound)
10875 return empty_array (value_type (array), low_bound, high_bound);
10876 else
10877 return ada_value_slice (array, longest_to_int (low_bound),
10878 longest_to_int (high_bound));
10879 }
10880
10881 case UNOP_IN_RANGE:
10882 (*pos) += 2;
10883 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10884 type = check_typedef (exp->elts[pc + 1].type);
10885
10886 if (noside == EVAL_SKIP)
10887 goto nosideret;
10888
10889 switch (type->code ())
10890 {
10891 default:
10892 lim_warning (_("Membership test incompletely implemented; "
10893 "always returns true"));
10894 type = language_bool_type (exp->language_defn, exp->gdbarch);
10895 return value_from_longest (type, (LONGEST) 1);
10896
10897 case TYPE_CODE_RANGE:
10898 arg2 = value_from_longest (type, TYPE_LOW_BOUND (type));
10899 arg3 = value_from_longest (type, TYPE_HIGH_BOUND (type));
10900 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10901 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg3);
10902 type = language_bool_type (exp->language_defn, exp->gdbarch);
10903 return
10904 value_from_longest (type,
10905 (value_less (arg1, arg3)
10906 || value_equal (arg1, arg3))
10907 && (value_less (arg2, arg1)
10908 || value_equal (arg2, arg1)));
10909 }
10910
10911 case BINOP_IN_BOUNDS:
10912 (*pos) += 2;
10913 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10914 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10915
10916 if (noside == EVAL_SKIP)
10917 goto nosideret;
10918
10919 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10920 {
10921 type = language_bool_type (exp->language_defn, exp->gdbarch);
10922 return value_zero (type, not_lval);
10923 }
10924
10925 tem = longest_to_int (exp->elts[pc + 1].longconst);
10926
10927 type = ada_index_type (value_type (arg2), tem, "range");
10928 if (!type)
10929 type = value_type (arg1);
10930
10931 arg3 = value_from_longest (type, ada_array_bound (arg2, tem, 1));
10932 arg2 = value_from_longest (type, ada_array_bound (arg2, tem, 0));
10933
10934 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10935 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg3);
10936 type = language_bool_type (exp->language_defn, exp->gdbarch);
10937 return
10938 value_from_longest (type,
10939 (value_less (arg1, arg3)
10940 || value_equal (arg1, arg3))
10941 && (value_less (arg2, arg1)
10942 || value_equal (arg2, arg1)));
10943
10944 case TERNOP_IN_RANGE:
10945 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10946 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10947 arg3 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10948
10949 if (noside == EVAL_SKIP)
10950 goto nosideret;
10951
10952 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10953 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg3);
10954 type = language_bool_type (exp->language_defn, exp->gdbarch);
10955 return
10956 value_from_longest (type,
10957 (value_less (arg1, arg3)
10958 || value_equal (arg1, arg3))
10959 && (value_less (arg2, arg1)
10960 || value_equal (arg2, arg1)));
10961
10962 case OP_ATR_FIRST:
10963 case OP_ATR_LAST:
10964 case OP_ATR_LENGTH:
10965 {
10966 struct type *type_arg;
10967
10968 if (exp->elts[*pos].opcode == OP_TYPE)
10969 {
10970 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
10971 arg1 = NULL;
10972 type_arg = check_typedef (exp->elts[pc + 2].type);
10973 }
10974 else
10975 {
10976 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10977 type_arg = NULL;
10978 }
10979
10980 if (exp->elts[*pos].opcode != OP_LONG)
10981 error (_("Invalid operand to '%s"), ada_attribute_name (op));
10982 tem = longest_to_int (exp->elts[*pos + 2].longconst);
10983 *pos += 4;
10984
10985 if (noside == EVAL_SKIP)
10986 goto nosideret;
10987 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
10988 {
10989 if (type_arg == NULL)
10990 type_arg = value_type (arg1);
10991
10992 if (ada_is_constrained_packed_array_type (type_arg))
10993 type_arg = decode_constrained_packed_array_type (type_arg);
10994
10995 if (!discrete_type_p (type_arg))
10996 {
10997 switch (op)
10998 {
10999 default: /* Should never happen. */
11000 error (_("unexpected attribute encountered"));
11001 case OP_ATR_FIRST:
11002 case OP_ATR_LAST:
11003 type_arg = ada_index_type (type_arg, tem,
11004 ada_attribute_name (op));
11005 break;
11006 case OP_ATR_LENGTH:
11007 type_arg = builtin_type (exp->gdbarch)->builtin_int;
11008 break;
11009 }
11010 }
11011
11012 return value_zero (type_arg, not_lval);
11013 }
11014 else if (type_arg == NULL)
11015 {
11016 arg1 = ada_coerce_ref (arg1);
11017
11018 if (ada_is_constrained_packed_array_type (value_type (arg1)))
11019 arg1 = ada_coerce_to_simple_array (arg1);
11020
11021 if (op == OP_ATR_LENGTH)
11022 type = builtin_type (exp->gdbarch)->builtin_int;
11023 else
11024 {
11025 type = ada_index_type (value_type (arg1), tem,
11026 ada_attribute_name (op));
11027 if (type == NULL)
11028 type = builtin_type (exp->gdbarch)->builtin_int;
11029 }
11030
11031 switch (op)
11032 {
11033 default: /* Should never happen. */
11034 error (_("unexpected attribute encountered"));
11035 case OP_ATR_FIRST:
11036 return value_from_longest
11037 (type, ada_array_bound (arg1, tem, 0));
11038 case OP_ATR_LAST:
11039 return value_from_longest
11040 (type, ada_array_bound (arg1, tem, 1));
11041 case OP_ATR_LENGTH:
11042 return value_from_longest
11043 (type, ada_array_length (arg1, tem));
11044 }
11045 }
11046 else if (discrete_type_p (type_arg))
11047 {
11048 struct type *range_type;
11049 const char *name = ada_type_name (type_arg);
11050
11051 range_type = NULL;
11052 if (name != NULL && type_arg->code () != TYPE_CODE_ENUM)
11053 range_type = to_fixed_range_type (type_arg, NULL);
11054 if (range_type == NULL)
11055 range_type = type_arg;
11056 switch (op)
11057 {
11058 default:
11059 error (_("unexpected attribute encountered"));
11060 case OP_ATR_FIRST:
11061 return value_from_longest
11062 (range_type, ada_discrete_type_low_bound (range_type));
11063 case OP_ATR_LAST:
11064 return value_from_longest
11065 (range_type, ada_discrete_type_high_bound (range_type));
11066 case OP_ATR_LENGTH:
11067 error (_("the 'length attribute applies only to array types"));
11068 }
11069 }
11070 else if (type_arg->code () == TYPE_CODE_FLT)
11071 error (_("unimplemented type attribute"));
11072 else
11073 {
11074 LONGEST low, high;
11075
11076 if (ada_is_constrained_packed_array_type (type_arg))
11077 type_arg = decode_constrained_packed_array_type (type_arg);
11078
11079 if (op == OP_ATR_LENGTH)
11080 type = builtin_type (exp->gdbarch)->builtin_int;
11081 else
11082 {
11083 type = ada_index_type (type_arg, tem, ada_attribute_name (op));
11084 if (type == NULL)
11085 type = builtin_type (exp->gdbarch)->builtin_int;
11086 }
11087
11088 switch (op)
11089 {
11090 default:
11091 error (_("unexpected attribute encountered"));
11092 case OP_ATR_FIRST:
11093 low = ada_array_bound_from_type (type_arg, tem, 0);
11094 return value_from_longest (type, low);
11095 case OP_ATR_LAST:
11096 high = ada_array_bound_from_type (type_arg, tem, 1);
11097 return value_from_longest (type, high);
11098 case OP_ATR_LENGTH:
11099 low = ada_array_bound_from_type (type_arg, tem, 0);
11100 high = ada_array_bound_from_type (type_arg, tem, 1);
11101 return value_from_longest (type, high - low + 1);
11102 }
11103 }
11104 }
11105
11106 case OP_ATR_TAG:
11107 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11108 if (noside == EVAL_SKIP)
11109 goto nosideret;
11110
11111 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11112 return value_zero (ada_tag_type (arg1), not_lval);
11113
11114 return ada_value_tag (arg1);
11115
11116 case OP_ATR_MIN:
11117 case OP_ATR_MAX:
11118 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11119 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11120 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11121 if (noside == EVAL_SKIP)
11122 goto nosideret;
11123 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11124 return value_zero (value_type (arg1), not_lval);
11125 else
11126 {
11127 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
11128 return value_binop (arg1, arg2,
11129 op == OP_ATR_MIN ? BINOP_MIN : BINOP_MAX);
11130 }
11131
11132 case OP_ATR_MODULUS:
11133 {
11134 struct type *type_arg = check_typedef (exp->elts[pc + 2].type);
11135
11136 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11137 if (noside == EVAL_SKIP)
11138 goto nosideret;
11139
11140 if (!ada_is_modular_type (type_arg))
11141 error (_("'modulus must be applied to modular type"));
11142
11143 return value_from_longest (TYPE_TARGET_TYPE (type_arg),
11144 ada_modulus (type_arg));
11145 }
11146
11147
11148 case OP_ATR_POS:
11149 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11150 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11151 if (noside == EVAL_SKIP)
11152 goto nosideret;
11153 type = builtin_type (exp->gdbarch)->builtin_int;
11154 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11155 return value_zero (type, not_lval);
11156 else
11157 return value_pos_atr (type, arg1);
11158
11159 case OP_ATR_SIZE:
11160 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11161 type = value_type (arg1);
11162
11163 /* If the argument is a reference, then dereference its type, since
11164 the user is really asking for the size of the actual object,
11165 not the size of the pointer. */
11166 if (type->code () == TYPE_CODE_REF)
11167 type = TYPE_TARGET_TYPE (type);
11168
11169 if (noside == EVAL_SKIP)
11170 goto nosideret;
11171 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11172 return value_zero (builtin_type (exp->gdbarch)->builtin_int, not_lval);
11173 else
11174 return value_from_longest (builtin_type (exp->gdbarch)->builtin_int,
11175 TARGET_CHAR_BIT * TYPE_LENGTH (type));
11176
11177 case OP_ATR_VAL:
11178 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11179 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11180 type = exp->elts[pc + 2].type;
11181 if (noside == EVAL_SKIP)
11182 goto nosideret;
11183 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11184 return value_zero (type, not_lval);
11185 else
11186 return value_val_atr (type, arg1);
11187
11188 case BINOP_EXP:
11189 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11190 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11191 if (noside == EVAL_SKIP)
11192 goto nosideret;
11193 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11194 return value_zero (value_type (arg1), not_lval);
11195 else
11196 {
11197 /* For integer exponentiation operations,
11198 only promote the first argument. */
11199 if (is_integral_type (value_type (arg2)))
11200 unop_promote (exp->language_defn, exp->gdbarch, &arg1);
11201 else
11202 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
11203
11204 return value_binop (arg1, arg2, op);
11205 }
11206
11207 case UNOP_PLUS:
11208 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11209 if (noside == EVAL_SKIP)
11210 goto nosideret;
11211 else
11212 return arg1;
11213
11214 case UNOP_ABS:
11215 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11216 if (noside == EVAL_SKIP)
11217 goto nosideret;
11218 unop_promote (exp->language_defn, exp->gdbarch, &arg1);
11219 if (value_less (arg1, value_zero (value_type (arg1), not_lval)))
11220 return value_neg (arg1);
11221 else
11222 return arg1;
11223
11224 case UNOP_IND:
11225 preeval_pos = *pos;
11226 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11227 if (noside == EVAL_SKIP)
11228 goto nosideret;
11229 type = ada_check_typedef (value_type (arg1));
11230 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11231 {
11232 if (ada_is_array_descriptor_type (type))
11233 /* GDB allows dereferencing GNAT array descriptors. */
11234 {
11235 struct type *arrType = ada_type_of_array (arg1, 0);
11236
11237 if (arrType == NULL)
11238 error (_("Attempt to dereference null array pointer."));
11239 return value_at_lazy (arrType, 0);
11240 }
11241 else if (type->code () == TYPE_CODE_PTR
11242 || type->code () == TYPE_CODE_REF
11243 /* In C you can dereference an array to get the 1st elt. */
11244 || type->code () == TYPE_CODE_ARRAY)
11245 {
11246 /* As mentioned in the OP_VAR_VALUE case, tagged types can
11247 only be determined by inspecting the object's tag.
11248 This means that we need to evaluate completely the
11249 expression in order to get its type. */
11250
11251 if ((type->code () == TYPE_CODE_REF
11252 || type->code () == TYPE_CODE_PTR)
11253 && ada_is_tagged_type (TYPE_TARGET_TYPE (type), 0))
11254 {
11255 arg1 = evaluate_subexp (NULL_TYPE, exp, &preeval_pos,
11256 EVAL_NORMAL);
11257 type = value_type (ada_value_ind (arg1));
11258 }
11259 else
11260 {
11261 type = to_static_fixed_type
11262 (ada_aligned_type
11263 (ada_check_typedef (TYPE_TARGET_TYPE (type))));
11264 }
11265 ada_ensure_varsize_limit (type);
11266 return value_zero (type, lval_memory);
11267 }
11268 else if (type->code () == TYPE_CODE_INT)
11269 {
11270 /* GDB allows dereferencing an int. */
11271 if (expect_type == NULL)
11272 return value_zero (builtin_type (exp->gdbarch)->builtin_int,
11273 lval_memory);
11274 else
11275 {
11276 expect_type =
11277 to_static_fixed_type (ada_aligned_type (expect_type));
11278 return value_zero (expect_type, lval_memory);
11279 }
11280 }
11281 else
11282 error (_("Attempt to take contents of a non-pointer value."));
11283 }
11284 arg1 = ada_coerce_ref (arg1); /* FIXME: What is this for?? */
11285 type = ada_check_typedef (value_type (arg1));
11286
11287 if (type->code () == TYPE_CODE_INT)
11288 /* GDB allows dereferencing an int. If we were given
11289 the expect_type, then use that as the target type.
11290 Otherwise, assume that the target type is an int. */
11291 {
11292 if (expect_type != NULL)
11293 return ada_value_ind (value_cast (lookup_pointer_type (expect_type),
11294 arg1));
11295 else
11296 return value_at_lazy (builtin_type (exp->gdbarch)->builtin_int,
11297 (CORE_ADDR) value_as_address (arg1));
11298 }
11299
11300 if (ada_is_array_descriptor_type (type))
11301 /* GDB allows dereferencing GNAT array descriptors. */
11302 return ada_coerce_to_simple_array (arg1);
11303 else
11304 return ada_value_ind (arg1);
11305
11306 case STRUCTOP_STRUCT:
11307 tem = longest_to_int (exp->elts[pc + 1].longconst);
11308 (*pos) += 3 + BYTES_TO_EXP_ELEM (tem + 1);
11309 preeval_pos = *pos;
11310 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11311 if (noside == EVAL_SKIP)
11312 goto nosideret;
11313 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11314 {
11315 struct type *type1 = value_type (arg1);
11316
11317 if (ada_is_tagged_type (type1, 1))
11318 {
11319 type = ada_lookup_struct_elt_type (type1,
11320 &exp->elts[pc + 2].string,
11321 1, 1);
11322
11323 /* If the field is not found, check if it exists in the
11324 extension of this object's type. This means that we
11325 need to evaluate completely the expression. */
11326
11327 if (type == NULL)
11328 {
11329 arg1 = evaluate_subexp (NULL_TYPE, exp, &preeval_pos,
11330 EVAL_NORMAL);
11331 arg1 = ada_value_struct_elt (arg1,
11332 &exp->elts[pc + 2].string,
11333 0);
11334 arg1 = unwrap_value (arg1);
11335 type = value_type (ada_to_fixed_value (arg1));
11336 }
11337 }
11338 else
11339 type =
11340 ada_lookup_struct_elt_type (type1, &exp->elts[pc + 2].string, 1,
11341 0);
11342
11343 return value_zero (ada_aligned_type (type), lval_memory);
11344 }
11345 else
11346 {
11347 arg1 = ada_value_struct_elt (arg1, &exp->elts[pc + 2].string, 0);
11348 arg1 = unwrap_value (arg1);
11349 return ada_to_fixed_value (arg1);
11350 }
11351
11352 case OP_TYPE:
11353 /* The value is not supposed to be used. This is here to make it
11354 easier to accommodate expressions that contain types. */
11355 (*pos) += 2;
11356 if (noside == EVAL_SKIP)
11357 goto nosideret;
11358 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11359 return allocate_value (exp->elts[pc + 1].type);
11360 else
11361 error (_("Attempt to use a type name as an expression"));
11362
11363 case OP_AGGREGATE:
11364 case OP_CHOICES:
11365 case OP_OTHERS:
11366 case OP_DISCRETE_RANGE:
11367 case OP_POSITIONAL:
11368 case OP_NAME:
11369 if (noside == EVAL_NORMAL)
11370 switch (op)
11371 {
11372 case OP_NAME:
11373 error (_("Undefined name, ambiguous name, or renaming used in "
11374 "component association: %s."), &exp->elts[pc+2].string);
11375 case OP_AGGREGATE:
11376 error (_("Aggregates only allowed on the right of an assignment"));
11377 default:
11378 internal_error (__FILE__, __LINE__,
11379 _("aggregate apparently mangled"));
11380 }
11381
11382 ada_forward_operator_length (exp, pc, &oplen, &nargs);
11383 *pos += oplen - 1;
11384 for (tem = 0; tem < nargs; tem += 1)
11385 ada_evaluate_subexp (NULL, exp, pos, noside);
11386 goto nosideret;
11387 }
11388
11389 nosideret:
11390 return eval_skip_value (exp);
11391 }
11392 \f
11393
11394 /* Fixed point */
11395
11396 /* If TYPE encodes an Ada fixed-point type, return the suffix of the
11397 type name that encodes the 'small and 'delta information.
11398 Otherwise, return NULL. */
11399
11400 static const char *
11401 gnat_encoded_fixed_type_info (struct type *type)
11402 {
11403 const char *name = ada_type_name (type);
11404 enum type_code code = (type == NULL) ? TYPE_CODE_UNDEF : type->code ();
11405
11406 if ((code == TYPE_CODE_INT || code == TYPE_CODE_RANGE) && name != NULL)
11407 {
11408 const char *tail = strstr (name, "___XF_");
11409
11410 if (tail == NULL)
11411 return NULL;
11412 else
11413 return tail + 5;
11414 }
11415 else if (code == TYPE_CODE_RANGE && TYPE_TARGET_TYPE (type) != type)
11416 return gnat_encoded_fixed_type_info (TYPE_TARGET_TYPE (type));
11417 else
11418 return NULL;
11419 }
11420
11421 /* Returns non-zero iff TYPE represents an Ada fixed-point type. */
11422
11423 int
11424 ada_is_gnat_encoded_fixed_point_type (struct type *type)
11425 {
11426 return gnat_encoded_fixed_type_info (type) != NULL;
11427 }
11428
11429 /* Return non-zero iff TYPE represents a System.Address type. */
11430
11431 int
11432 ada_is_system_address_type (struct type *type)
11433 {
11434 return (type->name () && strcmp (type->name (), "system__address") == 0);
11435 }
11436
11437 /* Assuming that TYPE is the representation of an Ada fixed-point
11438 type, return the target floating-point type to be used to represent
11439 of this type during internal computation. */
11440
11441 static struct type *
11442 ada_scaling_type (struct type *type)
11443 {
11444 return builtin_type (get_type_arch (type))->builtin_long_double;
11445 }
11446
11447 /* Assuming that TYPE is the representation of an Ada fixed-point
11448 type, return its delta, or NULL if the type is malformed and the
11449 delta cannot be determined. */
11450
11451 struct value *
11452 gnat_encoded_fixed_point_delta (struct type *type)
11453 {
11454 const char *encoding = gnat_encoded_fixed_type_info (type);
11455 struct type *scale_type = ada_scaling_type (type);
11456
11457 long long num, den;
11458
11459 if (sscanf (encoding, "_%lld_%lld", &num, &den) < 2)
11460 return nullptr;
11461 else
11462 return value_binop (value_from_longest (scale_type, num),
11463 value_from_longest (scale_type, den), BINOP_DIV);
11464 }
11465
11466 /* Assuming that ada_is_gnat_encoded_fixed_point_type (TYPE), return
11467 the scaling factor ('SMALL value) associated with the type. */
11468
11469 struct value *
11470 ada_scaling_factor (struct type *type)
11471 {
11472 const char *encoding = gnat_encoded_fixed_type_info (type);
11473 struct type *scale_type = ada_scaling_type (type);
11474
11475 long long num0, den0, num1, den1;
11476 int n;
11477
11478 n = sscanf (encoding, "_%lld_%lld_%lld_%lld",
11479 &num0, &den0, &num1, &den1);
11480
11481 if (n < 2)
11482 return value_from_longest (scale_type, 1);
11483 else if (n == 4)
11484 return value_binop (value_from_longest (scale_type, num1),
11485 value_from_longest (scale_type, den1), BINOP_DIV);
11486 else
11487 return value_binop (value_from_longest (scale_type, num0),
11488 value_from_longest (scale_type, den0), BINOP_DIV);
11489 }
11490
11491 \f
11492
11493 /* Range types */
11494
11495 /* Scan STR beginning at position K for a discriminant name, and
11496 return the value of that discriminant field of DVAL in *PX. If
11497 PNEW_K is not null, put the position of the character beyond the
11498 name scanned in *PNEW_K. Return 1 if successful; return 0 and do
11499 not alter *PX and *PNEW_K if unsuccessful. */
11500
11501 static int
11502 scan_discrim_bound (const char *str, int k, struct value *dval, LONGEST * px,
11503 int *pnew_k)
11504 {
11505 static char *bound_buffer = NULL;
11506 static size_t bound_buffer_len = 0;
11507 const char *pstart, *pend, *bound;
11508 struct value *bound_val;
11509
11510 if (dval == NULL || str == NULL || str[k] == '\0')
11511 return 0;
11512
11513 pstart = str + k;
11514 pend = strstr (pstart, "__");
11515 if (pend == NULL)
11516 {
11517 bound = pstart;
11518 k += strlen (bound);
11519 }
11520 else
11521 {
11522 int len = pend - pstart;
11523
11524 /* Strip __ and beyond. */
11525 GROW_VECT (bound_buffer, bound_buffer_len, len + 1);
11526 strncpy (bound_buffer, pstart, len);
11527 bound_buffer[len] = '\0';
11528
11529 bound = bound_buffer;
11530 k = pend - str;
11531 }
11532
11533 bound_val = ada_search_struct_field (bound, dval, 0, value_type (dval));
11534 if (bound_val == NULL)
11535 return 0;
11536
11537 *px = value_as_long (bound_val);
11538 if (pnew_k != NULL)
11539 *pnew_k = k;
11540 return 1;
11541 }
11542
11543 /* Value of variable named NAME in the current environment. If
11544 no such variable found, then if ERR_MSG is null, returns 0, and
11545 otherwise causes an error with message ERR_MSG. */
11546
11547 static struct value *
11548 get_var_value (const char *name, const char *err_msg)
11549 {
11550 lookup_name_info lookup_name (name, symbol_name_match_type::FULL);
11551
11552 std::vector<struct block_symbol> syms;
11553 int nsyms = ada_lookup_symbol_list_worker (lookup_name,
11554 get_selected_block (0),
11555 VAR_DOMAIN, &syms, 1);
11556
11557 if (nsyms != 1)
11558 {
11559 if (err_msg == NULL)
11560 return 0;
11561 else
11562 error (("%s"), err_msg);
11563 }
11564
11565 return value_of_variable (syms[0].symbol, syms[0].block);
11566 }
11567
11568 /* Value of integer variable named NAME in the current environment.
11569 If no such variable is found, returns false. Otherwise, sets VALUE
11570 to the variable's value and returns true. */
11571
11572 bool
11573 get_int_var_value (const char *name, LONGEST &value)
11574 {
11575 struct value *var_val = get_var_value (name, 0);
11576
11577 if (var_val == 0)
11578 return false;
11579
11580 value = value_as_long (var_val);
11581 return true;
11582 }
11583
11584
11585 /* Return a range type whose base type is that of the range type named
11586 NAME in the current environment, and whose bounds are calculated
11587 from NAME according to the GNAT range encoding conventions.
11588 Extract discriminant values, if needed, from DVAL. ORIG_TYPE is the
11589 corresponding range type from debug information; fall back to using it
11590 if symbol lookup fails. If a new type must be created, allocate it
11591 like ORIG_TYPE was. The bounds information, in general, is encoded
11592 in NAME, the base type given in the named range type. */
11593
11594 static struct type *
11595 to_fixed_range_type (struct type *raw_type, struct value *dval)
11596 {
11597 const char *name;
11598 struct type *base_type;
11599 const char *subtype_info;
11600
11601 gdb_assert (raw_type != NULL);
11602 gdb_assert (raw_type->name () != NULL);
11603
11604 if (raw_type->code () == TYPE_CODE_RANGE)
11605 base_type = TYPE_TARGET_TYPE (raw_type);
11606 else
11607 base_type = raw_type;
11608
11609 name = raw_type->name ();
11610 subtype_info = strstr (name, "___XD");
11611 if (subtype_info == NULL)
11612 {
11613 LONGEST L = ada_discrete_type_low_bound (raw_type);
11614 LONGEST U = ada_discrete_type_high_bound (raw_type);
11615
11616 if (L < INT_MIN || U > INT_MAX)
11617 return raw_type;
11618 else
11619 return create_static_range_type (alloc_type_copy (raw_type), raw_type,
11620 L, U);
11621 }
11622 else
11623 {
11624 static char *name_buf = NULL;
11625 static size_t name_len = 0;
11626 int prefix_len = subtype_info - name;
11627 LONGEST L, U;
11628 struct type *type;
11629 const char *bounds_str;
11630 int n;
11631
11632 GROW_VECT (name_buf, name_len, prefix_len + 5);
11633 strncpy (name_buf, name, prefix_len);
11634 name_buf[prefix_len] = '\0';
11635
11636 subtype_info += 5;
11637 bounds_str = strchr (subtype_info, '_');
11638 n = 1;
11639
11640 if (*subtype_info == 'L')
11641 {
11642 if (!ada_scan_number (bounds_str, n, &L, &n)
11643 && !scan_discrim_bound (bounds_str, n, dval, &L, &n))
11644 return raw_type;
11645 if (bounds_str[n] == '_')
11646 n += 2;
11647 else if (bounds_str[n] == '.') /* FIXME? SGI Workshop kludge. */
11648 n += 1;
11649 subtype_info += 1;
11650 }
11651 else
11652 {
11653 strcpy (name_buf + prefix_len, "___L");
11654 if (!get_int_var_value (name_buf, L))
11655 {
11656 lim_warning (_("Unknown lower bound, using 1."));
11657 L = 1;
11658 }
11659 }
11660
11661 if (*subtype_info == 'U')
11662 {
11663 if (!ada_scan_number (bounds_str, n, &U, &n)
11664 && !scan_discrim_bound (bounds_str, n, dval, &U, &n))
11665 return raw_type;
11666 }
11667 else
11668 {
11669 strcpy (name_buf + prefix_len, "___U");
11670 if (!get_int_var_value (name_buf, U))
11671 {
11672 lim_warning (_("Unknown upper bound, using %ld."), (long) L);
11673 U = L;
11674 }
11675 }
11676
11677 type = create_static_range_type (alloc_type_copy (raw_type),
11678 base_type, L, U);
11679 /* create_static_range_type alters the resulting type's length
11680 to match the size of the base_type, which is not what we want.
11681 Set it back to the original range type's length. */
11682 TYPE_LENGTH (type) = TYPE_LENGTH (raw_type);
11683 type->set_name (name);
11684 return type;
11685 }
11686 }
11687
11688 /* True iff NAME is the name of a range type. */
11689
11690 int
11691 ada_is_range_type_name (const char *name)
11692 {
11693 return (name != NULL && strstr (name, "___XD"));
11694 }
11695 \f
11696
11697 /* Modular types */
11698
11699 /* True iff TYPE is an Ada modular type. */
11700
11701 int
11702 ada_is_modular_type (struct type *type)
11703 {
11704 struct type *subranged_type = get_base_type (type);
11705
11706 return (subranged_type != NULL && type->code () == TYPE_CODE_RANGE
11707 && subranged_type->code () == TYPE_CODE_INT
11708 && TYPE_UNSIGNED (subranged_type));
11709 }
11710
11711 /* Assuming ada_is_modular_type (TYPE), the modulus of TYPE. */
11712
11713 ULONGEST
11714 ada_modulus (struct type *type)
11715 {
11716 return (ULONGEST) TYPE_HIGH_BOUND (type) + 1;
11717 }
11718 \f
11719
11720 /* Ada exception catchpoint support:
11721 ---------------------------------
11722
11723 We support 3 kinds of exception catchpoints:
11724 . catchpoints on Ada exceptions
11725 . catchpoints on unhandled Ada exceptions
11726 . catchpoints on failed assertions
11727
11728 Exceptions raised during failed assertions, or unhandled exceptions
11729 could perfectly be caught with the general catchpoint on Ada exceptions.
11730 However, we can easily differentiate these two special cases, and having
11731 the option to distinguish these two cases from the rest can be useful
11732 to zero-in on certain situations.
11733
11734 Exception catchpoints are a specialized form of breakpoint,
11735 since they rely on inserting breakpoints inside known routines
11736 of the GNAT runtime. The implementation therefore uses a standard
11737 breakpoint structure of the BP_BREAKPOINT type, but with its own set
11738 of breakpoint_ops.
11739
11740 Support in the runtime for exception catchpoints have been changed
11741 a few times already, and these changes affect the implementation
11742 of these catchpoints. In order to be able to support several
11743 variants of the runtime, we use a sniffer that will determine
11744 the runtime variant used by the program being debugged. */
11745
11746 /* Ada's standard exceptions.
11747
11748 The Ada 83 standard also defined Numeric_Error. But there so many
11749 situations where it was unclear from the Ada 83 Reference Manual
11750 (RM) whether Constraint_Error or Numeric_Error should be raised,
11751 that the ARG (Ada Rapporteur Group) eventually issued a Binding
11752 Interpretation saying that anytime the RM says that Numeric_Error
11753 should be raised, the implementation may raise Constraint_Error.
11754 Ada 95 went one step further and pretty much removed Numeric_Error
11755 from the list of standard exceptions (it made it a renaming of
11756 Constraint_Error, to help preserve compatibility when compiling
11757 an Ada83 compiler). As such, we do not include Numeric_Error from
11758 this list of standard exceptions. */
11759
11760 static const char *standard_exc[] = {
11761 "constraint_error",
11762 "program_error",
11763 "storage_error",
11764 "tasking_error"
11765 };
11766
11767 typedef CORE_ADDR (ada_unhandled_exception_name_addr_ftype) (void);
11768
11769 /* A structure that describes how to support exception catchpoints
11770 for a given executable. */
11771
11772 struct exception_support_info
11773 {
11774 /* The name of the symbol to break on in order to insert
11775 a catchpoint on exceptions. */
11776 const char *catch_exception_sym;
11777
11778 /* The name of the symbol to break on in order to insert
11779 a catchpoint on unhandled exceptions. */
11780 const char *catch_exception_unhandled_sym;
11781
11782 /* The name of the symbol to break on in order to insert
11783 a catchpoint on failed assertions. */
11784 const char *catch_assert_sym;
11785
11786 /* The name of the symbol to break on in order to insert
11787 a catchpoint on exception handling. */
11788 const char *catch_handlers_sym;
11789
11790 /* Assuming that the inferior just triggered an unhandled exception
11791 catchpoint, this function is responsible for returning the address
11792 in inferior memory where the name of that exception is stored.
11793 Return zero if the address could not be computed. */
11794 ada_unhandled_exception_name_addr_ftype *unhandled_exception_name_addr;
11795 };
11796
11797 static CORE_ADDR ada_unhandled_exception_name_addr (void);
11798 static CORE_ADDR ada_unhandled_exception_name_addr_from_raise (void);
11799
11800 /* The following exception support info structure describes how to
11801 implement exception catchpoints with the latest version of the
11802 Ada runtime (as of 2019-08-??). */
11803
11804 static const struct exception_support_info default_exception_support_info =
11805 {
11806 "__gnat_debug_raise_exception", /* catch_exception_sym */
11807 "__gnat_unhandled_exception", /* catch_exception_unhandled_sym */
11808 "__gnat_debug_raise_assert_failure", /* catch_assert_sym */
11809 "__gnat_begin_handler_v1", /* catch_handlers_sym */
11810 ada_unhandled_exception_name_addr
11811 };
11812
11813 /* The following exception support info structure describes how to
11814 implement exception catchpoints with an earlier version of the
11815 Ada runtime (as of 2007-03-06) using v0 of the EH ABI. */
11816
11817 static const struct exception_support_info exception_support_info_v0 =
11818 {
11819 "__gnat_debug_raise_exception", /* catch_exception_sym */
11820 "__gnat_unhandled_exception", /* catch_exception_unhandled_sym */
11821 "__gnat_debug_raise_assert_failure", /* catch_assert_sym */
11822 "__gnat_begin_handler", /* catch_handlers_sym */
11823 ada_unhandled_exception_name_addr
11824 };
11825
11826 /* The following exception support info structure describes how to
11827 implement exception catchpoints with a slightly older version
11828 of the Ada runtime. */
11829
11830 static const struct exception_support_info exception_support_info_fallback =
11831 {
11832 "__gnat_raise_nodefer_with_msg", /* catch_exception_sym */
11833 "__gnat_unhandled_exception", /* catch_exception_unhandled_sym */
11834 "system__assertions__raise_assert_failure", /* catch_assert_sym */
11835 "__gnat_begin_handler", /* catch_handlers_sym */
11836 ada_unhandled_exception_name_addr_from_raise
11837 };
11838
11839 /* Return nonzero if we can detect the exception support routines
11840 described in EINFO.
11841
11842 This function errors out if an abnormal situation is detected
11843 (for instance, if we find the exception support routines, but
11844 that support is found to be incomplete). */
11845
11846 static int
11847 ada_has_this_exception_support (const struct exception_support_info *einfo)
11848 {
11849 struct symbol *sym;
11850
11851 /* The symbol we're looking up is provided by a unit in the GNAT runtime
11852 that should be compiled with debugging information. As a result, we
11853 expect to find that symbol in the symtabs. */
11854
11855 sym = standard_lookup (einfo->catch_exception_sym, NULL, VAR_DOMAIN);
11856 if (sym == NULL)
11857 {
11858 /* Perhaps we did not find our symbol because the Ada runtime was
11859 compiled without debugging info, or simply stripped of it.
11860 It happens on some GNU/Linux distributions for instance, where
11861 users have to install a separate debug package in order to get
11862 the runtime's debugging info. In that situation, let the user
11863 know why we cannot insert an Ada exception catchpoint.
11864
11865 Note: Just for the purpose of inserting our Ada exception
11866 catchpoint, we could rely purely on the associated minimal symbol.
11867 But we would be operating in degraded mode anyway, since we are
11868 still lacking the debugging info needed later on to extract
11869 the name of the exception being raised (this name is printed in
11870 the catchpoint message, and is also used when trying to catch
11871 a specific exception). We do not handle this case for now. */
11872 struct bound_minimal_symbol msym
11873 = lookup_minimal_symbol (einfo->catch_exception_sym, NULL, NULL);
11874
11875 if (msym.minsym && MSYMBOL_TYPE (msym.minsym) != mst_solib_trampoline)
11876 error (_("Your Ada runtime appears to be missing some debugging "
11877 "information.\nCannot insert Ada exception catchpoint "
11878 "in this configuration."));
11879
11880 return 0;
11881 }
11882
11883 /* Make sure that the symbol we found corresponds to a function. */
11884
11885 if (SYMBOL_CLASS (sym) != LOC_BLOCK)
11886 {
11887 error (_("Symbol \"%s\" is not a function (class = %d)"),
11888 sym->linkage_name (), SYMBOL_CLASS (sym));
11889 return 0;
11890 }
11891
11892 sym = standard_lookup (einfo->catch_handlers_sym, NULL, VAR_DOMAIN);
11893 if (sym == NULL)
11894 {
11895 struct bound_minimal_symbol msym
11896 = lookup_minimal_symbol (einfo->catch_handlers_sym, NULL, NULL);
11897
11898 if (msym.minsym && MSYMBOL_TYPE (msym.minsym) != mst_solib_trampoline)
11899 error (_("Your Ada runtime appears to be missing some debugging "
11900 "information.\nCannot insert Ada exception catchpoint "
11901 "in this configuration."));
11902
11903 return 0;
11904 }
11905
11906 /* Make sure that the symbol we found corresponds to a function. */
11907
11908 if (SYMBOL_CLASS (sym) != LOC_BLOCK)
11909 {
11910 error (_("Symbol \"%s\" is not a function (class = %d)"),
11911 sym->linkage_name (), SYMBOL_CLASS (sym));
11912 return 0;
11913 }
11914
11915 return 1;
11916 }
11917
11918 /* Inspect the Ada runtime and determine which exception info structure
11919 should be used to provide support for exception catchpoints.
11920
11921 This function will always set the per-inferior exception_info,
11922 or raise an error. */
11923
11924 static void
11925 ada_exception_support_info_sniffer (void)
11926 {
11927 struct ada_inferior_data *data = get_ada_inferior_data (current_inferior ());
11928
11929 /* If the exception info is already known, then no need to recompute it. */
11930 if (data->exception_info != NULL)
11931 return;
11932
11933 /* Check the latest (default) exception support info. */
11934 if (ada_has_this_exception_support (&default_exception_support_info))
11935 {
11936 data->exception_info = &default_exception_support_info;
11937 return;
11938 }
11939
11940 /* Try the v0 exception suport info. */
11941 if (ada_has_this_exception_support (&exception_support_info_v0))
11942 {
11943 data->exception_info = &exception_support_info_v0;
11944 return;
11945 }
11946
11947 /* Try our fallback exception suport info. */
11948 if (ada_has_this_exception_support (&exception_support_info_fallback))
11949 {
11950 data->exception_info = &exception_support_info_fallback;
11951 return;
11952 }
11953
11954 /* Sometimes, it is normal for us to not be able to find the routine
11955 we are looking for. This happens when the program is linked with
11956 the shared version of the GNAT runtime, and the program has not been
11957 started yet. Inform the user of these two possible causes if
11958 applicable. */
11959
11960 if (ada_update_initial_language (language_unknown) != language_ada)
11961 error (_("Unable to insert catchpoint. Is this an Ada main program?"));
11962
11963 /* If the symbol does not exist, then check that the program is
11964 already started, to make sure that shared libraries have been
11965 loaded. If it is not started, this may mean that the symbol is
11966 in a shared library. */
11967
11968 if (inferior_ptid.pid () == 0)
11969 error (_("Unable to insert catchpoint. Try to start the program first."));
11970
11971 /* At this point, we know that we are debugging an Ada program and
11972 that the inferior has been started, but we still are not able to
11973 find the run-time symbols. That can mean that we are in
11974 configurable run time mode, or that a-except as been optimized
11975 out by the linker... In any case, at this point it is not worth
11976 supporting this feature. */
11977
11978 error (_("Cannot insert Ada exception catchpoints in this configuration."));
11979 }
11980
11981 /* True iff FRAME is very likely to be that of a function that is
11982 part of the runtime system. This is all very heuristic, but is
11983 intended to be used as advice as to what frames are uninteresting
11984 to most users. */
11985
11986 static int
11987 is_known_support_routine (struct frame_info *frame)
11988 {
11989 enum language func_lang;
11990 int i;
11991 const char *fullname;
11992
11993 /* If this code does not have any debugging information (no symtab),
11994 This cannot be any user code. */
11995
11996 symtab_and_line sal = find_frame_sal (frame);
11997 if (sal.symtab == NULL)
11998 return 1;
11999
12000 /* If there is a symtab, but the associated source file cannot be
12001 located, then assume this is not user code: Selecting a frame
12002 for which we cannot display the code would not be very helpful
12003 for the user. This should also take care of case such as VxWorks
12004 where the kernel has some debugging info provided for a few units. */
12005
12006 fullname = symtab_to_fullname (sal.symtab);
12007 if (access (fullname, R_OK) != 0)
12008 return 1;
12009
12010 /* Check the unit filename against the Ada runtime file naming.
12011 We also check the name of the objfile against the name of some
12012 known system libraries that sometimes come with debugging info
12013 too. */
12014
12015 for (i = 0; known_runtime_file_name_patterns[i] != NULL; i += 1)
12016 {
12017 re_comp (known_runtime_file_name_patterns[i]);
12018 if (re_exec (lbasename (sal.symtab->filename)))
12019 return 1;
12020 if (SYMTAB_OBJFILE (sal.symtab) != NULL
12021 && re_exec (objfile_name (SYMTAB_OBJFILE (sal.symtab))))
12022 return 1;
12023 }
12024
12025 /* Check whether the function is a GNAT-generated entity. */
12026
12027 gdb::unique_xmalloc_ptr<char> func_name
12028 = find_frame_funname (frame, &func_lang, NULL);
12029 if (func_name == NULL)
12030 return 1;
12031
12032 for (i = 0; known_auxiliary_function_name_patterns[i] != NULL; i += 1)
12033 {
12034 re_comp (known_auxiliary_function_name_patterns[i]);
12035 if (re_exec (func_name.get ()))
12036 return 1;
12037 }
12038
12039 return 0;
12040 }
12041
12042 /* Find the first frame that contains debugging information and that is not
12043 part of the Ada run-time, starting from FI and moving upward. */
12044
12045 void
12046 ada_find_printable_frame (struct frame_info *fi)
12047 {
12048 for (; fi != NULL; fi = get_prev_frame (fi))
12049 {
12050 if (!is_known_support_routine (fi))
12051 {
12052 select_frame (fi);
12053 break;
12054 }
12055 }
12056
12057 }
12058
12059 /* Assuming that the inferior just triggered an unhandled exception
12060 catchpoint, return the address in inferior memory where the name
12061 of the exception is stored.
12062
12063 Return zero if the address could not be computed. */
12064
12065 static CORE_ADDR
12066 ada_unhandled_exception_name_addr (void)
12067 {
12068 return parse_and_eval_address ("e.full_name");
12069 }
12070
12071 /* Same as ada_unhandled_exception_name_addr, except that this function
12072 should be used when the inferior uses an older version of the runtime,
12073 where the exception name needs to be extracted from a specific frame
12074 several frames up in the callstack. */
12075
12076 static CORE_ADDR
12077 ada_unhandled_exception_name_addr_from_raise (void)
12078 {
12079 int frame_level;
12080 struct frame_info *fi;
12081 struct ada_inferior_data *data = get_ada_inferior_data (current_inferior ());
12082
12083 /* To determine the name of this exception, we need to select
12084 the frame corresponding to RAISE_SYM_NAME. This frame is
12085 at least 3 levels up, so we simply skip the first 3 frames
12086 without checking the name of their associated function. */
12087 fi = get_current_frame ();
12088 for (frame_level = 0; frame_level < 3; frame_level += 1)
12089 if (fi != NULL)
12090 fi = get_prev_frame (fi);
12091
12092 while (fi != NULL)
12093 {
12094 enum language func_lang;
12095
12096 gdb::unique_xmalloc_ptr<char> func_name
12097 = find_frame_funname (fi, &func_lang, NULL);
12098 if (func_name != NULL)
12099 {
12100 if (strcmp (func_name.get (),
12101 data->exception_info->catch_exception_sym) == 0)
12102 break; /* We found the frame we were looking for... */
12103 }
12104 fi = get_prev_frame (fi);
12105 }
12106
12107 if (fi == NULL)
12108 return 0;
12109
12110 select_frame (fi);
12111 return parse_and_eval_address ("id.full_name");
12112 }
12113
12114 /* Assuming the inferior just triggered an Ada exception catchpoint
12115 (of any type), return the address in inferior memory where the name
12116 of the exception is stored, if applicable.
12117
12118 Assumes the selected frame is the current frame.
12119
12120 Return zero if the address could not be computed, or if not relevant. */
12121
12122 static CORE_ADDR
12123 ada_exception_name_addr_1 (enum ada_exception_catchpoint_kind ex,
12124 struct breakpoint *b)
12125 {
12126 struct ada_inferior_data *data = get_ada_inferior_data (current_inferior ());
12127
12128 switch (ex)
12129 {
12130 case ada_catch_exception:
12131 return (parse_and_eval_address ("e.full_name"));
12132 break;
12133
12134 case ada_catch_exception_unhandled:
12135 return data->exception_info->unhandled_exception_name_addr ();
12136 break;
12137
12138 case ada_catch_handlers:
12139 return 0; /* The runtimes does not provide access to the exception
12140 name. */
12141 break;
12142
12143 case ada_catch_assert:
12144 return 0; /* Exception name is not relevant in this case. */
12145 break;
12146
12147 default:
12148 internal_error (__FILE__, __LINE__, _("unexpected catchpoint type"));
12149 break;
12150 }
12151
12152 return 0; /* Should never be reached. */
12153 }
12154
12155 /* Assuming the inferior is stopped at an exception catchpoint,
12156 return the message which was associated to the exception, if
12157 available. Return NULL if the message could not be retrieved.
12158
12159 Note: The exception message can be associated to an exception
12160 either through the use of the Raise_Exception function, or
12161 more simply (Ada 2005 and later), via:
12162
12163 raise Exception_Name with "exception message";
12164
12165 */
12166
12167 static gdb::unique_xmalloc_ptr<char>
12168 ada_exception_message_1 (void)
12169 {
12170 struct value *e_msg_val;
12171 int e_msg_len;
12172
12173 /* For runtimes that support this feature, the exception message
12174 is passed as an unbounded string argument called "message". */
12175 e_msg_val = parse_and_eval ("message");
12176 if (e_msg_val == NULL)
12177 return NULL; /* Exception message not supported. */
12178
12179 e_msg_val = ada_coerce_to_simple_array (e_msg_val);
12180 gdb_assert (e_msg_val != NULL);
12181 e_msg_len = TYPE_LENGTH (value_type (e_msg_val));
12182
12183 /* If the message string is empty, then treat it as if there was
12184 no exception message. */
12185 if (e_msg_len <= 0)
12186 return NULL;
12187
12188 gdb::unique_xmalloc_ptr<char> e_msg ((char *) xmalloc (e_msg_len + 1));
12189 read_memory_string (value_address (e_msg_val), e_msg.get (), e_msg_len + 1);
12190 e_msg.get ()[e_msg_len] = '\0';
12191
12192 return e_msg;
12193 }
12194
12195 /* Same as ada_exception_message_1, except that all exceptions are
12196 contained here (returning NULL instead). */
12197
12198 static gdb::unique_xmalloc_ptr<char>
12199 ada_exception_message (void)
12200 {
12201 gdb::unique_xmalloc_ptr<char> e_msg;
12202
12203 try
12204 {
12205 e_msg = ada_exception_message_1 ();
12206 }
12207 catch (const gdb_exception_error &e)
12208 {
12209 e_msg.reset (nullptr);
12210 }
12211
12212 return e_msg;
12213 }
12214
12215 /* Same as ada_exception_name_addr_1, except that it intercepts and contains
12216 any error that ada_exception_name_addr_1 might cause to be thrown.
12217 When an error is intercepted, a warning with the error message is printed,
12218 and zero is returned. */
12219
12220 static CORE_ADDR
12221 ada_exception_name_addr (enum ada_exception_catchpoint_kind ex,
12222 struct breakpoint *b)
12223 {
12224 CORE_ADDR result = 0;
12225
12226 try
12227 {
12228 result = ada_exception_name_addr_1 (ex, b);
12229 }
12230
12231 catch (const gdb_exception_error &e)
12232 {
12233 warning (_("failed to get exception name: %s"), e.what ());
12234 return 0;
12235 }
12236
12237 return result;
12238 }
12239
12240 static std::string ada_exception_catchpoint_cond_string
12241 (const char *excep_string,
12242 enum ada_exception_catchpoint_kind ex);
12243
12244 /* Ada catchpoints.
12245
12246 In the case of catchpoints on Ada exceptions, the catchpoint will
12247 stop the target on every exception the program throws. When a user
12248 specifies the name of a specific exception, we translate this
12249 request into a condition expression (in text form), and then parse
12250 it into an expression stored in each of the catchpoint's locations.
12251 We then use this condition to check whether the exception that was
12252 raised is the one the user is interested in. If not, then the
12253 target is resumed again. We store the name of the requested
12254 exception, in order to be able to re-set the condition expression
12255 when symbols change. */
12256
12257 /* An instance of this type is used to represent an Ada catchpoint
12258 breakpoint location. */
12259
12260 class ada_catchpoint_location : public bp_location
12261 {
12262 public:
12263 ada_catchpoint_location (breakpoint *owner)
12264 : bp_location (owner, bp_loc_software_breakpoint)
12265 {}
12266
12267 /* The condition that checks whether the exception that was raised
12268 is the specific exception the user specified on catchpoint
12269 creation. */
12270 expression_up excep_cond_expr;
12271 };
12272
12273 /* An instance of this type is used to represent an Ada catchpoint. */
12274
12275 struct ada_catchpoint : public breakpoint
12276 {
12277 explicit ada_catchpoint (enum ada_exception_catchpoint_kind kind)
12278 : m_kind (kind)
12279 {
12280 }
12281
12282 /* The name of the specific exception the user specified. */
12283 std::string excep_string;
12284
12285 /* What kind of catchpoint this is. */
12286 enum ada_exception_catchpoint_kind m_kind;
12287 };
12288
12289 /* Parse the exception condition string in the context of each of the
12290 catchpoint's locations, and store them for later evaluation. */
12291
12292 static void
12293 create_excep_cond_exprs (struct ada_catchpoint *c,
12294 enum ada_exception_catchpoint_kind ex)
12295 {
12296 struct bp_location *bl;
12297
12298 /* Nothing to do if there's no specific exception to catch. */
12299 if (c->excep_string.empty ())
12300 return;
12301
12302 /* Same if there are no locations... */
12303 if (c->loc == NULL)
12304 return;
12305
12306 /* Compute the condition expression in text form, from the specific
12307 expection we want to catch. */
12308 std::string cond_string
12309 = ada_exception_catchpoint_cond_string (c->excep_string.c_str (), ex);
12310
12311 /* Iterate over all the catchpoint's locations, and parse an
12312 expression for each. */
12313 for (bl = c->loc; bl != NULL; bl = bl->next)
12314 {
12315 struct ada_catchpoint_location *ada_loc
12316 = (struct ada_catchpoint_location *) bl;
12317 expression_up exp;
12318
12319 if (!bl->shlib_disabled)
12320 {
12321 const char *s;
12322
12323 s = cond_string.c_str ();
12324 try
12325 {
12326 exp = parse_exp_1 (&s, bl->address,
12327 block_for_pc (bl->address),
12328 0);
12329 }
12330 catch (const gdb_exception_error &e)
12331 {
12332 warning (_("failed to reevaluate internal exception condition "
12333 "for catchpoint %d: %s"),
12334 c->number, e.what ());
12335 }
12336 }
12337
12338 ada_loc->excep_cond_expr = std::move (exp);
12339 }
12340 }
12341
12342 /* Implement the ALLOCATE_LOCATION method in the breakpoint_ops
12343 structure for all exception catchpoint kinds. */
12344
12345 static struct bp_location *
12346 allocate_location_exception (struct breakpoint *self)
12347 {
12348 return new ada_catchpoint_location (self);
12349 }
12350
12351 /* Implement the RE_SET method in the breakpoint_ops structure for all
12352 exception catchpoint kinds. */
12353
12354 static void
12355 re_set_exception (struct breakpoint *b)
12356 {
12357 struct ada_catchpoint *c = (struct ada_catchpoint *) b;
12358
12359 /* Call the base class's method. This updates the catchpoint's
12360 locations. */
12361 bkpt_breakpoint_ops.re_set (b);
12362
12363 /* Reparse the exception conditional expressions. One for each
12364 location. */
12365 create_excep_cond_exprs (c, c->m_kind);
12366 }
12367
12368 /* Returns true if we should stop for this breakpoint hit. If the
12369 user specified a specific exception, we only want to cause a stop
12370 if the program thrown that exception. */
12371
12372 static int
12373 should_stop_exception (const struct bp_location *bl)
12374 {
12375 struct ada_catchpoint *c = (struct ada_catchpoint *) bl->owner;
12376 const struct ada_catchpoint_location *ada_loc
12377 = (const struct ada_catchpoint_location *) bl;
12378 int stop;
12379
12380 struct internalvar *var = lookup_internalvar ("_ada_exception");
12381 if (c->m_kind == ada_catch_assert)
12382 clear_internalvar (var);
12383 else
12384 {
12385 try
12386 {
12387 const char *expr;
12388
12389 if (c->m_kind == ada_catch_handlers)
12390 expr = ("GNAT_GCC_exception_Access(gcc_exception)"
12391 ".all.occurrence.id");
12392 else
12393 expr = "e";
12394
12395 struct value *exc = parse_and_eval (expr);
12396 set_internalvar (var, exc);
12397 }
12398 catch (const gdb_exception_error &ex)
12399 {
12400 clear_internalvar (var);
12401 }
12402 }
12403
12404 /* With no specific exception, should always stop. */
12405 if (c->excep_string.empty ())
12406 return 1;
12407
12408 if (ada_loc->excep_cond_expr == NULL)
12409 {
12410 /* We will have a NULL expression if back when we were creating
12411 the expressions, this location's had failed to parse. */
12412 return 1;
12413 }
12414
12415 stop = 1;
12416 try
12417 {
12418 struct value *mark;
12419
12420 mark = value_mark ();
12421 stop = value_true (evaluate_expression (ada_loc->excep_cond_expr.get ()));
12422 value_free_to_mark (mark);
12423 }
12424 catch (const gdb_exception &ex)
12425 {
12426 exception_fprintf (gdb_stderr, ex,
12427 _("Error in testing exception condition:\n"));
12428 }
12429
12430 return stop;
12431 }
12432
12433 /* Implement the CHECK_STATUS method in the breakpoint_ops structure
12434 for all exception catchpoint kinds. */
12435
12436 static void
12437 check_status_exception (bpstat bs)
12438 {
12439 bs->stop = should_stop_exception (bs->bp_location_at);
12440 }
12441
12442 /* Implement the PRINT_IT method in the breakpoint_ops structure
12443 for all exception catchpoint kinds. */
12444
12445 static enum print_stop_action
12446 print_it_exception (bpstat bs)
12447 {
12448 struct ui_out *uiout = current_uiout;
12449 struct breakpoint *b = bs->breakpoint_at;
12450
12451 annotate_catchpoint (b->number);
12452
12453 if (uiout->is_mi_like_p ())
12454 {
12455 uiout->field_string ("reason",
12456 async_reason_lookup (EXEC_ASYNC_BREAKPOINT_HIT));
12457 uiout->field_string ("disp", bpdisp_text (b->disposition));
12458 }
12459
12460 uiout->text (b->disposition == disp_del
12461 ? "\nTemporary catchpoint " : "\nCatchpoint ");
12462 uiout->field_signed ("bkptno", b->number);
12463 uiout->text (", ");
12464
12465 /* ada_exception_name_addr relies on the selected frame being the
12466 current frame. Need to do this here because this function may be
12467 called more than once when printing a stop, and below, we'll
12468 select the first frame past the Ada run-time (see
12469 ada_find_printable_frame). */
12470 select_frame (get_current_frame ());
12471
12472 struct ada_catchpoint *c = (struct ada_catchpoint *) b;
12473 switch (c->m_kind)
12474 {
12475 case ada_catch_exception:
12476 case ada_catch_exception_unhandled:
12477 case ada_catch_handlers:
12478 {
12479 const CORE_ADDR addr = ada_exception_name_addr (c->m_kind, b);
12480 char exception_name[256];
12481
12482 if (addr != 0)
12483 {
12484 read_memory (addr, (gdb_byte *) exception_name,
12485 sizeof (exception_name) - 1);
12486 exception_name [sizeof (exception_name) - 1] = '\0';
12487 }
12488 else
12489 {
12490 /* For some reason, we were unable to read the exception
12491 name. This could happen if the Runtime was compiled
12492 without debugging info, for instance. In that case,
12493 just replace the exception name by the generic string
12494 "exception" - it will read as "an exception" in the
12495 notification we are about to print. */
12496 memcpy (exception_name, "exception", sizeof ("exception"));
12497 }
12498 /* In the case of unhandled exception breakpoints, we print
12499 the exception name as "unhandled EXCEPTION_NAME", to make
12500 it clearer to the user which kind of catchpoint just got
12501 hit. We used ui_out_text to make sure that this extra
12502 info does not pollute the exception name in the MI case. */
12503 if (c->m_kind == ada_catch_exception_unhandled)
12504 uiout->text ("unhandled ");
12505 uiout->field_string ("exception-name", exception_name);
12506 }
12507 break;
12508 case ada_catch_assert:
12509 /* In this case, the name of the exception is not really
12510 important. Just print "failed assertion" to make it clearer
12511 that his program just hit an assertion-failure catchpoint.
12512 We used ui_out_text because this info does not belong in
12513 the MI output. */
12514 uiout->text ("failed assertion");
12515 break;
12516 }
12517
12518 gdb::unique_xmalloc_ptr<char> exception_message = ada_exception_message ();
12519 if (exception_message != NULL)
12520 {
12521 uiout->text (" (");
12522 uiout->field_string ("exception-message", exception_message.get ());
12523 uiout->text (")");
12524 }
12525
12526 uiout->text (" at ");
12527 ada_find_printable_frame (get_current_frame ());
12528
12529 return PRINT_SRC_AND_LOC;
12530 }
12531
12532 /* Implement the PRINT_ONE method in the breakpoint_ops structure
12533 for all exception catchpoint kinds. */
12534
12535 static void
12536 print_one_exception (struct breakpoint *b, struct bp_location **last_loc)
12537 {
12538 struct ui_out *uiout = current_uiout;
12539 struct ada_catchpoint *c = (struct ada_catchpoint *) b;
12540 struct value_print_options opts;
12541
12542 get_user_print_options (&opts);
12543
12544 if (opts.addressprint)
12545 uiout->field_skip ("addr");
12546
12547 annotate_field (5);
12548 switch (c->m_kind)
12549 {
12550 case ada_catch_exception:
12551 if (!c->excep_string.empty ())
12552 {
12553 std::string msg = string_printf (_("`%s' Ada exception"),
12554 c->excep_string.c_str ());
12555
12556 uiout->field_string ("what", msg);
12557 }
12558 else
12559 uiout->field_string ("what", "all Ada exceptions");
12560
12561 break;
12562
12563 case ada_catch_exception_unhandled:
12564 uiout->field_string ("what", "unhandled Ada exceptions");
12565 break;
12566
12567 case ada_catch_handlers:
12568 if (!c->excep_string.empty ())
12569 {
12570 uiout->field_fmt ("what",
12571 _("`%s' Ada exception handlers"),
12572 c->excep_string.c_str ());
12573 }
12574 else
12575 uiout->field_string ("what", "all Ada exceptions handlers");
12576 break;
12577
12578 case ada_catch_assert:
12579 uiout->field_string ("what", "failed Ada assertions");
12580 break;
12581
12582 default:
12583 internal_error (__FILE__, __LINE__, _("unexpected catchpoint type"));
12584 break;
12585 }
12586 }
12587
12588 /* Implement the PRINT_MENTION method in the breakpoint_ops structure
12589 for all exception catchpoint kinds. */
12590
12591 static void
12592 print_mention_exception (struct breakpoint *b)
12593 {
12594 struct ada_catchpoint *c = (struct ada_catchpoint *) b;
12595 struct ui_out *uiout = current_uiout;
12596
12597 uiout->text (b->disposition == disp_del ? _("Temporary catchpoint ")
12598 : _("Catchpoint "));
12599 uiout->field_signed ("bkptno", b->number);
12600 uiout->text (": ");
12601
12602 switch (c->m_kind)
12603 {
12604 case ada_catch_exception:
12605 if (!c->excep_string.empty ())
12606 {
12607 std::string info = string_printf (_("`%s' Ada exception"),
12608 c->excep_string.c_str ());
12609 uiout->text (info.c_str ());
12610 }
12611 else
12612 uiout->text (_("all Ada exceptions"));
12613 break;
12614
12615 case ada_catch_exception_unhandled:
12616 uiout->text (_("unhandled Ada exceptions"));
12617 break;
12618
12619 case ada_catch_handlers:
12620 if (!c->excep_string.empty ())
12621 {
12622 std::string info
12623 = string_printf (_("`%s' Ada exception handlers"),
12624 c->excep_string.c_str ());
12625 uiout->text (info.c_str ());
12626 }
12627 else
12628 uiout->text (_("all Ada exceptions handlers"));
12629 break;
12630
12631 case ada_catch_assert:
12632 uiout->text (_("failed Ada assertions"));
12633 break;
12634
12635 default:
12636 internal_error (__FILE__, __LINE__, _("unexpected catchpoint type"));
12637 break;
12638 }
12639 }
12640
12641 /* Implement the PRINT_RECREATE method in the breakpoint_ops structure
12642 for all exception catchpoint kinds. */
12643
12644 static void
12645 print_recreate_exception (struct breakpoint *b, struct ui_file *fp)
12646 {
12647 struct ada_catchpoint *c = (struct ada_catchpoint *) b;
12648
12649 switch (c->m_kind)
12650 {
12651 case ada_catch_exception:
12652 fprintf_filtered (fp, "catch exception");
12653 if (!c->excep_string.empty ())
12654 fprintf_filtered (fp, " %s", c->excep_string.c_str ());
12655 break;
12656
12657 case ada_catch_exception_unhandled:
12658 fprintf_filtered (fp, "catch exception unhandled");
12659 break;
12660
12661 case ada_catch_handlers:
12662 fprintf_filtered (fp, "catch handlers");
12663 break;
12664
12665 case ada_catch_assert:
12666 fprintf_filtered (fp, "catch assert");
12667 break;
12668
12669 default:
12670 internal_error (__FILE__, __LINE__, _("unexpected catchpoint type"));
12671 }
12672 print_recreate_thread (b, fp);
12673 }
12674
12675 /* Virtual tables for various breakpoint types. */
12676 static struct breakpoint_ops catch_exception_breakpoint_ops;
12677 static struct breakpoint_ops catch_exception_unhandled_breakpoint_ops;
12678 static struct breakpoint_ops catch_assert_breakpoint_ops;
12679 static struct breakpoint_ops catch_handlers_breakpoint_ops;
12680
12681 /* See ada-lang.h. */
12682
12683 bool
12684 is_ada_exception_catchpoint (breakpoint *bp)
12685 {
12686 return (bp->ops == &catch_exception_breakpoint_ops
12687 || bp->ops == &catch_exception_unhandled_breakpoint_ops
12688 || bp->ops == &catch_assert_breakpoint_ops
12689 || bp->ops == &catch_handlers_breakpoint_ops);
12690 }
12691
12692 /* Split the arguments specified in a "catch exception" command.
12693 Set EX to the appropriate catchpoint type.
12694 Set EXCEP_STRING to the name of the specific exception if
12695 specified by the user.
12696 IS_CATCH_HANDLERS_CMD: True if the arguments are for a
12697 "catch handlers" command. False otherwise.
12698 If a condition is found at the end of the arguments, the condition
12699 expression is stored in COND_STRING (memory must be deallocated
12700 after use). Otherwise COND_STRING is set to NULL. */
12701
12702 static void
12703 catch_ada_exception_command_split (const char *args,
12704 bool is_catch_handlers_cmd,
12705 enum ada_exception_catchpoint_kind *ex,
12706 std::string *excep_string,
12707 std::string *cond_string)
12708 {
12709 std::string exception_name;
12710
12711 exception_name = extract_arg (&args);
12712 if (exception_name == "if")
12713 {
12714 /* This is not an exception name; this is the start of a condition
12715 expression for a catchpoint on all exceptions. So, "un-get"
12716 this token, and set exception_name to NULL. */
12717 exception_name.clear ();
12718 args -= 2;
12719 }
12720
12721 /* Check to see if we have a condition. */
12722
12723 args = skip_spaces (args);
12724 if (startswith (args, "if")
12725 && (isspace (args[2]) || args[2] == '\0'))
12726 {
12727 args += 2;
12728 args = skip_spaces (args);
12729
12730 if (args[0] == '\0')
12731 error (_("Condition missing after `if' keyword"));
12732 *cond_string = args;
12733
12734 args += strlen (args);
12735 }
12736
12737 /* Check that we do not have any more arguments. Anything else
12738 is unexpected. */
12739
12740 if (args[0] != '\0')
12741 error (_("Junk at end of expression"));
12742
12743 if (is_catch_handlers_cmd)
12744 {
12745 /* Catch handling of exceptions. */
12746 *ex = ada_catch_handlers;
12747 *excep_string = exception_name;
12748 }
12749 else if (exception_name.empty ())
12750 {
12751 /* Catch all exceptions. */
12752 *ex = ada_catch_exception;
12753 excep_string->clear ();
12754 }
12755 else if (exception_name == "unhandled")
12756 {
12757 /* Catch unhandled exceptions. */
12758 *ex = ada_catch_exception_unhandled;
12759 excep_string->clear ();
12760 }
12761 else
12762 {
12763 /* Catch a specific exception. */
12764 *ex = ada_catch_exception;
12765 *excep_string = exception_name;
12766 }
12767 }
12768
12769 /* Return the name of the symbol on which we should break in order to
12770 implement a catchpoint of the EX kind. */
12771
12772 static const char *
12773 ada_exception_sym_name (enum ada_exception_catchpoint_kind ex)
12774 {
12775 struct ada_inferior_data *data = get_ada_inferior_data (current_inferior ());
12776
12777 gdb_assert (data->exception_info != NULL);
12778
12779 switch (ex)
12780 {
12781 case ada_catch_exception:
12782 return (data->exception_info->catch_exception_sym);
12783 break;
12784 case ada_catch_exception_unhandled:
12785 return (data->exception_info->catch_exception_unhandled_sym);
12786 break;
12787 case ada_catch_assert:
12788 return (data->exception_info->catch_assert_sym);
12789 break;
12790 case ada_catch_handlers:
12791 return (data->exception_info->catch_handlers_sym);
12792 break;
12793 default:
12794 internal_error (__FILE__, __LINE__,
12795 _("unexpected catchpoint kind (%d)"), ex);
12796 }
12797 }
12798
12799 /* Return the breakpoint ops "virtual table" used for catchpoints
12800 of the EX kind. */
12801
12802 static const struct breakpoint_ops *
12803 ada_exception_breakpoint_ops (enum ada_exception_catchpoint_kind ex)
12804 {
12805 switch (ex)
12806 {
12807 case ada_catch_exception:
12808 return (&catch_exception_breakpoint_ops);
12809 break;
12810 case ada_catch_exception_unhandled:
12811 return (&catch_exception_unhandled_breakpoint_ops);
12812 break;
12813 case ada_catch_assert:
12814 return (&catch_assert_breakpoint_ops);
12815 break;
12816 case ada_catch_handlers:
12817 return (&catch_handlers_breakpoint_ops);
12818 break;
12819 default:
12820 internal_error (__FILE__, __LINE__,
12821 _("unexpected catchpoint kind (%d)"), ex);
12822 }
12823 }
12824
12825 /* Return the condition that will be used to match the current exception
12826 being raised with the exception that the user wants to catch. This
12827 assumes that this condition is used when the inferior just triggered
12828 an exception catchpoint.
12829 EX: the type of catchpoints used for catching Ada exceptions. */
12830
12831 static std::string
12832 ada_exception_catchpoint_cond_string (const char *excep_string,
12833 enum ada_exception_catchpoint_kind ex)
12834 {
12835 int i;
12836 bool is_standard_exc = false;
12837 std::string result;
12838
12839 if (ex == ada_catch_handlers)
12840 {
12841 /* For exception handlers catchpoints, the condition string does
12842 not use the same parameter as for the other exceptions. */
12843 result = ("long_integer (GNAT_GCC_exception_Access"
12844 "(gcc_exception).all.occurrence.id)");
12845 }
12846 else
12847 result = "long_integer (e)";
12848
12849 /* The standard exceptions are a special case. They are defined in
12850 runtime units that have been compiled without debugging info; if
12851 EXCEP_STRING is the not-fully-qualified name of a standard
12852 exception (e.g. "constraint_error") then, during the evaluation
12853 of the condition expression, the symbol lookup on this name would
12854 *not* return this standard exception. The catchpoint condition
12855 may then be set only on user-defined exceptions which have the
12856 same not-fully-qualified name (e.g. my_package.constraint_error).
12857
12858 To avoid this unexcepted behavior, these standard exceptions are
12859 systematically prefixed by "standard". This means that "catch
12860 exception constraint_error" is rewritten into "catch exception
12861 standard.constraint_error".
12862
12863 If an exception named constraint_error is defined in another package of
12864 the inferior program, then the only way to specify this exception as a
12865 breakpoint condition is to use its fully-qualified named:
12866 e.g. my_package.constraint_error. */
12867
12868 for (i = 0; i < sizeof (standard_exc) / sizeof (char *); i++)
12869 {
12870 if (strcmp (standard_exc [i], excep_string) == 0)
12871 {
12872 is_standard_exc = true;
12873 break;
12874 }
12875 }
12876
12877 result += " = ";
12878
12879 if (is_standard_exc)
12880 string_appendf (result, "long_integer (&standard.%s)", excep_string);
12881 else
12882 string_appendf (result, "long_integer (&%s)", excep_string);
12883
12884 return result;
12885 }
12886
12887 /* Return the symtab_and_line that should be used to insert an exception
12888 catchpoint of the TYPE kind.
12889
12890 ADDR_STRING returns the name of the function where the real
12891 breakpoint that implements the catchpoints is set, depending on the
12892 type of catchpoint we need to create. */
12893
12894 static struct symtab_and_line
12895 ada_exception_sal (enum ada_exception_catchpoint_kind ex,
12896 std::string *addr_string, const struct breakpoint_ops **ops)
12897 {
12898 const char *sym_name;
12899 struct symbol *sym;
12900
12901 /* First, find out which exception support info to use. */
12902 ada_exception_support_info_sniffer ();
12903
12904 /* Then lookup the function on which we will break in order to catch
12905 the Ada exceptions requested by the user. */
12906 sym_name = ada_exception_sym_name (ex);
12907 sym = standard_lookup (sym_name, NULL, VAR_DOMAIN);
12908
12909 if (sym == NULL)
12910 error (_("Catchpoint symbol not found: %s"), sym_name);
12911
12912 if (SYMBOL_CLASS (sym) != LOC_BLOCK)
12913 error (_("Unable to insert catchpoint. %s is not a function."), sym_name);
12914
12915 /* Set ADDR_STRING. */
12916 *addr_string = sym_name;
12917
12918 /* Set OPS. */
12919 *ops = ada_exception_breakpoint_ops (ex);
12920
12921 return find_function_start_sal (sym, 1);
12922 }
12923
12924 /* Create an Ada exception catchpoint.
12925
12926 EX_KIND is the kind of exception catchpoint to be created.
12927
12928 If EXCEPT_STRING is empty, this catchpoint is expected to trigger
12929 for all exceptions. Otherwise, EXCEPT_STRING indicates the name
12930 of the exception to which this catchpoint applies.
12931
12932 COND_STRING, if not empty, is the catchpoint condition.
12933
12934 TEMPFLAG, if nonzero, means that the underlying breakpoint
12935 should be temporary.
12936
12937 FROM_TTY is the usual argument passed to all commands implementations. */
12938
12939 void
12940 create_ada_exception_catchpoint (struct gdbarch *gdbarch,
12941 enum ada_exception_catchpoint_kind ex_kind,
12942 const std::string &excep_string,
12943 const std::string &cond_string,
12944 int tempflag,
12945 int disabled,
12946 int from_tty)
12947 {
12948 std::string addr_string;
12949 const struct breakpoint_ops *ops = NULL;
12950 struct symtab_and_line sal = ada_exception_sal (ex_kind, &addr_string, &ops);
12951
12952 std::unique_ptr<ada_catchpoint> c (new ada_catchpoint (ex_kind));
12953 init_ada_exception_breakpoint (c.get (), gdbarch, sal, addr_string.c_str (),
12954 ops, tempflag, disabled, from_tty);
12955 c->excep_string = excep_string;
12956 create_excep_cond_exprs (c.get (), ex_kind);
12957 if (!cond_string.empty ())
12958 set_breakpoint_condition (c.get (), cond_string.c_str (), from_tty);
12959 install_breakpoint (0, std::move (c), 1);
12960 }
12961
12962 /* Implement the "catch exception" command. */
12963
12964 static void
12965 catch_ada_exception_command (const char *arg_entry, int from_tty,
12966 struct cmd_list_element *command)
12967 {
12968 const char *arg = arg_entry;
12969 struct gdbarch *gdbarch = get_current_arch ();
12970 int tempflag;
12971 enum ada_exception_catchpoint_kind ex_kind;
12972 std::string excep_string;
12973 std::string cond_string;
12974
12975 tempflag = get_cmd_context (command) == CATCH_TEMPORARY;
12976
12977 if (!arg)
12978 arg = "";
12979 catch_ada_exception_command_split (arg, false, &ex_kind, &excep_string,
12980 &cond_string);
12981 create_ada_exception_catchpoint (gdbarch, ex_kind,
12982 excep_string, cond_string,
12983 tempflag, 1 /* enabled */,
12984 from_tty);
12985 }
12986
12987 /* Implement the "catch handlers" command. */
12988
12989 static void
12990 catch_ada_handlers_command (const char *arg_entry, int from_tty,
12991 struct cmd_list_element *command)
12992 {
12993 const char *arg = arg_entry;
12994 struct gdbarch *gdbarch = get_current_arch ();
12995 int tempflag;
12996 enum ada_exception_catchpoint_kind ex_kind;
12997 std::string excep_string;
12998 std::string cond_string;
12999
13000 tempflag = get_cmd_context (command) == CATCH_TEMPORARY;
13001
13002 if (!arg)
13003 arg = "";
13004 catch_ada_exception_command_split (arg, true, &ex_kind, &excep_string,
13005 &cond_string);
13006 create_ada_exception_catchpoint (gdbarch, ex_kind,
13007 excep_string, cond_string,
13008 tempflag, 1 /* enabled */,
13009 from_tty);
13010 }
13011
13012 /* Completion function for the Ada "catch" commands. */
13013
13014 static void
13015 catch_ada_completer (struct cmd_list_element *cmd, completion_tracker &tracker,
13016 const char *text, const char *word)
13017 {
13018 std::vector<ada_exc_info> exceptions = ada_exceptions_list (NULL);
13019
13020 for (const ada_exc_info &info : exceptions)
13021 {
13022 if (startswith (info.name, word))
13023 tracker.add_completion (make_unique_xstrdup (info.name));
13024 }
13025 }
13026
13027 /* Split the arguments specified in a "catch assert" command.
13028
13029 ARGS contains the command's arguments (or the empty string if
13030 no arguments were passed).
13031
13032 If ARGS contains a condition, set COND_STRING to that condition
13033 (the memory needs to be deallocated after use). */
13034
13035 static void
13036 catch_ada_assert_command_split (const char *args, std::string &cond_string)
13037 {
13038 args = skip_spaces (args);
13039
13040 /* Check whether a condition was provided. */
13041 if (startswith (args, "if")
13042 && (isspace (args[2]) || args[2] == '\0'))
13043 {
13044 args += 2;
13045 args = skip_spaces (args);
13046 if (args[0] == '\0')
13047 error (_("condition missing after `if' keyword"));
13048 cond_string.assign (args);
13049 }
13050
13051 /* Otherwise, there should be no other argument at the end of
13052 the command. */
13053 else if (args[0] != '\0')
13054 error (_("Junk at end of arguments."));
13055 }
13056
13057 /* Implement the "catch assert" command. */
13058
13059 static void
13060 catch_assert_command (const char *arg_entry, int from_tty,
13061 struct cmd_list_element *command)
13062 {
13063 const char *arg = arg_entry;
13064 struct gdbarch *gdbarch = get_current_arch ();
13065 int tempflag;
13066 std::string cond_string;
13067
13068 tempflag = get_cmd_context (command) == CATCH_TEMPORARY;
13069
13070 if (!arg)
13071 arg = "";
13072 catch_ada_assert_command_split (arg, cond_string);
13073 create_ada_exception_catchpoint (gdbarch, ada_catch_assert,
13074 "", cond_string,
13075 tempflag, 1 /* enabled */,
13076 from_tty);
13077 }
13078
13079 /* Return non-zero if the symbol SYM is an Ada exception object. */
13080
13081 static int
13082 ada_is_exception_sym (struct symbol *sym)
13083 {
13084 const char *type_name = SYMBOL_TYPE (sym)->name ();
13085
13086 return (SYMBOL_CLASS (sym) != LOC_TYPEDEF
13087 && SYMBOL_CLASS (sym) != LOC_BLOCK
13088 && SYMBOL_CLASS (sym) != LOC_CONST
13089 && SYMBOL_CLASS (sym) != LOC_UNRESOLVED
13090 && type_name != NULL && strcmp (type_name, "exception") == 0);
13091 }
13092
13093 /* Given a global symbol SYM, return non-zero iff SYM is a non-standard
13094 Ada exception object. This matches all exceptions except the ones
13095 defined by the Ada language. */
13096
13097 static int
13098 ada_is_non_standard_exception_sym (struct symbol *sym)
13099 {
13100 int i;
13101
13102 if (!ada_is_exception_sym (sym))
13103 return 0;
13104
13105 for (i = 0; i < ARRAY_SIZE (standard_exc); i++)
13106 if (strcmp (sym->linkage_name (), standard_exc[i]) == 0)
13107 return 0; /* A standard exception. */
13108
13109 /* Numeric_Error is also a standard exception, so exclude it.
13110 See the STANDARD_EXC description for more details as to why
13111 this exception is not listed in that array. */
13112 if (strcmp (sym->linkage_name (), "numeric_error") == 0)
13113 return 0;
13114
13115 return 1;
13116 }
13117
13118 /* A helper function for std::sort, comparing two struct ada_exc_info
13119 objects.
13120
13121 The comparison is determined first by exception name, and then
13122 by exception address. */
13123
13124 bool
13125 ada_exc_info::operator< (const ada_exc_info &other) const
13126 {
13127 int result;
13128
13129 result = strcmp (name, other.name);
13130 if (result < 0)
13131 return true;
13132 if (result == 0 && addr < other.addr)
13133 return true;
13134 return false;
13135 }
13136
13137 bool
13138 ada_exc_info::operator== (const ada_exc_info &other) const
13139 {
13140 return addr == other.addr && strcmp (name, other.name) == 0;
13141 }
13142
13143 /* Sort EXCEPTIONS using compare_ada_exception_info as the comparison
13144 routine, but keeping the first SKIP elements untouched.
13145
13146 All duplicates are also removed. */
13147
13148 static void
13149 sort_remove_dups_ada_exceptions_list (std::vector<ada_exc_info> *exceptions,
13150 int skip)
13151 {
13152 std::sort (exceptions->begin () + skip, exceptions->end ());
13153 exceptions->erase (std::unique (exceptions->begin () + skip, exceptions->end ()),
13154 exceptions->end ());
13155 }
13156
13157 /* Add all exceptions defined by the Ada standard whose name match
13158 a regular expression.
13159
13160 If PREG is not NULL, then this regexp_t object is used to
13161 perform the symbol name matching. Otherwise, no name-based
13162 filtering is performed.
13163
13164 EXCEPTIONS is a vector of exceptions to which matching exceptions
13165 gets pushed. */
13166
13167 static void
13168 ada_add_standard_exceptions (compiled_regex *preg,
13169 std::vector<ada_exc_info> *exceptions)
13170 {
13171 int i;
13172
13173 for (i = 0; i < ARRAY_SIZE (standard_exc); i++)
13174 {
13175 if (preg == NULL
13176 || preg->exec (standard_exc[i], 0, NULL, 0) == 0)
13177 {
13178 struct bound_minimal_symbol msymbol
13179 = ada_lookup_simple_minsym (standard_exc[i]);
13180
13181 if (msymbol.minsym != NULL)
13182 {
13183 struct ada_exc_info info
13184 = {standard_exc[i], BMSYMBOL_VALUE_ADDRESS (msymbol)};
13185
13186 exceptions->push_back (info);
13187 }
13188 }
13189 }
13190 }
13191
13192 /* Add all Ada exceptions defined locally and accessible from the given
13193 FRAME.
13194
13195 If PREG is not NULL, then this regexp_t object is used to
13196 perform the symbol name matching. Otherwise, no name-based
13197 filtering is performed.
13198
13199 EXCEPTIONS is a vector of exceptions to which matching exceptions
13200 gets pushed. */
13201
13202 static void
13203 ada_add_exceptions_from_frame (compiled_regex *preg,
13204 struct frame_info *frame,
13205 std::vector<ada_exc_info> *exceptions)
13206 {
13207 const struct block *block = get_frame_block (frame, 0);
13208
13209 while (block != 0)
13210 {
13211 struct block_iterator iter;
13212 struct symbol *sym;
13213
13214 ALL_BLOCK_SYMBOLS (block, iter, sym)
13215 {
13216 switch (SYMBOL_CLASS (sym))
13217 {
13218 case LOC_TYPEDEF:
13219 case LOC_BLOCK:
13220 case LOC_CONST:
13221 break;
13222 default:
13223 if (ada_is_exception_sym (sym))
13224 {
13225 struct ada_exc_info info = {sym->print_name (),
13226 SYMBOL_VALUE_ADDRESS (sym)};
13227
13228 exceptions->push_back (info);
13229 }
13230 }
13231 }
13232 if (BLOCK_FUNCTION (block) != NULL)
13233 break;
13234 block = BLOCK_SUPERBLOCK (block);
13235 }
13236 }
13237
13238 /* Return true if NAME matches PREG or if PREG is NULL. */
13239
13240 static bool
13241 name_matches_regex (const char *name, compiled_regex *preg)
13242 {
13243 return (preg == NULL
13244 || preg->exec (ada_decode (name).c_str (), 0, NULL, 0) == 0);
13245 }
13246
13247 /* Add all exceptions defined globally whose name name match
13248 a regular expression, excluding standard exceptions.
13249
13250 The reason we exclude standard exceptions is that they need
13251 to be handled separately: Standard exceptions are defined inside
13252 a runtime unit which is normally not compiled with debugging info,
13253 and thus usually do not show up in our symbol search. However,
13254 if the unit was in fact built with debugging info, we need to
13255 exclude them because they would duplicate the entry we found
13256 during the special loop that specifically searches for those
13257 standard exceptions.
13258
13259 If PREG is not NULL, then this regexp_t object is used to
13260 perform the symbol name matching. Otherwise, no name-based
13261 filtering is performed.
13262
13263 EXCEPTIONS is a vector of exceptions to which matching exceptions
13264 gets pushed. */
13265
13266 static void
13267 ada_add_global_exceptions (compiled_regex *preg,
13268 std::vector<ada_exc_info> *exceptions)
13269 {
13270 /* In Ada, the symbol "search name" is a linkage name, whereas the
13271 regular expression used to do the matching refers to the natural
13272 name. So match against the decoded name. */
13273 expand_symtabs_matching (NULL,
13274 lookup_name_info::match_any (),
13275 [&] (const char *search_name)
13276 {
13277 std::string decoded = ada_decode (search_name);
13278 return name_matches_regex (decoded.c_str (), preg);
13279 },
13280 NULL,
13281 VARIABLES_DOMAIN);
13282
13283 for (objfile *objfile : current_program_space->objfiles ())
13284 {
13285 for (compunit_symtab *s : objfile->compunits ())
13286 {
13287 const struct blockvector *bv = COMPUNIT_BLOCKVECTOR (s);
13288 int i;
13289
13290 for (i = GLOBAL_BLOCK; i <= STATIC_BLOCK; i++)
13291 {
13292 const struct block *b = BLOCKVECTOR_BLOCK (bv, i);
13293 struct block_iterator iter;
13294 struct symbol *sym;
13295
13296 ALL_BLOCK_SYMBOLS (b, iter, sym)
13297 if (ada_is_non_standard_exception_sym (sym)
13298 && name_matches_regex (sym->natural_name (), preg))
13299 {
13300 struct ada_exc_info info
13301 = {sym->print_name (), SYMBOL_VALUE_ADDRESS (sym)};
13302
13303 exceptions->push_back (info);
13304 }
13305 }
13306 }
13307 }
13308 }
13309
13310 /* Implements ada_exceptions_list with the regular expression passed
13311 as a regex_t, rather than a string.
13312
13313 If not NULL, PREG is used to filter out exceptions whose names
13314 do not match. Otherwise, all exceptions are listed. */
13315
13316 static std::vector<ada_exc_info>
13317 ada_exceptions_list_1 (compiled_regex *preg)
13318 {
13319 std::vector<ada_exc_info> result;
13320 int prev_len;
13321
13322 /* First, list the known standard exceptions. These exceptions
13323 need to be handled separately, as they are usually defined in
13324 runtime units that have been compiled without debugging info. */
13325
13326 ada_add_standard_exceptions (preg, &result);
13327
13328 /* Next, find all exceptions whose scope is local and accessible
13329 from the currently selected frame. */
13330
13331 if (has_stack_frames ())
13332 {
13333 prev_len = result.size ();
13334 ada_add_exceptions_from_frame (preg, get_selected_frame (NULL),
13335 &result);
13336 if (result.size () > prev_len)
13337 sort_remove_dups_ada_exceptions_list (&result, prev_len);
13338 }
13339
13340 /* Add all exceptions whose scope is global. */
13341
13342 prev_len = result.size ();
13343 ada_add_global_exceptions (preg, &result);
13344 if (result.size () > prev_len)
13345 sort_remove_dups_ada_exceptions_list (&result, prev_len);
13346
13347 return result;
13348 }
13349
13350 /* Return a vector of ada_exc_info.
13351
13352 If REGEXP is NULL, all exceptions are included in the result.
13353 Otherwise, it should contain a valid regular expression,
13354 and only the exceptions whose names match that regular expression
13355 are included in the result.
13356
13357 The exceptions are sorted in the following order:
13358 - Standard exceptions (defined by the Ada language), in
13359 alphabetical order;
13360 - Exceptions only visible from the current frame, in
13361 alphabetical order;
13362 - Exceptions whose scope is global, in alphabetical order. */
13363
13364 std::vector<ada_exc_info>
13365 ada_exceptions_list (const char *regexp)
13366 {
13367 if (regexp == NULL)
13368 return ada_exceptions_list_1 (NULL);
13369
13370 compiled_regex reg (regexp, REG_NOSUB, _("invalid regular expression"));
13371 return ada_exceptions_list_1 (&reg);
13372 }
13373
13374 /* Implement the "info exceptions" command. */
13375
13376 static void
13377 info_exceptions_command (const char *regexp, int from_tty)
13378 {
13379 struct gdbarch *gdbarch = get_current_arch ();
13380
13381 std::vector<ada_exc_info> exceptions = ada_exceptions_list (regexp);
13382
13383 if (regexp != NULL)
13384 printf_filtered
13385 (_("All Ada exceptions matching regular expression \"%s\":\n"), regexp);
13386 else
13387 printf_filtered (_("All defined Ada exceptions:\n"));
13388
13389 for (const ada_exc_info &info : exceptions)
13390 printf_filtered ("%s: %s\n", info.name, paddress (gdbarch, info.addr));
13391 }
13392
13393 /* Operators */
13394 /* Information about operators given special treatment in functions
13395 below. */
13396 /* Format: OP_DEFN (<operator>, <operator length>, <# args>, <binop>). */
13397
13398 #define ADA_OPERATORS \
13399 OP_DEFN (OP_VAR_VALUE, 4, 0, 0) \
13400 OP_DEFN (BINOP_IN_BOUNDS, 3, 2, 0) \
13401 OP_DEFN (TERNOP_IN_RANGE, 1, 3, 0) \
13402 OP_DEFN (OP_ATR_FIRST, 1, 2, 0) \
13403 OP_DEFN (OP_ATR_LAST, 1, 2, 0) \
13404 OP_DEFN (OP_ATR_LENGTH, 1, 2, 0) \
13405 OP_DEFN (OP_ATR_IMAGE, 1, 2, 0) \
13406 OP_DEFN (OP_ATR_MAX, 1, 3, 0) \
13407 OP_DEFN (OP_ATR_MIN, 1, 3, 0) \
13408 OP_DEFN (OP_ATR_MODULUS, 1, 1, 0) \
13409 OP_DEFN (OP_ATR_POS, 1, 2, 0) \
13410 OP_DEFN (OP_ATR_SIZE, 1, 1, 0) \
13411 OP_DEFN (OP_ATR_TAG, 1, 1, 0) \
13412 OP_DEFN (OP_ATR_VAL, 1, 2, 0) \
13413 OP_DEFN (UNOP_QUAL, 3, 1, 0) \
13414 OP_DEFN (UNOP_IN_RANGE, 3, 1, 0) \
13415 OP_DEFN (OP_OTHERS, 1, 1, 0) \
13416 OP_DEFN (OP_POSITIONAL, 3, 1, 0) \
13417 OP_DEFN (OP_DISCRETE_RANGE, 1, 2, 0)
13418
13419 static void
13420 ada_operator_length (const struct expression *exp, int pc, int *oplenp,
13421 int *argsp)
13422 {
13423 switch (exp->elts[pc - 1].opcode)
13424 {
13425 default:
13426 operator_length_standard (exp, pc, oplenp, argsp);
13427 break;
13428
13429 #define OP_DEFN(op, len, args, binop) \
13430 case op: *oplenp = len; *argsp = args; break;
13431 ADA_OPERATORS;
13432 #undef OP_DEFN
13433
13434 case OP_AGGREGATE:
13435 *oplenp = 3;
13436 *argsp = longest_to_int (exp->elts[pc - 2].longconst);
13437 break;
13438
13439 case OP_CHOICES:
13440 *oplenp = 3;
13441 *argsp = longest_to_int (exp->elts[pc - 2].longconst) + 1;
13442 break;
13443 }
13444 }
13445
13446 /* Implementation of the exp_descriptor method operator_check. */
13447
13448 static int
13449 ada_operator_check (struct expression *exp, int pos,
13450 int (*objfile_func) (struct objfile *objfile, void *data),
13451 void *data)
13452 {
13453 const union exp_element *const elts = exp->elts;
13454 struct type *type = NULL;
13455
13456 switch (elts[pos].opcode)
13457 {
13458 case UNOP_IN_RANGE:
13459 case UNOP_QUAL:
13460 type = elts[pos + 1].type;
13461 break;
13462
13463 default:
13464 return operator_check_standard (exp, pos, objfile_func, data);
13465 }
13466
13467 /* Invoke callbacks for TYPE and OBJFILE if they were set as non-NULL. */
13468
13469 if (type && TYPE_OBJFILE (type)
13470 && (*objfile_func) (TYPE_OBJFILE (type), data))
13471 return 1;
13472
13473 return 0;
13474 }
13475
13476 static const char *
13477 ada_op_name (enum exp_opcode opcode)
13478 {
13479 switch (opcode)
13480 {
13481 default:
13482 return op_name_standard (opcode);
13483
13484 #define OP_DEFN(op, len, args, binop) case op: return #op;
13485 ADA_OPERATORS;
13486 #undef OP_DEFN
13487
13488 case OP_AGGREGATE:
13489 return "OP_AGGREGATE";
13490 case OP_CHOICES:
13491 return "OP_CHOICES";
13492 case OP_NAME:
13493 return "OP_NAME";
13494 }
13495 }
13496
13497 /* As for operator_length, but assumes PC is pointing at the first
13498 element of the operator, and gives meaningful results only for the
13499 Ada-specific operators, returning 0 for *OPLENP and *ARGSP otherwise. */
13500
13501 static void
13502 ada_forward_operator_length (struct expression *exp, int pc,
13503 int *oplenp, int *argsp)
13504 {
13505 switch (exp->elts[pc].opcode)
13506 {
13507 default:
13508 *oplenp = *argsp = 0;
13509 break;
13510
13511 #define OP_DEFN(op, len, args, binop) \
13512 case op: *oplenp = len; *argsp = args; break;
13513 ADA_OPERATORS;
13514 #undef OP_DEFN
13515
13516 case OP_AGGREGATE:
13517 *oplenp = 3;
13518 *argsp = longest_to_int (exp->elts[pc + 1].longconst);
13519 break;
13520
13521 case OP_CHOICES:
13522 *oplenp = 3;
13523 *argsp = longest_to_int (exp->elts[pc + 1].longconst) + 1;
13524 break;
13525
13526 case OP_STRING:
13527 case OP_NAME:
13528 {
13529 int len = longest_to_int (exp->elts[pc + 1].longconst);
13530
13531 *oplenp = 4 + BYTES_TO_EXP_ELEM (len + 1);
13532 *argsp = 0;
13533 break;
13534 }
13535 }
13536 }
13537
13538 static int
13539 ada_dump_subexp_body (struct expression *exp, struct ui_file *stream, int elt)
13540 {
13541 enum exp_opcode op = exp->elts[elt].opcode;
13542 int oplen, nargs;
13543 int pc = elt;
13544 int i;
13545
13546 ada_forward_operator_length (exp, elt, &oplen, &nargs);
13547
13548 switch (op)
13549 {
13550 /* Ada attributes ('Foo). */
13551 case OP_ATR_FIRST:
13552 case OP_ATR_LAST:
13553 case OP_ATR_LENGTH:
13554 case OP_ATR_IMAGE:
13555 case OP_ATR_MAX:
13556 case OP_ATR_MIN:
13557 case OP_ATR_MODULUS:
13558 case OP_ATR_POS:
13559 case OP_ATR_SIZE:
13560 case OP_ATR_TAG:
13561 case OP_ATR_VAL:
13562 break;
13563
13564 case UNOP_IN_RANGE:
13565 case UNOP_QUAL:
13566 /* XXX: gdb_sprint_host_address, type_sprint */
13567 fprintf_filtered (stream, _("Type @"));
13568 gdb_print_host_address (exp->elts[pc + 1].type, stream);
13569 fprintf_filtered (stream, " (");
13570 type_print (exp->elts[pc + 1].type, NULL, stream, 0);
13571 fprintf_filtered (stream, ")");
13572 break;
13573 case BINOP_IN_BOUNDS:
13574 fprintf_filtered (stream, " (%d)",
13575 longest_to_int (exp->elts[pc + 2].longconst));
13576 break;
13577 case TERNOP_IN_RANGE:
13578 break;
13579
13580 case OP_AGGREGATE:
13581 case OP_OTHERS:
13582 case OP_DISCRETE_RANGE:
13583 case OP_POSITIONAL:
13584 case OP_CHOICES:
13585 break;
13586
13587 case OP_NAME:
13588 case OP_STRING:
13589 {
13590 char *name = &exp->elts[elt + 2].string;
13591 int len = longest_to_int (exp->elts[elt + 1].longconst);
13592
13593 fprintf_filtered (stream, "Text: `%.*s'", len, name);
13594 break;
13595 }
13596
13597 default:
13598 return dump_subexp_body_standard (exp, stream, elt);
13599 }
13600
13601 elt += oplen;
13602 for (i = 0; i < nargs; i += 1)
13603 elt = dump_subexp (exp, stream, elt);
13604
13605 return elt;
13606 }
13607
13608 /* The Ada extension of print_subexp (q.v.). */
13609
13610 static void
13611 ada_print_subexp (struct expression *exp, int *pos,
13612 struct ui_file *stream, enum precedence prec)
13613 {
13614 int oplen, nargs, i;
13615 int pc = *pos;
13616 enum exp_opcode op = exp->elts[pc].opcode;
13617
13618 ada_forward_operator_length (exp, pc, &oplen, &nargs);
13619
13620 *pos += oplen;
13621 switch (op)
13622 {
13623 default:
13624 *pos -= oplen;
13625 print_subexp_standard (exp, pos, stream, prec);
13626 return;
13627
13628 case OP_VAR_VALUE:
13629 fputs_filtered (exp->elts[pc + 2].symbol->natural_name (), stream);
13630 return;
13631
13632 case BINOP_IN_BOUNDS:
13633 /* XXX: sprint_subexp */
13634 print_subexp (exp, pos, stream, PREC_SUFFIX);
13635 fputs_filtered (" in ", stream);
13636 print_subexp (exp, pos, stream, PREC_SUFFIX);
13637 fputs_filtered ("'range", stream);
13638 if (exp->elts[pc + 1].longconst > 1)
13639 fprintf_filtered (stream, "(%ld)",
13640 (long) exp->elts[pc + 1].longconst);
13641 return;
13642
13643 case TERNOP_IN_RANGE:
13644 if (prec >= PREC_EQUAL)
13645 fputs_filtered ("(", stream);
13646 /* XXX: sprint_subexp */
13647 print_subexp (exp, pos, stream, PREC_SUFFIX);
13648 fputs_filtered (" in ", stream);
13649 print_subexp (exp, pos, stream, PREC_EQUAL);
13650 fputs_filtered (" .. ", stream);
13651 print_subexp (exp, pos, stream, PREC_EQUAL);
13652 if (prec >= PREC_EQUAL)
13653 fputs_filtered (")", stream);
13654 return;
13655
13656 case OP_ATR_FIRST:
13657 case OP_ATR_LAST:
13658 case OP_ATR_LENGTH:
13659 case OP_ATR_IMAGE:
13660 case OP_ATR_MAX:
13661 case OP_ATR_MIN:
13662 case OP_ATR_MODULUS:
13663 case OP_ATR_POS:
13664 case OP_ATR_SIZE:
13665 case OP_ATR_TAG:
13666 case OP_ATR_VAL:
13667 if (exp->elts[*pos].opcode == OP_TYPE)
13668 {
13669 if (exp->elts[*pos + 1].type->code () != TYPE_CODE_VOID)
13670 LA_PRINT_TYPE (exp->elts[*pos + 1].type, "", stream, 0, 0,
13671 &type_print_raw_options);
13672 *pos += 3;
13673 }
13674 else
13675 print_subexp (exp, pos, stream, PREC_SUFFIX);
13676 fprintf_filtered (stream, "'%s", ada_attribute_name (op));
13677 if (nargs > 1)
13678 {
13679 int tem;
13680
13681 for (tem = 1; tem < nargs; tem += 1)
13682 {
13683 fputs_filtered ((tem == 1) ? " (" : ", ", stream);
13684 print_subexp (exp, pos, stream, PREC_ABOVE_COMMA);
13685 }
13686 fputs_filtered (")", stream);
13687 }
13688 return;
13689
13690 case UNOP_QUAL:
13691 type_print (exp->elts[pc + 1].type, "", stream, 0);
13692 fputs_filtered ("'(", stream);
13693 print_subexp (exp, pos, stream, PREC_PREFIX);
13694 fputs_filtered (")", stream);
13695 return;
13696
13697 case UNOP_IN_RANGE:
13698 /* XXX: sprint_subexp */
13699 print_subexp (exp, pos, stream, PREC_SUFFIX);
13700 fputs_filtered (" in ", stream);
13701 LA_PRINT_TYPE (exp->elts[pc + 1].type, "", stream, 1, 0,
13702 &type_print_raw_options);
13703 return;
13704
13705 case OP_DISCRETE_RANGE:
13706 print_subexp (exp, pos, stream, PREC_SUFFIX);
13707 fputs_filtered ("..", stream);
13708 print_subexp (exp, pos, stream, PREC_SUFFIX);
13709 return;
13710
13711 case OP_OTHERS:
13712 fputs_filtered ("others => ", stream);
13713 print_subexp (exp, pos, stream, PREC_SUFFIX);
13714 return;
13715
13716 case OP_CHOICES:
13717 for (i = 0; i < nargs-1; i += 1)
13718 {
13719 if (i > 0)
13720 fputs_filtered ("|", stream);
13721 print_subexp (exp, pos, stream, PREC_SUFFIX);
13722 }
13723 fputs_filtered (" => ", stream);
13724 print_subexp (exp, pos, stream, PREC_SUFFIX);
13725 return;
13726
13727 case OP_POSITIONAL:
13728 print_subexp (exp, pos, stream, PREC_SUFFIX);
13729 return;
13730
13731 case OP_AGGREGATE:
13732 fputs_filtered ("(", stream);
13733 for (i = 0; i < nargs; i += 1)
13734 {
13735 if (i > 0)
13736 fputs_filtered (", ", stream);
13737 print_subexp (exp, pos, stream, PREC_SUFFIX);
13738 }
13739 fputs_filtered (")", stream);
13740 return;
13741 }
13742 }
13743
13744 /* Table mapping opcodes into strings for printing operators
13745 and precedences of the operators. */
13746
13747 static const struct op_print ada_op_print_tab[] = {
13748 {":=", BINOP_ASSIGN, PREC_ASSIGN, 1},
13749 {"or else", BINOP_LOGICAL_OR, PREC_LOGICAL_OR, 0},
13750 {"and then", BINOP_LOGICAL_AND, PREC_LOGICAL_AND, 0},
13751 {"or", BINOP_BITWISE_IOR, PREC_BITWISE_IOR, 0},
13752 {"xor", BINOP_BITWISE_XOR, PREC_BITWISE_XOR, 0},
13753 {"and", BINOP_BITWISE_AND, PREC_BITWISE_AND, 0},
13754 {"=", BINOP_EQUAL, PREC_EQUAL, 0},
13755 {"/=", BINOP_NOTEQUAL, PREC_EQUAL, 0},
13756 {"<=", BINOP_LEQ, PREC_ORDER, 0},
13757 {">=", BINOP_GEQ, PREC_ORDER, 0},
13758 {">", BINOP_GTR, PREC_ORDER, 0},
13759 {"<", BINOP_LESS, PREC_ORDER, 0},
13760 {">>", BINOP_RSH, PREC_SHIFT, 0},
13761 {"<<", BINOP_LSH, PREC_SHIFT, 0},
13762 {"+", BINOP_ADD, PREC_ADD, 0},
13763 {"-", BINOP_SUB, PREC_ADD, 0},
13764 {"&", BINOP_CONCAT, PREC_ADD, 0},
13765 {"*", BINOP_MUL, PREC_MUL, 0},
13766 {"/", BINOP_DIV, PREC_MUL, 0},
13767 {"rem", BINOP_REM, PREC_MUL, 0},
13768 {"mod", BINOP_MOD, PREC_MUL, 0},
13769 {"**", BINOP_EXP, PREC_REPEAT, 0},
13770 {"@", BINOP_REPEAT, PREC_REPEAT, 0},
13771 {"-", UNOP_NEG, PREC_PREFIX, 0},
13772 {"+", UNOP_PLUS, PREC_PREFIX, 0},
13773 {"not ", UNOP_LOGICAL_NOT, PREC_PREFIX, 0},
13774 {"not ", UNOP_COMPLEMENT, PREC_PREFIX, 0},
13775 {"abs ", UNOP_ABS, PREC_PREFIX, 0},
13776 {".all", UNOP_IND, PREC_SUFFIX, 1},
13777 {"'access", UNOP_ADDR, PREC_SUFFIX, 1},
13778 {"'size", OP_ATR_SIZE, PREC_SUFFIX, 1},
13779 {NULL, OP_NULL, PREC_SUFFIX, 0}
13780 };
13781 \f
13782 enum ada_primitive_types {
13783 ada_primitive_type_int,
13784 ada_primitive_type_long,
13785 ada_primitive_type_short,
13786 ada_primitive_type_char,
13787 ada_primitive_type_float,
13788 ada_primitive_type_double,
13789 ada_primitive_type_void,
13790 ada_primitive_type_long_long,
13791 ada_primitive_type_long_double,
13792 ada_primitive_type_natural,
13793 ada_primitive_type_positive,
13794 ada_primitive_type_system_address,
13795 ada_primitive_type_storage_offset,
13796 nr_ada_primitive_types
13797 };
13798
13799 static void
13800 ada_language_arch_info (struct gdbarch *gdbarch,
13801 struct language_arch_info *lai)
13802 {
13803 const struct builtin_type *builtin = builtin_type (gdbarch);
13804
13805 lai->primitive_type_vector
13806 = GDBARCH_OBSTACK_CALLOC (gdbarch, nr_ada_primitive_types + 1,
13807 struct type *);
13808
13809 lai->primitive_type_vector [ada_primitive_type_int]
13810 = arch_integer_type (gdbarch, gdbarch_int_bit (gdbarch),
13811 0, "integer");
13812 lai->primitive_type_vector [ada_primitive_type_long]
13813 = arch_integer_type (gdbarch, gdbarch_long_bit (gdbarch),
13814 0, "long_integer");
13815 lai->primitive_type_vector [ada_primitive_type_short]
13816 = arch_integer_type (gdbarch, gdbarch_short_bit (gdbarch),
13817 0, "short_integer");
13818 lai->string_char_type
13819 = lai->primitive_type_vector [ada_primitive_type_char]
13820 = arch_character_type (gdbarch, TARGET_CHAR_BIT, 0, "character");
13821 lai->primitive_type_vector [ada_primitive_type_float]
13822 = arch_float_type (gdbarch, gdbarch_float_bit (gdbarch),
13823 "float", gdbarch_float_format (gdbarch));
13824 lai->primitive_type_vector [ada_primitive_type_double]
13825 = arch_float_type (gdbarch, gdbarch_double_bit (gdbarch),
13826 "long_float", gdbarch_double_format (gdbarch));
13827 lai->primitive_type_vector [ada_primitive_type_long_long]
13828 = arch_integer_type (gdbarch, gdbarch_long_long_bit (gdbarch),
13829 0, "long_long_integer");
13830 lai->primitive_type_vector [ada_primitive_type_long_double]
13831 = arch_float_type (gdbarch, gdbarch_long_double_bit (gdbarch),
13832 "long_long_float", gdbarch_long_double_format (gdbarch));
13833 lai->primitive_type_vector [ada_primitive_type_natural]
13834 = arch_integer_type (gdbarch, gdbarch_int_bit (gdbarch),
13835 0, "natural");
13836 lai->primitive_type_vector [ada_primitive_type_positive]
13837 = arch_integer_type (gdbarch, gdbarch_int_bit (gdbarch),
13838 0, "positive");
13839 lai->primitive_type_vector [ada_primitive_type_void]
13840 = builtin->builtin_void;
13841
13842 lai->primitive_type_vector [ada_primitive_type_system_address]
13843 = lookup_pointer_type (arch_type (gdbarch, TYPE_CODE_VOID, TARGET_CHAR_BIT,
13844 "void"));
13845 lai->primitive_type_vector [ada_primitive_type_system_address]
13846 ->set_name ("system__address");
13847
13848 /* Create the equivalent of the System.Storage_Elements.Storage_Offset
13849 type. This is a signed integral type whose size is the same as
13850 the size of addresses. */
13851 {
13852 unsigned int addr_length = TYPE_LENGTH
13853 (lai->primitive_type_vector [ada_primitive_type_system_address]);
13854
13855 lai->primitive_type_vector [ada_primitive_type_storage_offset]
13856 = arch_integer_type (gdbarch, addr_length * HOST_CHAR_BIT, 0,
13857 "storage_offset");
13858 }
13859
13860 lai->bool_type_symbol = NULL;
13861 lai->bool_type_default = builtin->builtin_bool;
13862 }
13863 \f
13864 /* Language vector */
13865
13866 /* Not really used, but needed in the ada_language_defn. */
13867
13868 static void
13869 emit_char (int c, struct type *type, struct ui_file *stream, int quoter)
13870 {
13871 ada_emit_char (c, type, stream, quoter, 1);
13872 }
13873
13874 static int
13875 parse (struct parser_state *ps)
13876 {
13877 warnings_issued = 0;
13878 return ada_parse (ps);
13879 }
13880
13881 static const struct exp_descriptor ada_exp_descriptor = {
13882 ada_print_subexp,
13883 ada_operator_length,
13884 ada_operator_check,
13885 ada_op_name,
13886 ada_dump_subexp_body,
13887 ada_evaluate_subexp
13888 };
13889
13890 /* symbol_name_matcher_ftype adapter for wild_match. */
13891
13892 static bool
13893 do_wild_match (const char *symbol_search_name,
13894 const lookup_name_info &lookup_name,
13895 completion_match_result *comp_match_res)
13896 {
13897 return wild_match (symbol_search_name, ada_lookup_name (lookup_name));
13898 }
13899
13900 /* symbol_name_matcher_ftype adapter for full_match. */
13901
13902 static bool
13903 do_full_match (const char *symbol_search_name,
13904 const lookup_name_info &lookup_name,
13905 completion_match_result *comp_match_res)
13906 {
13907 return full_match (symbol_search_name, ada_lookup_name (lookup_name));
13908 }
13909
13910 /* symbol_name_matcher_ftype for exact (verbatim) matches. */
13911
13912 static bool
13913 do_exact_match (const char *symbol_search_name,
13914 const lookup_name_info &lookup_name,
13915 completion_match_result *comp_match_res)
13916 {
13917 return strcmp (symbol_search_name, ada_lookup_name (lookup_name)) == 0;
13918 }
13919
13920 /* Build the Ada lookup name for LOOKUP_NAME. */
13921
13922 ada_lookup_name_info::ada_lookup_name_info (const lookup_name_info &lookup_name)
13923 {
13924 gdb::string_view user_name = lookup_name.name ();
13925
13926 if (user_name[0] == '<')
13927 {
13928 if (user_name.back () == '>')
13929 m_encoded_name
13930 = user_name.substr (1, user_name.size () - 2).to_string ();
13931 else
13932 m_encoded_name
13933 = user_name.substr (1, user_name.size () - 1).to_string ();
13934 m_encoded_p = true;
13935 m_verbatim_p = true;
13936 m_wild_match_p = false;
13937 m_standard_p = false;
13938 }
13939 else
13940 {
13941 m_verbatim_p = false;
13942
13943 m_encoded_p = user_name.find ("__") != gdb::string_view::npos;
13944
13945 if (!m_encoded_p)
13946 {
13947 const char *folded = ada_fold_name (user_name);
13948 const char *encoded = ada_encode_1 (folded, false);
13949 if (encoded != NULL)
13950 m_encoded_name = encoded;
13951 else
13952 m_encoded_name = user_name.to_string ();
13953 }
13954 else
13955 m_encoded_name = user_name.to_string ();
13956
13957 /* Handle the 'package Standard' special case. See description
13958 of m_standard_p. */
13959 if (startswith (m_encoded_name.c_str (), "standard__"))
13960 {
13961 m_encoded_name = m_encoded_name.substr (sizeof ("standard__") - 1);
13962 m_standard_p = true;
13963 }
13964 else
13965 m_standard_p = false;
13966
13967 /* If the name contains a ".", then the user is entering a fully
13968 qualified entity name, and the match must not be done in wild
13969 mode. Similarly, if the user wants to complete what looks
13970 like an encoded name, the match must not be done in wild
13971 mode. Also, in the standard__ special case always do
13972 non-wild matching. */
13973 m_wild_match_p
13974 = (lookup_name.match_type () != symbol_name_match_type::FULL
13975 && !m_encoded_p
13976 && !m_standard_p
13977 && user_name.find ('.') == std::string::npos);
13978 }
13979 }
13980
13981 /* symbol_name_matcher_ftype method for Ada. This only handles
13982 completion mode. */
13983
13984 static bool
13985 ada_symbol_name_matches (const char *symbol_search_name,
13986 const lookup_name_info &lookup_name,
13987 completion_match_result *comp_match_res)
13988 {
13989 return lookup_name.ada ().matches (symbol_search_name,
13990 lookup_name.match_type (),
13991 comp_match_res);
13992 }
13993
13994 /* A name matcher that matches the symbol name exactly, with
13995 strcmp. */
13996
13997 static bool
13998 literal_symbol_name_matcher (const char *symbol_search_name,
13999 const lookup_name_info &lookup_name,
14000 completion_match_result *comp_match_res)
14001 {
14002 gdb::string_view name_view = lookup_name.name ();
14003
14004 if (lookup_name.completion_mode ()
14005 ? (strncmp (symbol_search_name, name_view.data (),
14006 name_view.size ()) == 0)
14007 : symbol_search_name == name_view)
14008 {
14009 if (comp_match_res != NULL)
14010 comp_match_res->set_match (symbol_search_name);
14011 return true;
14012 }
14013 else
14014 return false;
14015 }
14016
14017 /* Implement the "la_get_symbol_name_matcher" language_defn method for
14018 Ada. */
14019
14020 static symbol_name_matcher_ftype *
14021 ada_get_symbol_name_matcher (const lookup_name_info &lookup_name)
14022 {
14023 if (lookup_name.match_type () == symbol_name_match_type::SEARCH_NAME)
14024 return literal_symbol_name_matcher;
14025
14026 if (lookup_name.completion_mode ())
14027 return ada_symbol_name_matches;
14028 else
14029 {
14030 if (lookup_name.ada ().wild_match_p ())
14031 return do_wild_match;
14032 else if (lookup_name.ada ().verbatim_p ())
14033 return do_exact_match;
14034 else
14035 return do_full_match;
14036 }
14037 }
14038
14039 /* Implement the "la_read_var_value" language_defn method for Ada. */
14040
14041 static struct value *
14042 ada_read_var_value (struct symbol *var, const struct block *var_block,
14043 struct frame_info *frame)
14044 {
14045 /* The only case where default_read_var_value is not sufficient
14046 is when VAR is a renaming... */
14047 if (frame != nullptr)
14048 {
14049 const struct block *frame_block = get_frame_block (frame, NULL);
14050 if (frame_block != nullptr && ada_is_renaming_symbol (var))
14051 return ada_read_renaming_var_value (var, frame_block);
14052 }
14053
14054 /* This is a typical case where we expect the default_read_var_value
14055 function to work. */
14056 return default_read_var_value (var, var_block, frame);
14057 }
14058
14059 static const char *ada_extensions[] =
14060 {
14061 ".adb", ".ads", ".a", ".ada", ".dg", NULL
14062 };
14063
14064 extern const struct language_defn ada_language_defn = {
14065 "ada", /* Language name */
14066 "Ada",
14067 language_ada,
14068 range_check_off,
14069 case_sensitive_on, /* Yes, Ada is case-insensitive, but
14070 that's not quite what this means. */
14071 array_row_major,
14072 macro_expansion_no,
14073 ada_extensions,
14074 &ada_exp_descriptor,
14075 parse,
14076 resolve,
14077 ada_printchar, /* Print a character constant */
14078 ada_printstr, /* Function to print string constant */
14079 emit_char, /* Function to print single char (not used) */
14080 ada_print_type, /* Print a type using appropriate syntax */
14081 ada_print_typedef, /* Print a typedef using appropriate syntax */
14082 ada_value_print_inner, /* la_value_print_inner */
14083 ada_value_print, /* Print a top-level value */
14084 ada_read_var_value, /* la_read_var_value */
14085 NULL, /* Language specific skip_trampoline */
14086 NULL, /* name_of_this */
14087 true, /* la_store_sym_names_in_linkage_form_p */
14088 ada_lookup_symbol_nonlocal, /* Looking up non-local symbols. */
14089 basic_lookup_transparent_type, /* lookup_transparent_type */
14090 ada_la_decode, /* Language specific symbol demangler */
14091 ada_sniff_from_mangled_name,
14092 NULL, /* Language specific
14093 class_name_from_physname */
14094 ada_op_print_tab, /* expression operators for printing */
14095 0, /* c-style arrays */
14096 1, /* String lower bound */
14097 ada_get_gdb_completer_word_break_characters,
14098 ada_collect_symbol_completion_matches,
14099 ada_language_arch_info,
14100 ada_print_array_index,
14101 default_pass_by_reference,
14102 ada_watch_location_expression,
14103 ada_get_symbol_name_matcher, /* la_get_symbol_name_matcher */
14104 ada_iterate_over_symbols,
14105 default_search_name_hash,
14106 &ada_varobj_ops,
14107 NULL,
14108 NULL,
14109 ada_is_string_type,
14110 "(...)" /* la_struct_too_deep_ellipsis */
14111 };
14112
14113 /* Command-list for the "set/show ada" prefix command. */
14114 static struct cmd_list_element *set_ada_list;
14115 static struct cmd_list_element *show_ada_list;
14116
14117 static void
14118 initialize_ada_catchpoint_ops (void)
14119 {
14120 struct breakpoint_ops *ops;
14121
14122 initialize_breakpoint_ops ();
14123
14124 ops = &catch_exception_breakpoint_ops;
14125 *ops = bkpt_breakpoint_ops;
14126 ops->allocate_location = allocate_location_exception;
14127 ops->re_set = re_set_exception;
14128 ops->check_status = check_status_exception;
14129 ops->print_it = print_it_exception;
14130 ops->print_one = print_one_exception;
14131 ops->print_mention = print_mention_exception;
14132 ops->print_recreate = print_recreate_exception;
14133
14134 ops = &catch_exception_unhandled_breakpoint_ops;
14135 *ops = bkpt_breakpoint_ops;
14136 ops->allocate_location = allocate_location_exception;
14137 ops->re_set = re_set_exception;
14138 ops->check_status = check_status_exception;
14139 ops->print_it = print_it_exception;
14140 ops->print_one = print_one_exception;
14141 ops->print_mention = print_mention_exception;
14142 ops->print_recreate = print_recreate_exception;
14143
14144 ops = &catch_assert_breakpoint_ops;
14145 *ops = bkpt_breakpoint_ops;
14146 ops->allocate_location = allocate_location_exception;
14147 ops->re_set = re_set_exception;
14148 ops->check_status = check_status_exception;
14149 ops->print_it = print_it_exception;
14150 ops->print_one = print_one_exception;
14151 ops->print_mention = print_mention_exception;
14152 ops->print_recreate = print_recreate_exception;
14153
14154 ops = &catch_handlers_breakpoint_ops;
14155 *ops = bkpt_breakpoint_ops;
14156 ops->allocate_location = allocate_location_exception;
14157 ops->re_set = re_set_exception;
14158 ops->check_status = check_status_exception;
14159 ops->print_it = print_it_exception;
14160 ops->print_one = print_one_exception;
14161 ops->print_mention = print_mention_exception;
14162 ops->print_recreate = print_recreate_exception;
14163 }
14164
14165 /* This module's 'new_objfile' observer. */
14166
14167 static void
14168 ada_new_objfile_observer (struct objfile *objfile)
14169 {
14170 ada_clear_symbol_cache ();
14171 }
14172
14173 /* This module's 'free_objfile' observer. */
14174
14175 static void
14176 ada_free_objfile_observer (struct objfile *objfile)
14177 {
14178 ada_clear_symbol_cache ();
14179 }
14180
14181 void _initialize_ada_language ();
14182 void
14183 _initialize_ada_language ()
14184 {
14185 initialize_ada_catchpoint_ops ();
14186
14187 add_basic_prefix_cmd ("ada", no_class,
14188 _("Prefix command for changing Ada-specific settings."),
14189 &set_ada_list, "set ada ", 0, &setlist);
14190
14191 add_show_prefix_cmd ("ada", no_class,
14192 _("Generic command for showing Ada-specific settings."),
14193 &show_ada_list, "show ada ", 0, &showlist);
14194
14195 add_setshow_boolean_cmd ("trust-PAD-over-XVS", class_obscure,
14196 &trust_pad_over_xvs, _("\
14197 Enable or disable an optimization trusting PAD types over XVS types."), _("\
14198 Show whether an optimization trusting PAD types over XVS types is activated."),
14199 _("\
14200 This is related to the encoding used by the GNAT compiler. The debugger\n\
14201 should normally trust the contents of PAD types, but certain older versions\n\
14202 of GNAT have a bug that sometimes causes the information in the PAD type\n\
14203 to be incorrect. Turning this setting \"off\" allows the debugger to\n\
14204 work around this bug. It is always safe to turn this option \"off\", but\n\
14205 this incurs a slight performance penalty, so it is recommended to NOT change\n\
14206 this option to \"off\" unless necessary."),
14207 NULL, NULL, &set_ada_list, &show_ada_list);
14208
14209 add_setshow_boolean_cmd ("print-signatures", class_vars,
14210 &print_signatures, _("\
14211 Enable or disable the output of formal and return types for functions in the \
14212 overloads selection menu."), _("\
14213 Show whether the output of formal and return types for functions in the \
14214 overloads selection menu is activated."),
14215 NULL, NULL, NULL, &set_ada_list, &show_ada_list);
14216
14217 add_catch_command ("exception", _("\
14218 Catch Ada exceptions, when raised.\n\
14219 Usage: catch exception [ARG] [if CONDITION]\n\
14220 Without any argument, stop when any Ada exception is raised.\n\
14221 If ARG is \"unhandled\" (without the quotes), only stop when the exception\n\
14222 being raised does not have a handler (and will therefore lead to the task's\n\
14223 termination).\n\
14224 Otherwise, the catchpoint only stops when the name of the exception being\n\
14225 raised is the same as ARG.\n\
14226 CONDITION is a boolean expression that is evaluated to see whether the\n\
14227 exception should cause a stop."),
14228 catch_ada_exception_command,
14229 catch_ada_completer,
14230 CATCH_PERMANENT,
14231 CATCH_TEMPORARY);
14232
14233 add_catch_command ("handlers", _("\
14234 Catch Ada exceptions, when handled.\n\
14235 Usage: catch handlers [ARG] [if CONDITION]\n\
14236 Without any argument, stop when any Ada exception is handled.\n\
14237 With an argument, catch only exceptions with the given name.\n\
14238 CONDITION is a boolean expression that is evaluated to see whether the\n\
14239 exception should cause a stop."),
14240 catch_ada_handlers_command,
14241 catch_ada_completer,
14242 CATCH_PERMANENT,
14243 CATCH_TEMPORARY);
14244 add_catch_command ("assert", _("\
14245 Catch failed Ada assertions, when raised.\n\
14246 Usage: catch assert [if CONDITION]\n\
14247 CONDITION is a boolean expression that is evaluated to see whether the\n\
14248 exception should cause a stop."),
14249 catch_assert_command,
14250 NULL,
14251 CATCH_PERMANENT,
14252 CATCH_TEMPORARY);
14253
14254 varsize_limit = 65536;
14255 add_setshow_uinteger_cmd ("varsize-limit", class_support,
14256 &varsize_limit, _("\
14257 Set the maximum number of bytes allowed in a variable-size object."), _("\
14258 Show the maximum number of bytes allowed in a variable-size object."), _("\
14259 Attempts to access an object whose size is not a compile-time constant\n\
14260 and exceeds this limit will cause an error."),
14261 NULL, NULL, &setlist, &showlist);
14262
14263 add_info ("exceptions", info_exceptions_command,
14264 _("\
14265 List all Ada exception names.\n\
14266 Usage: info exceptions [REGEXP]\n\
14267 If a regular expression is passed as an argument, only those matching\n\
14268 the regular expression are listed."));
14269
14270 add_basic_prefix_cmd ("ada", class_maintenance,
14271 _("Set Ada maintenance-related variables."),
14272 &maint_set_ada_cmdlist, "maintenance set ada ",
14273 0/*allow-unknown*/, &maintenance_set_cmdlist);
14274
14275 add_show_prefix_cmd ("ada", class_maintenance,
14276 _("Show Ada maintenance-related variables."),
14277 &maint_show_ada_cmdlist, "maintenance show ada ",
14278 0/*allow-unknown*/, &maintenance_show_cmdlist);
14279
14280 add_setshow_boolean_cmd
14281 ("ignore-descriptive-types", class_maintenance,
14282 &ada_ignore_descriptive_types_p,
14283 _("Set whether descriptive types generated by GNAT should be ignored."),
14284 _("Show whether descriptive types generated by GNAT should be ignored."),
14285 _("\
14286 When enabled, the debugger will stop using the DW_AT_GNAT_descriptive_type\n\
14287 DWARF attribute."),
14288 NULL, NULL, &maint_set_ada_cmdlist, &maint_show_ada_cmdlist);
14289
14290 decoded_names_store = htab_create_alloc (256, htab_hash_string, streq_hash,
14291 NULL, xcalloc, xfree);
14292
14293 /* The ada-lang observers. */
14294 gdb::observers::new_objfile.attach (ada_new_objfile_observer);
14295 gdb::observers::free_objfile.attach (ada_free_objfile_observer);
14296 gdb::observers::inferior_exit.attach (ada_inferior_exit);
14297 }
This page took 0.298916 seconds and 3 git commands to generate.