dee3a83f98c24854777f967b7998616e65e90b2c
[deliverable/binutils-gdb.git] / gdb / ada-lang.c
1 /* Ada language support routines for GDB, the GNU debugger.
2
3 Copyright (C) 1992-2019 Free Software Foundation, Inc.
4
5 This file is part of GDB.
6
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 3 of the License, or
10 (at your option) any later version.
11
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
16
17 You should have received a copy of the GNU General Public License
18 along with this program. If not, see <http://www.gnu.org/licenses/>. */
19
20
21 #include "defs.h"
22 #include <ctype.h>
23 #include "demangle.h"
24 #include "gdb_regex.h"
25 #include "frame.h"
26 #include "symtab.h"
27 #include "gdbtypes.h"
28 #include "gdbcmd.h"
29 #include "expression.h"
30 #include "parser-defs.h"
31 #include "language.h"
32 #include "varobj.h"
33 #include "c-lang.h"
34 #include "inferior.h"
35 #include "symfile.h"
36 #include "objfiles.h"
37 #include "breakpoint.h"
38 #include "gdbcore.h"
39 #include "hashtab.h"
40 #include "gdb_obstack.h"
41 #include "ada-lang.h"
42 #include "completer.h"
43 #include <sys/stat.h>
44 #include "ui-out.h"
45 #include "block.h"
46 #include "infcall.h"
47 #include "dictionary.h"
48 #include "annotate.h"
49 #include "valprint.h"
50 #include "source.h"
51 #include "observable.h"
52 #include "common/vec.h"
53 #include "stack.h"
54 #include "common/gdb_vecs.h"
55 #include "typeprint.h"
56 #include "namespace.h"
57
58 #include "psymtab.h"
59 #include "value.h"
60 #include "mi/mi-common.h"
61 #include "arch-utils.h"
62 #include "cli/cli-utils.h"
63 #include "common/function-view.h"
64 #include "common/byte-vector.h"
65 #include <algorithm>
66 #include <map>
67
68 /* Define whether or not the C operator '/' truncates towards zero for
69 differently signed operands (truncation direction is undefined in C).
70 Copied from valarith.c. */
71
72 #ifndef TRUNCATION_TOWARDS_ZERO
73 #define TRUNCATION_TOWARDS_ZERO ((-5 / 2) == -2)
74 #endif
75
76 static struct type *desc_base_type (struct type *);
77
78 static struct type *desc_bounds_type (struct type *);
79
80 static struct value *desc_bounds (struct value *);
81
82 static int fat_pntr_bounds_bitpos (struct type *);
83
84 static int fat_pntr_bounds_bitsize (struct type *);
85
86 static struct type *desc_data_target_type (struct type *);
87
88 static struct value *desc_data (struct value *);
89
90 static int fat_pntr_data_bitpos (struct type *);
91
92 static int fat_pntr_data_bitsize (struct type *);
93
94 static struct value *desc_one_bound (struct value *, int, int);
95
96 static int desc_bound_bitpos (struct type *, int, int);
97
98 static int desc_bound_bitsize (struct type *, int, int);
99
100 static struct type *desc_index_type (struct type *, int);
101
102 static int desc_arity (struct type *);
103
104 static int ada_type_match (struct type *, struct type *, int);
105
106 static int ada_args_match (struct symbol *, struct value **, int);
107
108 static struct value *make_array_descriptor (struct type *, struct value *);
109
110 static void ada_add_block_symbols (struct obstack *,
111 const struct block *,
112 const lookup_name_info &lookup_name,
113 domain_enum, struct objfile *);
114
115 static void ada_add_all_symbols (struct obstack *, const struct block *,
116 const lookup_name_info &lookup_name,
117 domain_enum, int, int *);
118
119 static int is_nonfunction (struct block_symbol *, int);
120
121 static void add_defn_to_vec (struct obstack *, struct symbol *,
122 const struct block *);
123
124 static int num_defns_collected (struct obstack *);
125
126 static struct block_symbol *defns_collected (struct obstack *, int);
127
128 static struct value *resolve_subexp (expression_up *, int *, int,
129 struct type *, int,
130 innermost_block_tracker *);
131
132 static void replace_operator_with_call (expression_up *, int, int, int,
133 struct symbol *, const struct block *);
134
135 static int possible_user_operator_p (enum exp_opcode, struct value **);
136
137 static const char *ada_op_name (enum exp_opcode);
138
139 static const char *ada_decoded_op_name (enum exp_opcode);
140
141 static int numeric_type_p (struct type *);
142
143 static int integer_type_p (struct type *);
144
145 static int scalar_type_p (struct type *);
146
147 static int discrete_type_p (struct type *);
148
149 static enum ada_renaming_category parse_old_style_renaming (struct type *,
150 const char **,
151 int *,
152 const char **);
153
154 static struct symbol *find_old_style_renaming_symbol (const char *,
155 const struct block *);
156
157 static struct type *ada_lookup_struct_elt_type (struct type *, const char *,
158 int, int);
159
160 static struct value *evaluate_subexp_type (struct expression *, int *);
161
162 static struct type *ada_find_parallel_type_with_name (struct type *,
163 const char *);
164
165 static int is_dynamic_field (struct type *, int);
166
167 static struct type *to_fixed_variant_branch_type (struct type *,
168 const gdb_byte *,
169 CORE_ADDR, struct value *);
170
171 static struct type *to_fixed_array_type (struct type *, struct value *, int);
172
173 static struct type *to_fixed_range_type (struct type *, struct value *);
174
175 static struct type *to_static_fixed_type (struct type *);
176 static struct type *static_unwrap_type (struct type *type);
177
178 static struct value *unwrap_value (struct value *);
179
180 static struct type *constrained_packed_array_type (struct type *, long *);
181
182 static struct type *decode_constrained_packed_array_type (struct type *);
183
184 static long decode_packed_array_bitsize (struct type *);
185
186 static struct value *decode_constrained_packed_array (struct value *);
187
188 static int ada_is_packed_array_type (struct type *);
189
190 static int ada_is_unconstrained_packed_array_type (struct type *);
191
192 static struct value *value_subscript_packed (struct value *, int,
193 struct value **);
194
195 static struct value *coerce_unspec_val_to_type (struct value *,
196 struct type *);
197
198 static int lesseq_defined_than (struct symbol *, struct symbol *);
199
200 static int equiv_types (struct type *, struct type *);
201
202 static int is_name_suffix (const char *);
203
204 static int advance_wild_match (const char **, const char *, int);
205
206 static bool wild_match (const char *name, const char *patn);
207
208 static struct value *ada_coerce_ref (struct value *);
209
210 static LONGEST pos_atr (struct value *);
211
212 static struct value *value_pos_atr (struct type *, struct value *);
213
214 static struct value *value_val_atr (struct type *, struct value *);
215
216 static struct symbol *standard_lookup (const char *, const struct block *,
217 domain_enum);
218
219 static struct value *ada_search_struct_field (const char *, struct value *, int,
220 struct type *);
221
222 static struct value *ada_value_primitive_field (struct value *, int, int,
223 struct type *);
224
225 static int find_struct_field (const char *, struct type *, int,
226 struct type **, int *, int *, int *, int *);
227
228 static int ada_resolve_function (struct block_symbol *, int,
229 struct value **, int, const char *,
230 struct type *, int);
231
232 static int ada_is_direct_array_type (struct type *);
233
234 static void ada_language_arch_info (struct gdbarch *,
235 struct language_arch_info *);
236
237 static struct value *ada_index_struct_field (int, struct value *, int,
238 struct type *);
239
240 static struct value *assign_aggregate (struct value *, struct value *,
241 struct expression *,
242 int *, enum noside);
243
244 static void aggregate_assign_from_choices (struct value *, struct value *,
245 struct expression *,
246 int *, LONGEST *, int *,
247 int, LONGEST, LONGEST);
248
249 static void aggregate_assign_positional (struct value *, struct value *,
250 struct expression *,
251 int *, LONGEST *, int *, int,
252 LONGEST, LONGEST);
253
254
255 static void aggregate_assign_others (struct value *, struct value *,
256 struct expression *,
257 int *, LONGEST *, int, LONGEST, LONGEST);
258
259
260 static void add_component_interval (LONGEST, LONGEST, LONGEST *, int *, int);
261
262
263 static struct value *ada_evaluate_subexp (struct type *, struct expression *,
264 int *, enum noside);
265
266 static void ada_forward_operator_length (struct expression *, int, int *,
267 int *);
268
269 static struct type *ada_find_any_type (const char *name);
270
271 static symbol_name_matcher_ftype *ada_get_symbol_name_matcher
272 (const lookup_name_info &lookup_name);
273
274 \f
275
276 /* The result of a symbol lookup to be stored in our symbol cache. */
277
278 struct cache_entry
279 {
280 /* The name used to perform the lookup. */
281 const char *name;
282 /* The namespace used during the lookup. */
283 domain_enum domain;
284 /* The symbol returned by the lookup, or NULL if no matching symbol
285 was found. */
286 struct symbol *sym;
287 /* The block where the symbol was found, or NULL if no matching
288 symbol was found. */
289 const struct block *block;
290 /* A pointer to the next entry with the same hash. */
291 struct cache_entry *next;
292 };
293
294 /* The Ada symbol cache, used to store the result of Ada-mode symbol
295 lookups in the course of executing the user's commands.
296
297 The cache is implemented using a simple, fixed-sized hash.
298 The size is fixed on the grounds that there are not likely to be
299 all that many symbols looked up during any given session, regardless
300 of the size of the symbol table. If we decide to go to a resizable
301 table, let's just use the stuff from libiberty instead. */
302
303 #define HASH_SIZE 1009
304
305 struct ada_symbol_cache
306 {
307 /* An obstack used to store the entries in our cache. */
308 struct obstack cache_space;
309
310 /* The root of the hash table used to implement our symbol cache. */
311 struct cache_entry *root[HASH_SIZE];
312 };
313
314 static void ada_free_symbol_cache (struct ada_symbol_cache *sym_cache);
315
316 /* Maximum-sized dynamic type. */
317 static unsigned int varsize_limit;
318
319 static const char ada_completer_word_break_characters[] =
320 #ifdef VMS
321 " \t\n!@#%^&*()+=|~`}{[]\";:?/,-";
322 #else
323 " \t\n!@#$%^&*()+=|~`}{[]\";:?/,-";
324 #endif
325
326 /* The name of the symbol to use to get the name of the main subprogram. */
327 static const char ADA_MAIN_PROGRAM_SYMBOL_NAME[]
328 = "__gnat_ada_main_program_name";
329
330 /* Limit on the number of warnings to raise per expression evaluation. */
331 static int warning_limit = 2;
332
333 /* Number of warning messages issued; reset to 0 by cleanups after
334 expression evaluation. */
335 static int warnings_issued = 0;
336
337 static const char *known_runtime_file_name_patterns[] = {
338 ADA_KNOWN_RUNTIME_FILE_NAME_PATTERNS NULL
339 };
340
341 static const char *known_auxiliary_function_name_patterns[] = {
342 ADA_KNOWN_AUXILIARY_FUNCTION_NAME_PATTERNS NULL
343 };
344
345 /* Maintenance-related settings for this module. */
346
347 static struct cmd_list_element *maint_set_ada_cmdlist;
348 static struct cmd_list_element *maint_show_ada_cmdlist;
349
350 /* Implement the "maintenance set ada" (prefix) command. */
351
352 static void
353 maint_set_ada_cmd (const char *args, int from_tty)
354 {
355 help_list (maint_set_ada_cmdlist, "maintenance set ada ", all_commands,
356 gdb_stdout);
357 }
358
359 /* Implement the "maintenance show ada" (prefix) command. */
360
361 static void
362 maint_show_ada_cmd (const char *args, int from_tty)
363 {
364 cmd_show_list (maint_show_ada_cmdlist, from_tty, "");
365 }
366
367 /* The "maintenance ada set/show ignore-descriptive-type" value. */
368
369 static int ada_ignore_descriptive_types_p = 0;
370
371 /* Inferior-specific data. */
372
373 /* Per-inferior data for this module. */
374
375 struct ada_inferior_data
376 {
377 /* The ada__tags__type_specific_data type, which is used when decoding
378 tagged types. With older versions of GNAT, this type was directly
379 accessible through a component ("tsd") in the object tag. But this
380 is no longer the case, so we cache it for each inferior. */
381 struct type *tsd_type = nullptr;
382
383 /* The exception_support_info data. This data is used to determine
384 how to implement support for Ada exception catchpoints in a given
385 inferior. */
386 const struct exception_support_info *exception_info = nullptr;
387 };
388
389 /* Our key to this module's inferior data. */
390 static const struct inferior_key<ada_inferior_data> ada_inferior_data;
391
392 /* Return our inferior data for the given inferior (INF).
393
394 This function always returns a valid pointer to an allocated
395 ada_inferior_data structure. If INF's inferior data has not
396 been previously set, this functions creates a new one with all
397 fields set to zero, sets INF's inferior to it, and then returns
398 a pointer to that newly allocated ada_inferior_data. */
399
400 static struct ada_inferior_data *
401 get_ada_inferior_data (struct inferior *inf)
402 {
403 struct ada_inferior_data *data;
404
405 data = ada_inferior_data.get (inf);
406 if (data == NULL)
407 data = ada_inferior_data.emplace (inf);
408
409 return data;
410 }
411
412 /* Perform all necessary cleanups regarding our module's inferior data
413 that is required after the inferior INF just exited. */
414
415 static void
416 ada_inferior_exit (struct inferior *inf)
417 {
418 ada_inferior_data.clear (inf);
419 }
420
421
422 /* program-space-specific data. */
423
424 /* This module's per-program-space data. */
425 struct ada_pspace_data
426 {
427 ~ada_pspace_data ()
428 {
429 if (sym_cache != NULL)
430 ada_free_symbol_cache (sym_cache);
431 }
432
433 /* The Ada symbol cache. */
434 struct ada_symbol_cache *sym_cache = nullptr;
435 };
436
437 /* Key to our per-program-space data. */
438 static const struct program_space_key<ada_pspace_data> ada_pspace_data_handle;
439
440 /* Return this module's data for the given program space (PSPACE).
441 If not is found, add a zero'ed one now.
442
443 This function always returns a valid object. */
444
445 static struct ada_pspace_data *
446 get_ada_pspace_data (struct program_space *pspace)
447 {
448 struct ada_pspace_data *data;
449
450 data = ada_pspace_data_handle.get (pspace);
451 if (data == NULL)
452 data = ada_pspace_data_handle.emplace (pspace);
453
454 return data;
455 }
456
457 /* Utilities */
458
459 /* If TYPE is a TYPE_CODE_TYPEDEF type, return the target type after
460 all typedef layers have been peeled. Otherwise, return TYPE.
461
462 Normally, we really expect a typedef type to only have 1 typedef layer.
463 In other words, we really expect the target type of a typedef type to be
464 a non-typedef type. This is particularly true for Ada units, because
465 the language does not have a typedef vs not-typedef distinction.
466 In that respect, the Ada compiler has been trying to eliminate as many
467 typedef definitions in the debugging information, since they generally
468 do not bring any extra information (we still use typedef under certain
469 circumstances related mostly to the GNAT encoding).
470
471 Unfortunately, we have seen situations where the debugging information
472 generated by the compiler leads to such multiple typedef layers. For
473 instance, consider the following example with stabs:
474
475 .stabs "pck__float_array___XUP:Tt(0,46)=s16P_ARRAY:(0,47)=[...]"[...]
476 .stabs "pck__float_array___XUP:t(0,36)=(0,46)",128,0,6,0
477
478 This is an error in the debugging information which causes type
479 pck__float_array___XUP to be defined twice, and the second time,
480 it is defined as a typedef of a typedef.
481
482 This is on the fringe of legality as far as debugging information is
483 concerned, and certainly unexpected. But it is easy to handle these
484 situations correctly, so we can afford to be lenient in this case. */
485
486 static struct type *
487 ada_typedef_target_type (struct type *type)
488 {
489 while (TYPE_CODE (type) == TYPE_CODE_TYPEDEF)
490 type = TYPE_TARGET_TYPE (type);
491 return type;
492 }
493
494 /* Given DECODED_NAME a string holding a symbol name in its
495 decoded form (ie using the Ada dotted notation), returns
496 its unqualified name. */
497
498 static const char *
499 ada_unqualified_name (const char *decoded_name)
500 {
501 const char *result;
502
503 /* If the decoded name starts with '<', it means that the encoded
504 name does not follow standard naming conventions, and thus that
505 it is not your typical Ada symbol name. Trying to unqualify it
506 is therefore pointless and possibly erroneous. */
507 if (decoded_name[0] == '<')
508 return decoded_name;
509
510 result = strrchr (decoded_name, '.');
511 if (result != NULL)
512 result++; /* Skip the dot... */
513 else
514 result = decoded_name;
515
516 return result;
517 }
518
519 /* Return a string starting with '<', followed by STR, and '>'. */
520
521 static std::string
522 add_angle_brackets (const char *str)
523 {
524 return string_printf ("<%s>", str);
525 }
526
527 static const char *
528 ada_get_gdb_completer_word_break_characters (void)
529 {
530 return ada_completer_word_break_characters;
531 }
532
533 /* Print an array element index using the Ada syntax. */
534
535 static void
536 ada_print_array_index (struct value *index_value, struct ui_file *stream,
537 const struct value_print_options *options)
538 {
539 LA_VALUE_PRINT (index_value, stream, options);
540 fprintf_filtered (stream, " => ");
541 }
542
543 /* la_watch_location_expression for Ada. */
544
545 gdb::unique_xmalloc_ptr<char>
546 ada_watch_location_expression (struct type *type, CORE_ADDR addr)
547 {
548 type = check_typedef (TYPE_TARGET_TYPE (check_typedef (type)));
549 std::string name = type_to_string (type);
550 return gdb::unique_xmalloc_ptr<char>
551 (xstrprintf ("{%s} %s", name.c_str (), core_addr_to_string (addr)));
552 }
553
554 /* Assuming VECT points to an array of *SIZE objects of size
555 ELEMENT_SIZE, grow it to contain at least MIN_SIZE objects,
556 updating *SIZE as necessary and returning the (new) array. */
557
558 void *
559 grow_vect (void *vect, size_t *size, size_t min_size, int element_size)
560 {
561 if (*size < min_size)
562 {
563 *size *= 2;
564 if (*size < min_size)
565 *size = min_size;
566 vect = xrealloc (vect, *size * element_size);
567 }
568 return vect;
569 }
570
571 /* True (non-zero) iff TARGET matches FIELD_NAME up to any trailing
572 suffix of FIELD_NAME beginning "___". */
573
574 static int
575 field_name_match (const char *field_name, const char *target)
576 {
577 int len = strlen (target);
578
579 return
580 (strncmp (field_name, target, len) == 0
581 && (field_name[len] == '\0'
582 || (startswith (field_name + len, "___")
583 && strcmp (field_name + strlen (field_name) - 6,
584 "___XVN") != 0)));
585 }
586
587
588 /* Assuming TYPE is a TYPE_CODE_STRUCT or a TYPE_CODE_TYPDEF to
589 a TYPE_CODE_STRUCT, find the field whose name matches FIELD_NAME,
590 and return its index. This function also handles fields whose name
591 have ___ suffixes because the compiler sometimes alters their name
592 by adding such a suffix to represent fields with certain constraints.
593 If the field could not be found, return a negative number if
594 MAYBE_MISSING is set. Otherwise raise an error. */
595
596 int
597 ada_get_field_index (const struct type *type, const char *field_name,
598 int maybe_missing)
599 {
600 int fieldno;
601 struct type *struct_type = check_typedef ((struct type *) type);
602
603 for (fieldno = 0; fieldno < TYPE_NFIELDS (struct_type); fieldno++)
604 if (field_name_match (TYPE_FIELD_NAME (struct_type, fieldno), field_name))
605 return fieldno;
606
607 if (!maybe_missing)
608 error (_("Unable to find field %s in struct %s. Aborting"),
609 field_name, TYPE_NAME (struct_type));
610
611 return -1;
612 }
613
614 /* The length of the prefix of NAME prior to any "___" suffix. */
615
616 int
617 ada_name_prefix_len (const char *name)
618 {
619 if (name == NULL)
620 return 0;
621 else
622 {
623 const char *p = strstr (name, "___");
624
625 if (p == NULL)
626 return strlen (name);
627 else
628 return p - name;
629 }
630 }
631
632 /* Return non-zero if SUFFIX is a suffix of STR.
633 Return zero if STR is null. */
634
635 static int
636 is_suffix (const char *str, const char *suffix)
637 {
638 int len1, len2;
639
640 if (str == NULL)
641 return 0;
642 len1 = strlen (str);
643 len2 = strlen (suffix);
644 return (len1 >= len2 && strcmp (str + len1 - len2, suffix) == 0);
645 }
646
647 /* The contents of value VAL, treated as a value of type TYPE. The
648 result is an lval in memory if VAL is. */
649
650 static struct value *
651 coerce_unspec_val_to_type (struct value *val, struct type *type)
652 {
653 type = ada_check_typedef (type);
654 if (value_type (val) == type)
655 return val;
656 else
657 {
658 struct value *result;
659
660 /* Make sure that the object size is not unreasonable before
661 trying to allocate some memory for it. */
662 ada_ensure_varsize_limit (type);
663
664 if (value_lazy (val)
665 || TYPE_LENGTH (type) > TYPE_LENGTH (value_type (val)))
666 result = allocate_value_lazy (type);
667 else
668 {
669 result = allocate_value (type);
670 value_contents_copy_raw (result, 0, val, 0, TYPE_LENGTH (type));
671 }
672 set_value_component_location (result, val);
673 set_value_bitsize (result, value_bitsize (val));
674 set_value_bitpos (result, value_bitpos (val));
675 set_value_address (result, value_address (val));
676 return result;
677 }
678 }
679
680 static const gdb_byte *
681 cond_offset_host (const gdb_byte *valaddr, long offset)
682 {
683 if (valaddr == NULL)
684 return NULL;
685 else
686 return valaddr + offset;
687 }
688
689 static CORE_ADDR
690 cond_offset_target (CORE_ADDR address, long offset)
691 {
692 if (address == 0)
693 return 0;
694 else
695 return address + offset;
696 }
697
698 /* Issue a warning (as for the definition of warning in utils.c, but
699 with exactly one argument rather than ...), unless the limit on the
700 number of warnings has passed during the evaluation of the current
701 expression. */
702
703 /* FIXME: cagney/2004-10-10: This function is mimicking the behavior
704 provided by "complaint". */
705 static void lim_warning (const char *format, ...) ATTRIBUTE_PRINTF (1, 2);
706
707 static void
708 lim_warning (const char *format, ...)
709 {
710 va_list args;
711
712 va_start (args, format);
713 warnings_issued += 1;
714 if (warnings_issued <= warning_limit)
715 vwarning (format, args);
716
717 va_end (args);
718 }
719
720 /* Issue an error if the size of an object of type T is unreasonable,
721 i.e. if it would be a bad idea to allocate a value of this type in
722 GDB. */
723
724 void
725 ada_ensure_varsize_limit (const struct type *type)
726 {
727 if (TYPE_LENGTH (type) > varsize_limit)
728 error (_("object size is larger than varsize-limit"));
729 }
730
731 /* Maximum value of a SIZE-byte signed integer type. */
732 static LONGEST
733 max_of_size (int size)
734 {
735 LONGEST top_bit = (LONGEST) 1 << (size * 8 - 2);
736
737 return top_bit | (top_bit - 1);
738 }
739
740 /* Minimum value of a SIZE-byte signed integer type. */
741 static LONGEST
742 min_of_size (int size)
743 {
744 return -max_of_size (size) - 1;
745 }
746
747 /* Maximum value of a SIZE-byte unsigned integer type. */
748 static ULONGEST
749 umax_of_size (int size)
750 {
751 ULONGEST top_bit = (ULONGEST) 1 << (size * 8 - 1);
752
753 return top_bit | (top_bit - 1);
754 }
755
756 /* Maximum value of integral type T, as a signed quantity. */
757 static LONGEST
758 max_of_type (struct type *t)
759 {
760 if (TYPE_UNSIGNED (t))
761 return (LONGEST) umax_of_size (TYPE_LENGTH (t));
762 else
763 return max_of_size (TYPE_LENGTH (t));
764 }
765
766 /* Minimum value of integral type T, as a signed quantity. */
767 static LONGEST
768 min_of_type (struct type *t)
769 {
770 if (TYPE_UNSIGNED (t))
771 return 0;
772 else
773 return min_of_size (TYPE_LENGTH (t));
774 }
775
776 /* The largest value in the domain of TYPE, a discrete type, as an integer. */
777 LONGEST
778 ada_discrete_type_high_bound (struct type *type)
779 {
780 type = resolve_dynamic_type (type, NULL, 0);
781 switch (TYPE_CODE (type))
782 {
783 case TYPE_CODE_RANGE:
784 return TYPE_HIGH_BOUND (type);
785 case TYPE_CODE_ENUM:
786 return TYPE_FIELD_ENUMVAL (type, TYPE_NFIELDS (type) - 1);
787 case TYPE_CODE_BOOL:
788 return 1;
789 case TYPE_CODE_CHAR:
790 case TYPE_CODE_INT:
791 return max_of_type (type);
792 default:
793 error (_("Unexpected type in ada_discrete_type_high_bound."));
794 }
795 }
796
797 /* The smallest value in the domain of TYPE, a discrete type, as an integer. */
798 LONGEST
799 ada_discrete_type_low_bound (struct type *type)
800 {
801 type = resolve_dynamic_type (type, NULL, 0);
802 switch (TYPE_CODE (type))
803 {
804 case TYPE_CODE_RANGE:
805 return TYPE_LOW_BOUND (type);
806 case TYPE_CODE_ENUM:
807 return TYPE_FIELD_ENUMVAL (type, 0);
808 case TYPE_CODE_BOOL:
809 return 0;
810 case TYPE_CODE_CHAR:
811 case TYPE_CODE_INT:
812 return min_of_type (type);
813 default:
814 error (_("Unexpected type in ada_discrete_type_low_bound."));
815 }
816 }
817
818 /* The identity on non-range types. For range types, the underlying
819 non-range scalar type. */
820
821 static struct type *
822 get_base_type (struct type *type)
823 {
824 while (type != NULL && TYPE_CODE (type) == TYPE_CODE_RANGE)
825 {
826 if (type == TYPE_TARGET_TYPE (type) || TYPE_TARGET_TYPE (type) == NULL)
827 return type;
828 type = TYPE_TARGET_TYPE (type);
829 }
830 return type;
831 }
832
833 /* Return a decoded version of the given VALUE. This means returning
834 a value whose type is obtained by applying all the GNAT-specific
835 encondings, making the resulting type a static but standard description
836 of the initial type. */
837
838 struct value *
839 ada_get_decoded_value (struct value *value)
840 {
841 struct type *type = ada_check_typedef (value_type (value));
842
843 if (ada_is_array_descriptor_type (type)
844 || (ada_is_constrained_packed_array_type (type)
845 && TYPE_CODE (type) != TYPE_CODE_PTR))
846 {
847 if (TYPE_CODE (type) == TYPE_CODE_TYPEDEF) /* array access type. */
848 value = ada_coerce_to_simple_array_ptr (value);
849 else
850 value = ada_coerce_to_simple_array (value);
851 }
852 else
853 value = ada_to_fixed_value (value);
854
855 return value;
856 }
857
858 /* Same as ada_get_decoded_value, but with the given TYPE.
859 Because there is no associated actual value for this type,
860 the resulting type might be a best-effort approximation in
861 the case of dynamic types. */
862
863 struct type *
864 ada_get_decoded_type (struct type *type)
865 {
866 type = to_static_fixed_type (type);
867 if (ada_is_constrained_packed_array_type (type))
868 type = ada_coerce_to_simple_array_type (type);
869 return type;
870 }
871
872 \f
873
874 /* Language Selection */
875
876 /* If the main program is in Ada, return language_ada, otherwise return LANG
877 (the main program is in Ada iif the adainit symbol is found). */
878
879 enum language
880 ada_update_initial_language (enum language lang)
881 {
882 if (lookup_minimal_symbol ("adainit", (const char *) NULL,
883 (struct objfile *) NULL).minsym != NULL)
884 return language_ada;
885
886 return lang;
887 }
888
889 /* If the main procedure is written in Ada, then return its name.
890 The result is good until the next call. Return NULL if the main
891 procedure doesn't appear to be in Ada. */
892
893 char *
894 ada_main_name (void)
895 {
896 struct bound_minimal_symbol msym;
897 static gdb::unique_xmalloc_ptr<char> main_program_name;
898
899 /* For Ada, the name of the main procedure is stored in a specific
900 string constant, generated by the binder. Look for that symbol,
901 extract its address, and then read that string. If we didn't find
902 that string, then most probably the main procedure is not written
903 in Ada. */
904 msym = lookup_minimal_symbol (ADA_MAIN_PROGRAM_SYMBOL_NAME, NULL, NULL);
905
906 if (msym.minsym != NULL)
907 {
908 CORE_ADDR main_program_name_addr;
909 int err_code;
910
911 main_program_name_addr = BMSYMBOL_VALUE_ADDRESS (msym);
912 if (main_program_name_addr == 0)
913 error (_("Invalid address for Ada main program name."));
914
915 target_read_string (main_program_name_addr, &main_program_name,
916 1024, &err_code);
917
918 if (err_code != 0)
919 return NULL;
920 return main_program_name.get ();
921 }
922
923 /* The main procedure doesn't seem to be in Ada. */
924 return NULL;
925 }
926 \f
927 /* Symbols */
928
929 /* Table of Ada operators and their GNAT-encoded names. Last entry is pair
930 of NULLs. */
931
932 const struct ada_opname_map ada_opname_table[] = {
933 {"Oadd", "\"+\"", BINOP_ADD},
934 {"Osubtract", "\"-\"", BINOP_SUB},
935 {"Omultiply", "\"*\"", BINOP_MUL},
936 {"Odivide", "\"/\"", BINOP_DIV},
937 {"Omod", "\"mod\"", BINOP_MOD},
938 {"Orem", "\"rem\"", BINOP_REM},
939 {"Oexpon", "\"**\"", BINOP_EXP},
940 {"Olt", "\"<\"", BINOP_LESS},
941 {"Ole", "\"<=\"", BINOP_LEQ},
942 {"Ogt", "\">\"", BINOP_GTR},
943 {"Oge", "\">=\"", BINOP_GEQ},
944 {"Oeq", "\"=\"", BINOP_EQUAL},
945 {"One", "\"/=\"", BINOP_NOTEQUAL},
946 {"Oand", "\"and\"", BINOP_BITWISE_AND},
947 {"Oor", "\"or\"", BINOP_BITWISE_IOR},
948 {"Oxor", "\"xor\"", BINOP_BITWISE_XOR},
949 {"Oconcat", "\"&\"", BINOP_CONCAT},
950 {"Oabs", "\"abs\"", UNOP_ABS},
951 {"Onot", "\"not\"", UNOP_LOGICAL_NOT},
952 {"Oadd", "\"+\"", UNOP_PLUS},
953 {"Osubtract", "\"-\"", UNOP_NEG},
954 {NULL, NULL}
955 };
956
957 /* The "encoded" form of DECODED, according to GNAT conventions. The
958 result is valid until the next call to ada_encode. If
959 THROW_ERRORS, throw an error if invalid operator name is found.
960 Otherwise, return NULL in that case. */
961
962 static char *
963 ada_encode_1 (const char *decoded, bool throw_errors)
964 {
965 static char *encoding_buffer = NULL;
966 static size_t encoding_buffer_size = 0;
967 const char *p;
968 int k;
969
970 if (decoded == NULL)
971 return NULL;
972
973 GROW_VECT (encoding_buffer, encoding_buffer_size,
974 2 * strlen (decoded) + 10);
975
976 k = 0;
977 for (p = decoded; *p != '\0'; p += 1)
978 {
979 if (*p == '.')
980 {
981 encoding_buffer[k] = encoding_buffer[k + 1] = '_';
982 k += 2;
983 }
984 else if (*p == '"')
985 {
986 const struct ada_opname_map *mapping;
987
988 for (mapping = ada_opname_table;
989 mapping->encoded != NULL
990 && !startswith (p, mapping->decoded); mapping += 1)
991 ;
992 if (mapping->encoded == NULL)
993 {
994 if (throw_errors)
995 error (_("invalid Ada operator name: %s"), p);
996 else
997 return NULL;
998 }
999 strcpy (encoding_buffer + k, mapping->encoded);
1000 k += strlen (mapping->encoded);
1001 break;
1002 }
1003 else
1004 {
1005 encoding_buffer[k] = *p;
1006 k += 1;
1007 }
1008 }
1009
1010 encoding_buffer[k] = '\0';
1011 return encoding_buffer;
1012 }
1013
1014 /* The "encoded" form of DECODED, according to GNAT conventions.
1015 The result is valid until the next call to ada_encode. */
1016
1017 char *
1018 ada_encode (const char *decoded)
1019 {
1020 return ada_encode_1 (decoded, true);
1021 }
1022
1023 /* Return NAME folded to lower case, or, if surrounded by single
1024 quotes, unfolded, but with the quotes stripped away. Result good
1025 to next call. */
1026
1027 char *
1028 ada_fold_name (const char *name)
1029 {
1030 static char *fold_buffer = NULL;
1031 static size_t fold_buffer_size = 0;
1032
1033 int len = strlen (name);
1034 GROW_VECT (fold_buffer, fold_buffer_size, len + 1);
1035
1036 if (name[0] == '\'')
1037 {
1038 strncpy (fold_buffer, name + 1, len - 2);
1039 fold_buffer[len - 2] = '\000';
1040 }
1041 else
1042 {
1043 int i;
1044
1045 for (i = 0; i <= len; i += 1)
1046 fold_buffer[i] = tolower (name[i]);
1047 }
1048
1049 return fold_buffer;
1050 }
1051
1052 /* Return nonzero if C is either a digit or a lowercase alphabet character. */
1053
1054 static int
1055 is_lower_alphanum (const char c)
1056 {
1057 return (isdigit (c) || (isalpha (c) && islower (c)));
1058 }
1059
1060 /* ENCODED is the linkage name of a symbol and LEN contains its length.
1061 This function saves in LEN the length of that same symbol name but
1062 without either of these suffixes:
1063 . .{DIGIT}+
1064 . ${DIGIT}+
1065 . ___{DIGIT}+
1066 . __{DIGIT}+.
1067
1068 These are suffixes introduced by the compiler for entities such as
1069 nested subprogram for instance, in order to avoid name clashes.
1070 They do not serve any purpose for the debugger. */
1071
1072 static void
1073 ada_remove_trailing_digits (const char *encoded, int *len)
1074 {
1075 if (*len > 1 && isdigit (encoded[*len - 1]))
1076 {
1077 int i = *len - 2;
1078
1079 while (i > 0 && isdigit (encoded[i]))
1080 i--;
1081 if (i >= 0 && encoded[i] == '.')
1082 *len = i;
1083 else if (i >= 0 && encoded[i] == '$')
1084 *len = i;
1085 else if (i >= 2 && startswith (encoded + i - 2, "___"))
1086 *len = i - 2;
1087 else if (i >= 1 && startswith (encoded + i - 1, "__"))
1088 *len = i - 1;
1089 }
1090 }
1091
1092 /* Remove the suffix introduced by the compiler for protected object
1093 subprograms. */
1094
1095 static void
1096 ada_remove_po_subprogram_suffix (const char *encoded, int *len)
1097 {
1098 /* Remove trailing N. */
1099
1100 /* Protected entry subprograms are broken into two
1101 separate subprograms: The first one is unprotected, and has
1102 a 'N' suffix; the second is the protected version, and has
1103 the 'P' suffix. The second calls the first one after handling
1104 the protection. Since the P subprograms are internally generated,
1105 we leave these names undecoded, giving the user a clue that this
1106 entity is internal. */
1107
1108 if (*len > 1
1109 && encoded[*len - 1] == 'N'
1110 && (isdigit (encoded[*len - 2]) || islower (encoded[*len - 2])))
1111 *len = *len - 1;
1112 }
1113
1114 /* Remove trailing X[bn]* suffixes (indicating names in package bodies). */
1115
1116 static void
1117 ada_remove_Xbn_suffix (const char *encoded, int *len)
1118 {
1119 int i = *len - 1;
1120
1121 while (i > 0 && (encoded[i] == 'b' || encoded[i] == 'n'))
1122 i--;
1123
1124 if (encoded[i] != 'X')
1125 return;
1126
1127 if (i == 0)
1128 return;
1129
1130 if (isalnum (encoded[i-1]))
1131 *len = i;
1132 }
1133
1134 /* If ENCODED follows the GNAT entity encoding conventions, then return
1135 the decoded form of ENCODED. Otherwise, return "<%s>" where "%s" is
1136 replaced by ENCODED.
1137
1138 The resulting string is valid until the next call of ada_decode.
1139 If the string is unchanged by decoding, the original string pointer
1140 is returned. */
1141
1142 const char *
1143 ada_decode (const char *encoded)
1144 {
1145 int i, j;
1146 int len0;
1147 const char *p;
1148 char *decoded;
1149 int at_start_name;
1150 static char *decoding_buffer = NULL;
1151 static size_t decoding_buffer_size = 0;
1152
1153 /* With function descriptors on PPC64, the value of a symbol named
1154 ".FN", if it exists, is the entry point of the function "FN". */
1155 if (encoded[0] == '.')
1156 encoded += 1;
1157
1158 /* The name of the Ada main procedure starts with "_ada_".
1159 This prefix is not part of the decoded name, so skip this part
1160 if we see this prefix. */
1161 if (startswith (encoded, "_ada_"))
1162 encoded += 5;
1163
1164 /* If the name starts with '_', then it is not a properly encoded
1165 name, so do not attempt to decode it. Similarly, if the name
1166 starts with '<', the name should not be decoded. */
1167 if (encoded[0] == '_' || encoded[0] == '<')
1168 goto Suppress;
1169
1170 len0 = strlen (encoded);
1171
1172 ada_remove_trailing_digits (encoded, &len0);
1173 ada_remove_po_subprogram_suffix (encoded, &len0);
1174
1175 /* Remove the ___X.* suffix if present. Do not forget to verify that
1176 the suffix is located before the current "end" of ENCODED. We want
1177 to avoid re-matching parts of ENCODED that have previously been
1178 marked as discarded (by decrementing LEN0). */
1179 p = strstr (encoded, "___");
1180 if (p != NULL && p - encoded < len0 - 3)
1181 {
1182 if (p[3] == 'X')
1183 len0 = p - encoded;
1184 else
1185 goto Suppress;
1186 }
1187
1188 /* Remove any trailing TKB suffix. It tells us that this symbol
1189 is for the body of a task, but that information does not actually
1190 appear in the decoded name. */
1191
1192 if (len0 > 3 && startswith (encoded + len0 - 3, "TKB"))
1193 len0 -= 3;
1194
1195 /* Remove any trailing TB suffix. The TB suffix is slightly different
1196 from the TKB suffix because it is used for non-anonymous task
1197 bodies. */
1198
1199 if (len0 > 2 && startswith (encoded + len0 - 2, "TB"))
1200 len0 -= 2;
1201
1202 /* Remove trailing "B" suffixes. */
1203 /* FIXME: brobecker/2006-04-19: Not sure what this are used for... */
1204
1205 if (len0 > 1 && startswith (encoded + len0 - 1, "B"))
1206 len0 -= 1;
1207
1208 /* Make decoded big enough for possible expansion by operator name. */
1209
1210 GROW_VECT (decoding_buffer, decoding_buffer_size, 2 * len0 + 1);
1211 decoded = decoding_buffer;
1212
1213 /* Remove trailing __{digit}+ or trailing ${digit}+. */
1214
1215 if (len0 > 1 && isdigit (encoded[len0 - 1]))
1216 {
1217 i = len0 - 2;
1218 while ((i >= 0 && isdigit (encoded[i]))
1219 || (i >= 1 && encoded[i] == '_' && isdigit (encoded[i - 1])))
1220 i -= 1;
1221 if (i > 1 && encoded[i] == '_' && encoded[i - 1] == '_')
1222 len0 = i - 1;
1223 else if (encoded[i] == '$')
1224 len0 = i;
1225 }
1226
1227 /* The first few characters that are not alphabetic are not part
1228 of any encoding we use, so we can copy them over verbatim. */
1229
1230 for (i = 0, j = 0; i < len0 && !isalpha (encoded[i]); i += 1, j += 1)
1231 decoded[j] = encoded[i];
1232
1233 at_start_name = 1;
1234 while (i < len0)
1235 {
1236 /* Is this a symbol function? */
1237 if (at_start_name && encoded[i] == 'O')
1238 {
1239 int k;
1240
1241 for (k = 0; ada_opname_table[k].encoded != NULL; k += 1)
1242 {
1243 int op_len = strlen (ada_opname_table[k].encoded);
1244 if ((strncmp (ada_opname_table[k].encoded + 1, encoded + i + 1,
1245 op_len - 1) == 0)
1246 && !isalnum (encoded[i + op_len]))
1247 {
1248 strcpy (decoded + j, ada_opname_table[k].decoded);
1249 at_start_name = 0;
1250 i += op_len;
1251 j += strlen (ada_opname_table[k].decoded);
1252 break;
1253 }
1254 }
1255 if (ada_opname_table[k].encoded != NULL)
1256 continue;
1257 }
1258 at_start_name = 0;
1259
1260 /* Replace "TK__" with "__", which will eventually be translated
1261 into "." (just below). */
1262
1263 if (i < len0 - 4 && startswith (encoded + i, "TK__"))
1264 i += 2;
1265
1266 /* Replace "__B_{DIGITS}+__" sequences by "__", which will eventually
1267 be translated into "." (just below). These are internal names
1268 generated for anonymous blocks inside which our symbol is nested. */
1269
1270 if (len0 - i > 5 && encoded [i] == '_' && encoded [i+1] == '_'
1271 && encoded [i+2] == 'B' && encoded [i+3] == '_'
1272 && isdigit (encoded [i+4]))
1273 {
1274 int k = i + 5;
1275
1276 while (k < len0 && isdigit (encoded[k]))
1277 k++; /* Skip any extra digit. */
1278
1279 /* Double-check that the "__B_{DIGITS}+" sequence we found
1280 is indeed followed by "__". */
1281 if (len0 - k > 2 && encoded [k] == '_' && encoded [k+1] == '_')
1282 i = k;
1283 }
1284
1285 /* Remove _E{DIGITS}+[sb] */
1286
1287 /* Just as for protected object subprograms, there are 2 categories
1288 of subprograms created by the compiler for each entry. The first
1289 one implements the actual entry code, and has a suffix following
1290 the convention above; the second one implements the barrier and
1291 uses the same convention as above, except that the 'E' is replaced
1292 by a 'B'.
1293
1294 Just as above, we do not decode the name of barrier functions
1295 to give the user a clue that the code he is debugging has been
1296 internally generated. */
1297
1298 if (len0 - i > 3 && encoded [i] == '_' && encoded[i+1] == 'E'
1299 && isdigit (encoded[i+2]))
1300 {
1301 int k = i + 3;
1302
1303 while (k < len0 && isdigit (encoded[k]))
1304 k++;
1305
1306 if (k < len0
1307 && (encoded[k] == 'b' || encoded[k] == 's'))
1308 {
1309 k++;
1310 /* Just as an extra precaution, make sure that if this
1311 suffix is followed by anything else, it is a '_'.
1312 Otherwise, we matched this sequence by accident. */
1313 if (k == len0
1314 || (k < len0 && encoded[k] == '_'))
1315 i = k;
1316 }
1317 }
1318
1319 /* Remove trailing "N" in [a-z0-9]+N__. The N is added by
1320 the GNAT front-end in protected object subprograms. */
1321
1322 if (i < len0 + 3
1323 && encoded[i] == 'N' && encoded[i+1] == '_' && encoded[i+2] == '_')
1324 {
1325 /* Backtrack a bit up until we reach either the begining of
1326 the encoded name, or "__". Make sure that we only find
1327 digits or lowercase characters. */
1328 const char *ptr = encoded + i - 1;
1329
1330 while (ptr >= encoded && is_lower_alphanum (ptr[0]))
1331 ptr--;
1332 if (ptr < encoded
1333 || (ptr > encoded && ptr[0] == '_' && ptr[-1] == '_'))
1334 i++;
1335 }
1336
1337 if (encoded[i] == 'X' && i != 0 && isalnum (encoded[i - 1]))
1338 {
1339 /* This is a X[bn]* sequence not separated from the previous
1340 part of the name with a non-alpha-numeric character (in other
1341 words, immediately following an alpha-numeric character), then
1342 verify that it is placed at the end of the encoded name. If
1343 not, then the encoding is not valid and we should abort the
1344 decoding. Otherwise, just skip it, it is used in body-nested
1345 package names. */
1346 do
1347 i += 1;
1348 while (i < len0 && (encoded[i] == 'b' || encoded[i] == 'n'));
1349 if (i < len0)
1350 goto Suppress;
1351 }
1352 else if (i < len0 - 2 && encoded[i] == '_' && encoded[i + 1] == '_')
1353 {
1354 /* Replace '__' by '.'. */
1355 decoded[j] = '.';
1356 at_start_name = 1;
1357 i += 2;
1358 j += 1;
1359 }
1360 else
1361 {
1362 /* It's a character part of the decoded name, so just copy it
1363 over. */
1364 decoded[j] = encoded[i];
1365 i += 1;
1366 j += 1;
1367 }
1368 }
1369 decoded[j] = '\000';
1370
1371 /* Decoded names should never contain any uppercase character.
1372 Double-check this, and abort the decoding if we find one. */
1373
1374 for (i = 0; decoded[i] != '\0'; i += 1)
1375 if (isupper (decoded[i]) || decoded[i] == ' ')
1376 goto Suppress;
1377
1378 if (strcmp (decoded, encoded) == 0)
1379 return encoded;
1380 else
1381 return decoded;
1382
1383 Suppress:
1384 GROW_VECT (decoding_buffer, decoding_buffer_size, strlen (encoded) + 3);
1385 decoded = decoding_buffer;
1386 if (encoded[0] == '<')
1387 strcpy (decoded, encoded);
1388 else
1389 xsnprintf (decoded, decoding_buffer_size, "<%s>", encoded);
1390 return decoded;
1391
1392 }
1393
1394 /* Table for keeping permanent unique copies of decoded names. Once
1395 allocated, names in this table are never released. While this is a
1396 storage leak, it should not be significant unless there are massive
1397 changes in the set of decoded names in successive versions of a
1398 symbol table loaded during a single session. */
1399 static struct htab *decoded_names_store;
1400
1401 /* Returns the decoded name of GSYMBOL, as for ada_decode, caching it
1402 in the language-specific part of GSYMBOL, if it has not been
1403 previously computed. Tries to save the decoded name in the same
1404 obstack as GSYMBOL, if possible, and otherwise on the heap (so that,
1405 in any case, the decoded symbol has a lifetime at least that of
1406 GSYMBOL).
1407 The GSYMBOL parameter is "mutable" in the C++ sense: logically
1408 const, but nevertheless modified to a semantically equivalent form
1409 when a decoded name is cached in it. */
1410
1411 const char *
1412 ada_decode_symbol (const struct general_symbol_info *arg)
1413 {
1414 struct general_symbol_info *gsymbol = (struct general_symbol_info *) arg;
1415 const char **resultp =
1416 &gsymbol->language_specific.demangled_name;
1417
1418 if (!gsymbol->ada_mangled)
1419 {
1420 const char *decoded = ada_decode (gsymbol->name);
1421 struct obstack *obstack = gsymbol->language_specific.obstack;
1422
1423 gsymbol->ada_mangled = 1;
1424
1425 if (obstack != NULL)
1426 *resultp
1427 = (const char *) obstack_copy0 (obstack, decoded, strlen (decoded));
1428 else
1429 {
1430 /* Sometimes, we can't find a corresponding objfile, in
1431 which case, we put the result on the heap. Since we only
1432 decode when needed, we hope this usually does not cause a
1433 significant memory leak (FIXME). */
1434
1435 char **slot = (char **) htab_find_slot (decoded_names_store,
1436 decoded, INSERT);
1437
1438 if (*slot == NULL)
1439 *slot = xstrdup (decoded);
1440 *resultp = *slot;
1441 }
1442 }
1443
1444 return *resultp;
1445 }
1446
1447 static char *
1448 ada_la_decode (const char *encoded, int options)
1449 {
1450 return xstrdup (ada_decode (encoded));
1451 }
1452
1453 /* Implement la_sniff_from_mangled_name for Ada. */
1454
1455 static int
1456 ada_sniff_from_mangled_name (const char *mangled, char **out)
1457 {
1458 const char *demangled = ada_decode (mangled);
1459
1460 *out = NULL;
1461
1462 if (demangled != mangled && demangled != NULL && demangled[0] != '<')
1463 {
1464 /* Set the gsymbol language to Ada, but still return 0.
1465 Two reasons for that:
1466
1467 1. For Ada, we prefer computing the symbol's decoded name
1468 on the fly rather than pre-compute it, in order to save
1469 memory (Ada projects are typically very large).
1470
1471 2. There are some areas in the definition of the GNAT
1472 encoding where, with a bit of bad luck, we might be able
1473 to decode a non-Ada symbol, generating an incorrect
1474 demangled name (Eg: names ending with "TB" for instance
1475 are identified as task bodies and so stripped from
1476 the decoded name returned).
1477
1478 Returning 1, here, but not setting *DEMANGLED, helps us get a
1479 little bit of the best of both worlds. Because we're last,
1480 we should not affect any of the other languages that were
1481 able to demangle the symbol before us; we get to correctly
1482 tag Ada symbols as such; and even if we incorrectly tagged a
1483 non-Ada symbol, which should be rare, any routing through the
1484 Ada language should be transparent (Ada tries to behave much
1485 like C/C++ with non-Ada symbols). */
1486 return 1;
1487 }
1488
1489 return 0;
1490 }
1491
1492 \f
1493
1494 /* Arrays */
1495
1496 /* Assuming that INDEX_DESC_TYPE is an ___XA structure, a structure
1497 generated by the GNAT compiler to describe the index type used
1498 for each dimension of an array, check whether it follows the latest
1499 known encoding. If not, fix it up to conform to the latest encoding.
1500 Otherwise, do nothing. This function also does nothing if
1501 INDEX_DESC_TYPE is NULL.
1502
1503 The GNAT encoding used to describle the array index type evolved a bit.
1504 Initially, the information would be provided through the name of each
1505 field of the structure type only, while the type of these fields was
1506 described as unspecified and irrelevant. The debugger was then expected
1507 to perform a global type lookup using the name of that field in order
1508 to get access to the full index type description. Because these global
1509 lookups can be very expensive, the encoding was later enhanced to make
1510 the global lookup unnecessary by defining the field type as being
1511 the full index type description.
1512
1513 The purpose of this routine is to allow us to support older versions
1514 of the compiler by detecting the use of the older encoding, and by
1515 fixing up the INDEX_DESC_TYPE to follow the new one (at this point,
1516 we essentially replace each field's meaningless type by the associated
1517 index subtype). */
1518
1519 void
1520 ada_fixup_array_indexes_type (struct type *index_desc_type)
1521 {
1522 int i;
1523
1524 if (index_desc_type == NULL)
1525 return;
1526 gdb_assert (TYPE_NFIELDS (index_desc_type) > 0);
1527
1528 /* Check if INDEX_DESC_TYPE follows the older encoding (it is sufficient
1529 to check one field only, no need to check them all). If not, return
1530 now.
1531
1532 If our INDEX_DESC_TYPE was generated using the older encoding,
1533 the field type should be a meaningless integer type whose name
1534 is not equal to the field name. */
1535 if (TYPE_NAME (TYPE_FIELD_TYPE (index_desc_type, 0)) != NULL
1536 && strcmp (TYPE_NAME (TYPE_FIELD_TYPE (index_desc_type, 0)),
1537 TYPE_FIELD_NAME (index_desc_type, 0)) == 0)
1538 return;
1539
1540 /* Fixup each field of INDEX_DESC_TYPE. */
1541 for (i = 0; i < TYPE_NFIELDS (index_desc_type); i++)
1542 {
1543 const char *name = TYPE_FIELD_NAME (index_desc_type, i);
1544 struct type *raw_type = ada_check_typedef (ada_find_any_type (name));
1545
1546 if (raw_type)
1547 TYPE_FIELD_TYPE (index_desc_type, i) = raw_type;
1548 }
1549 }
1550
1551 /* Names of MAX_ADA_DIMENS bounds in P_BOUNDS fields of array descriptors. */
1552
1553 static const char *bound_name[] = {
1554 "LB0", "UB0", "LB1", "UB1", "LB2", "UB2", "LB3", "UB3",
1555 "LB4", "UB4", "LB5", "UB5", "LB6", "UB6", "LB7", "UB7"
1556 };
1557
1558 /* Maximum number of array dimensions we are prepared to handle. */
1559
1560 #define MAX_ADA_DIMENS (sizeof(bound_name) / (2*sizeof(char *)))
1561
1562
1563 /* The desc_* routines return primitive portions of array descriptors
1564 (fat pointers). */
1565
1566 /* The descriptor or array type, if any, indicated by TYPE; removes
1567 level of indirection, if needed. */
1568
1569 static struct type *
1570 desc_base_type (struct type *type)
1571 {
1572 if (type == NULL)
1573 return NULL;
1574 type = ada_check_typedef (type);
1575 if (TYPE_CODE (type) == TYPE_CODE_TYPEDEF)
1576 type = ada_typedef_target_type (type);
1577
1578 if (type != NULL
1579 && (TYPE_CODE (type) == TYPE_CODE_PTR
1580 || TYPE_CODE (type) == TYPE_CODE_REF))
1581 return ada_check_typedef (TYPE_TARGET_TYPE (type));
1582 else
1583 return type;
1584 }
1585
1586 /* True iff TYPE indicates a "thin" array pointer type. */
1587
1588 static int
1589 is_thin_pntr (struct type *type)
1590 {
1591 return
1592 is_suffix (ada_type_name (desc_base_type (type)), "___XUT")
1593 || is_suffix (ada_type_name (desc_base_type (type)), "___XUT___XVE");
1594 }
1595
1596 /* The descriptor type for thin pointer type TYPE. */
1597
1598 static struct type *
1599 thin_descriptor_type (struct type *type)
1600 {
1601 struct type *base_type = desc_base_type (type);
1602
1603 if (base_type == NULL)
1604 return NULL;
1605 if (is_suffix (ada_type_name (base_type), "___XVE"))
1606 return base_type;
1607 else
1608 {
1609 struct type *alt_type = ada_find_parallel_type (base_type, "___XVE");
1610
1611 if (alt_type == NULL)
1612 return base_type;
1613 else
1614 return alt_type;
1615 }
1616 }
1617
1618 /* A pointer to the array data for thin-pointer value VAL. */
1619
1620 static struct value *
1621 thin_data_pntr (struct value *val)
1622 {
1623 struct type *type = ada_check_typedef (value_type (val));
1624 struct type *data_type = desc_data_target_type (thin_descriptor_type (type));
1625
1626 data_type = lookup_pointer_type (data_type);
1627
1628 if (TYPE_CODE (type) == TYPE_CODE_PTR)
1629 return value_cast (data_type, value_copy (val));
1630 else
1631 return value_from_longest (data_type, value_address (val));
1632 }
1633
1634 /* True iff TYPE indicates a "thick" array pointer type. */
1635
1636 static int
1637 is_thick_pntr (struct type *type)
1638 {
1639 type = desc_base_type (type);
1640 return (type != NULL && TYPE_CODE (type) == TYPE_CODE_STRUCT
1641 && lookup_struct_elt_type (type, "P_BOUNDS", 1) != NULL);
1642 }
1643
1644 /* If TYPE is the type of an array descriptor (fat or thin pointer) or a
1645 pointer to one, the type of its bounds data; otherwise, NULL. */
1646
1647 static struct type *
1648 desc_bounds_type (struct type *type)
1649 {
1650 struct type *r;
1651
1652 type = desc_base_type (type);
1653
1654 if (type == NULL)
1655 return NULL;
1656 else if (is_thin_pntr (type))
1657 {
1658 type = thin_descriptor_type (type);
1659 if (type == NULL)
1660 return NULL;
1661 r = lookup_struct_elt_type (type, "BOUNDS", 1);
1662 if (r != NULL)
1663 return ada_check_typedef (r);
1664 }
1665 else if (TYPE_CODE (type) == TYPE_CODE_STRUCT)
1666 {
1667 r = lookup_struct_elt_type (type, "P_BOUNDS", 1);
1668 if (r != NULL)
1669 return ada_check_typedef (TYPE_TARGET_TYPE (ada_check_typedef (r)));
1670 }
1671 return NULL;
1672 }
1673
1674 /* If ARR is an array descriptor (fat or thin pointer), or pointer to
1675 one, a pointer to its bounds data. Otherwise NULL. */
1676
1677 static struct value *
1678 desc_bounds (struct value *arr)
1679 {
1680 struct type *type = ada_check_typedef (value_type (arr));
1681
1682 if (is_thin_pntr (type))
1683 {
1684 struct type *bounds_type =
1685 desc_bounds_type (thin_descriptor_type (type));
1686 LONGEST addr;
1687
1688 if (bounds_type == NULL)
1689 error (_("Bad GNAT array descriptor"));
1690
1691 /* NOTE: The following calculation is not really kosher, but
1692 since desc_type is an XVE-encoded type (and shouldn't be),
1693 the correct calculation is a real pain. FIXME (and fix GCC). */
1694 if (TYPE_CODE (type) == TYPE_CODE_PTR)
1695 addr = value_as_long (arr);
1696 else
1697 addr = value_address (arr);
1698
1699 return
1700 value_from_longest (lookup_pointer_type (bounds_type),
1701 addr - TYPE_LENGTH (bounds_type));
1702 }
1703
1704 else if (is_thick_pntr (type))
1705 {
1706 struct value *p_bounds = value_struct_elt (&arr, NULL, "P_BOUNDS", NULL,
1707 _("Bad GNAT array descriptor"));
1708 struct type *p_bounds_type = value_type (p_bounds);
1709
1710 if (p_bounds_type
1711 && TYPE_CODE (p_bounds_type) == TYPE_CODE_PTR)
1712 {
1713 struct type *target_type = TYPE_TARGET_TYPE (p_bounds_type);
1714
1715 if (TYPE_STUB (target_type))
1716 p_bounds = value_cast (lookup_pointer_type
1717 (ada_check_typedef (target_type)),
1718 p_bounds);
1719 }
1720 else
1721 error (_("Bad GNAT array descriptor"));
1722
1723 return p_bounds;
1724 }
1725 else
1726 return NULL;
1727 }
1728
1729 /* If TYPE is the type of an array-descriptor (fat pointer), the bit
1730 position of the field containing the address of the bounds data. */
1731
1732 static int
1733 fat_pntr_bounds_bitpos (struct type *type)
1734 {
1735 return TYPE_FIELD_BITPOS (desc_base_type (type), 1);
1736 }
1737
1738 /* If TYPE is the type of an array-descriptor (fat pointer), the bit
1739 size of the field containing the address of the bounds data. */
1740
1741 static int
1742 fat_pntr_bounds_bitsize (struct type *type)
1743 {
1744 type = desc_base_type (type);
1745
1746 if (TYPE_FIELD_BITSIZE (type, 1) > 0)
1747 return TYPE_FIELD_BITSIZE (type, 1);
1748 else
1749 return 8 * TYPE_LENGTH (ada_check_typedef (TYPE_FIELD_TYPE (type, 1)));
1750 }
1751
1752 /* If TYPE is the type of an array descriptor (fat or thin pointer) or a
1753 pointer to one, the type of its array data (a array-with-no-bounds type);
1754 otherwise, NULL. Use ada_type_of_array to get an array type with bounds
1755 data. */
1756
1757 static struct type *
1758 desc_data_target_type (struct type *type)
1759 {
1760 type = desc_base_type (type);
1761
1762 /* NOTE: The following is bogus; see comment in desc_bounds. */
1763 if (is_thin_pntr (type))
1764 return desc_base_type (TYPE_FIELD_TYPE (thin_descriptor_type (type), 1));
1765 else if (is_thick_pntr (type))
1766 {
1767 struct type *data_type = lookup_struct_elt_type (type, "P_ARRAY", 1);
1768
1769 if (data_type
1770 && TYPE_CODE (ada_check_typedef (data_type)) == TYPE_CODE_PTR)
1771 return ada_check_typedef (TYPE_TARGET_TYPE (data_type));
1772 }
1773
1774 return NULL;
1775 }
1776
1777 /* If ARR is an array descriptor (fat or thin pointer), a pointer to
1778 its array data. */
1779
1780 static struct value *
1781 desc_data (struct value *arr)
1782 {
1783 struct type *type = value_type (arr);
1784
1785 if (is_thin_pntr (type))
1786 return thin_data_pntr (arr);
1787 else if (is_thick_pntr (type))
1788 return value_struct_elt (&arr, NULL, "P_ARRAY", NULL,
1789 _("Bad GNAT array descriptor"));
1790 else
1791 return NULL;
1792 }
1793
1794
1795 /* If TYPE is the type of an array-descriptor (fat pointer), the bit
1796 position of the field containing the address of the data. */
1797
1798 static int
1799 fat_pntr_data_bitpos (struct type *type)
1800 {
1801 return TYPE_FIELD_BITPOS (desc_base_type (type), 0);
1802 }
1803
1804 /* If TYPE is the type of an array-descriptor (fat pointer), the bit
1805 size of the field containing the address of the data. */
1806
1807 static int
1808 fat_pntr_data_bitsize (struct type *type)
1809 {
1810 type = desc_base_type (type);
1811
1812 if (TYPE_FIELD_BITSIZE (type, 0) > 0)
1813 return TYPE_FIELD_BITSIZE (type, 0);
1814 else
1815 return TARGET_CHAR_BIT * TYPE_LENGTH (TYPE_FIELD_TYPE (type, 0));
1816 }
1817
1818 /* If BOUNDS is an array-bounds structure (or pointer to one), return
1819 the Ith lower bound stored in it, if WHICH is 0, and the Ith upper
1820 bound, if WHICH is 1. The first bound is I=1. */
1821
1822 static struct value *
1823 desc_one_bound (struct value *bounds, int i, int which)
1824 {
1825 return value_struct_elt (&bounds, NULL, bound_name[2 * i + which - 2], NULL,
1826 _("Bad GNAT array descriptor bounds"));
1827 }
1828
1829 /* If BOUNDS is an array-bounds structure type, return the bit position
1830 of the Ith lower bound stored in it, if WHICH is 0, and the Ith upper
1831 bound, if WHICH is 1. The first bound is I=1. */
1832
1833 static int
1834 desc_bound_bitpos (struct type *type, int i, int which)
1835 {
1836 return TYPE_FIELD_BITPOS (desc_base_type (type), 2 * i + which - 2);
1837 }
1838
1839 /* If BOUNDS is an array-bounds structure type, return the bit field size
1840 of the Ith lower bound stored in it, if WHICH is 0, and the Ith upper
1841 bound, if WHICH is 1. The first bound is I=1. */
1842
1843 static int
1844 desc_bound_bitsize (struct type *type, int i, int which)
1845 {
1846 type = desc_base_type (type);
1847
1848 if (TYPE_FIELD_BITSIZE (type, 2 * i + which - 2) > 0)
1849 return TYPE_FIELD_BITSIZE (type, 2 * i + which - 2);
1850 else
1851 return 8 * TYPE_LENGTH (TYPE_FIELD_TYPE (type, 2 * i + which - 2));
1852 }
1853
1854 /* If TYPE is the type of an array-bounds structure, the type of its
1855 Ith bound (numbering from 1). Otherwise, NULL. */
1856
1857 static struct type *
1858 desc_index_type (struct type *type, int i)
1859 {
1860 type = desc_base_type (type);
1861
1862 if (TYPE_CODE (type) == TYPE_CODE_STRUCT)
1863 return lookup_struct_elt_type (type, bound_name[2 * i - 2], 1);
1864 else
1865 return NULL;
1866 }
1867
1868 /* The number of index positions in the array-bounds type TYPE.
1869 Return 0 if TYPE is NULL. */
1870
1871 static int
1872 desc_arity (struct type *type)
1873 {
1874 type = desc_base_type (type);
1875
1876 if (type != NULL)
1877 return TYPE_NFIELDS (type) / 2;
1878 return 0;
1879 }
1880
1881 /* Non-zero iff TYPE is a simple array type (not a pointer to one) or
1882 an array descriptor type (representing an unconstrained array
1883 type). */
1884
1885 static int
1886 ada_is_direct_array_type (struct type *type)
1887 {
1888 if (type == NULL)
1889 return 0;
1890 type = ada_check_typedef (type);
1891 return (TYPE_CODE (type) == TYPE_CODE_ARRAY
1892 || ada_is_array_descriptor_type (type));
1893 }
1894
1895 /* Non-zero iff TYPE represents any kind of array in Ada, or a pointer
1896 * to one. */
1897
1898 static int
1899 ada_is_array_type (struct type *type)
1900 {
1901 while (type != NULL
1902 && (TYPE_CODE (type) == TYPE_CODE_PTR
1903 || TYPE_CODE (type) == TYPE_CODE_REF))
1904 type = TYPE_TARGET_TYPE (type);
1905 return ada_is_direct_array_type (type);
1906 }
1907
1908 /* Non-zero iff TYPE is a simple array type or pointer to one. */
1909
1910 int
1911 ada_is_simple_array_type (struct type *type)
1912 {
1913 if (type == NULL)
1914 return 0;
1915 type = ada_check_typedef (type);
1916 return (TYPE_CODE (type) == TYPE_CODE_ARRAY
1917 || (TYPE_CODE (type) == TYPE_CODE_PTR
1918 && TYPE_CODE (ada_check_typedef (TYPE_TARGET_TYPE (type)))
1919 == TYPE_CODE_ARRAY));
1920 }
1921
1922 /* Non-zero iff TYPE belongs to a GNAT array descriptor. */
1923
1924 int
1925 ada_is_array_descriptor_type (struct type *type)
1926 {
1927 struct type *data_type = desc_data_target_type (type);
1928
1929 if (type == NULL)
1930 return 0;
1931 type = ada_check_typedef (type);
1932 return (data_type != NULL
1933 && TYPE_CODE (data_type) == TYPE_CODE_ARRAY
1934 && desc_arity (desc_bounds_type (type)) > 0);
1935 }
1936
1937 /* Non-zero iff type is a partially mal-formed GNAT array
1938 descriptor. FIXME: This is to compensate for some problems with
1939 debugging output from GNAT. Re-examine periodically to see if it
1940 is still needed. */
1941
1942 int
1943 ada_is_bogus_array_descriptor (struct type *type)
1944 {
1945 return
1946 type != NULL
1947 && TYPE_CODE (type) == TYPE_CODE_STRUCT
1948 && (lookup_struct_elt_type (type, "P_BOUNDS", 1) != NULL
1949 || lookup_struct_elt_type (type, "P_ARRAY", 1) != NULL)
1950 && !ada_is_array_descriptor_type (type);
1951 }
1952
1953
1954 /* If ARR has a record type in the form of a standard GNAT array descriptor,
1955 (fat pointer) returns the type of the array data described---specifically,
1956 a pointer-to-array type. If BOUNDS is non-zero, the bounds data are filled
1957 in from the descriptor; otherwise, they are left unspecified. If
1958 the ARR denotes a null array descriptor and BOUNDS is non-zero,
1959 returns NULL. The result is simply the type of ARR if ARR is not
1960 a descriptor. */
1961 struct type *
1962 ada_type_of_array (struct value *arr, int bounds)
1963 {
1964 if (ada_is_constrained_packed_array_type (value_type (arr)))
1965 return decode_constrained_packed_array_type (value_type (arr));
1966
1967 if (!ada_is_array_descriptor_type (value_type (arr)))
1968 return value_type (arr);
1969
1970 if (!bounds)
1971 {
1972 struct type *array_type =
1973 ada_check_typedef (desc_data_target_type (value_type (arr)));
1974
1975 if (ada_is_unconstrained_packed_array_type (value_type (arr)))
1976 TYPE_FIELD_BITSIZE (array_type, 0) =
1977 decode_packed_array_bitsize (value_type (arr));
1978
1979 return array_type;
1980 }
1981 else
1982 {
1983 struct type *elt_type;
1984 int arity;
1985 struct value *descriptor;
1986
1987 elt_type = ada_array_element_type (value_type (arr), -1);
1988 arity = ada_array_arity (value_type (arr));
1989
1990 if (elt_type == NULL || arity == 0)
1991 return ada_check_typedef (value_type (arr));
1992
1993 descriptor = desc_bounds (arr);
1994 if (value_as_long (descriptor) == 0)
1995 return NULL;
1996 while (arity > 0)
1997 {
1998 struct type *range_type = alloc_type_copy (value_type (arr));
1999 struct type *array_type = alloc_type_copy (value_type (arr));
2000 struct value *low = desc_one_bound (descriptor, arity, 0);
2001 struct value *high = desc_one_bound (descriptor, arity, 1);
2002
2003 arity -= 1;
2004 create_static_range_type (range_type, value_type (low),
2005 longest_to_int (value_as_long (low)),
2006 longest_to_int (value_as_long (high)));
2007 elt_type = create_array_type (array_type, elt_type, range_type);
2008
2009 if (ada_is_unconstrained_packed_array_type (value_type (arr)))
2010 {
2011 /* We need to store the element packed bitsize, as well as
2012 recompute the array size, because it was previously
2013 computed based on the unpacked element size. */
2014 LONGEST lo = value_as_long (low);
2015 LONGEST hi = value_as_long (high);
2016
2017 TYPE_FIELD_BITSIZE (elt_type, 0) =
2018 decode_packed_array_bitsize (value_type (arr));
2019 /* If the array has no element, then the size is already
2020 zero, and does not need to be recomputed. */
2021 if (lo < hi)
2022 {
2023 int array_bitsize =
2024 (hi - lo + 1) * TYPE_FIELD_BITSIZE (elt_type, 0);
2025
2026 TYPE_LENGTH (array_type) = (array_bitsize + 7) / 8;
2027 }
2028 }
2029 }
2030
2031 return lookup_pointer_type (elt_type);
2032 }
2033 }
2034
2035 /* If ARR does not represent an array, returns ARR unchanged.
2036 Otherwise, returns either a standard GDB array with bounds set
2037 appropriately or, if ARR is a non-null fat pointer, a pointer to a standard
2038 GDB array. Returns NULL if ARR is a null fat pointer. */
2039
2040 struct value *
2041 ada_coerce_to_simple_array_ptr (struct value *arr)
2042 {
2043 if (ada_is_array_descriptor_type (value_type (arr)))
2044 {
2045 struct type *arrType = ada_type_of_array (arr, 1);
2046
2047 if (arrType == NULL)
2048 return NULL;
2049 return value_cast (arrType, value_copy (desc_data (arr)));
2050 }
2051 else if (ada_is_constrained_packed_array_type (value_type (arr)))
2052 return decode_constrained_packed_array (arr);
2053 else
2054 return arr;
2055 }
2056
2057 /* If ARR does not represent an array, returns ARR unchanged.
2058 Otherwise, returns a standard GDB array describing ARR (which may
2059 be ARR itself if it already is in the proper form). */
2060
2061 struct value *
2062 ada_coerce_to_simple_array (struct value *arr)
2063 {
2064 if (ada_is_array_descriptor_type (value_type (arr)))
2065 {
2066 struct value *arrVal = ada_coerce_to_simple_array_ptr (arr);
2067
2068 if (arrVal == NULL)
2069 error (_("Bounds unavailable for null array pointer."));
2070 ada_ensure_varsize_limit (TYPE_TARGET_TYPE (value_type (arrVal)));
2071 return value_ind (arrVal);
2072 }
2073 else if (ada_is_constrained_packed_array_type (value_type (arr)))
2074 return decode_constrained_packed_array (arr);
2075 else
2076 return arr;
2077 }
2078
2079 /* If TYPE represents a GNAT array type, return it translated to an
2080 ordinary GDB array type (possibly with BITSIZE fields indicating
2081 packing). For other types, is the identity. */
2082
2083 struct type *
2084 ada_coerce_to_simple_array_type (struct type *type)
2085 {
2086 if (ada_is_constrained_packed_array_type (type))
2087 return decode_constrained_packed_array_type (type);
2088
2089 if (ada_is_array_descriptor_type (type))
2090 return ada_check_typedef (desc_data_target_type (type));
2091
2092 return type;
2093 }
2094
2095 /* Non-zero iff TYPE represents a standard GNAT packed-array type. */
2096
2097 static int
2098 ada_is_packed_array_type (struct type *type)
2099 {
2100 if (type == NULL)
2101 return 0;
2102 type = desc_base_type (type);
2103 type = ada_check_typedef (type);
2104 return
2105 ada_type_name (type) != NULL
2106 && strstr (ada_type_name (type), "___XP") != NULL;
2107 }
2108
2109 /* Non-zero iff TYPE represents a standard GNAT constrained
2110 packed-array type. */
2111
2112 int
2113 ada_is_constrained_packed_array_type (struct type *type)
2114 {
2115 return ada_is_packed_array_type (type)
2116 && !ada_is_array_descriptor_type (type);
2117 }
2118
2119 /* Non-zero iff TYPE represents an array descriptor for a
2120 unconstrained packed-array type. */
2121
2122 static int
2123 ada_is_unconstrained_packed_array_type (struct type *type)
2124 {
2125 return ada_is_packed_array_type (type)
2126 && ada_is_array_descriptor_type (type);
2127 }
2128
2129 /* Given that TYPE encodes a packed array type (constrained or unconstrained),
2130 return the size of its elements in bits. */
2131
2132 static long
2133 decode_packed_array_bitsize (struct type *type)
2134 {
2135 const char *raw_name;
2136 const char *tail;
2137 long bits;
2138
2139 /* Access to arrays implemented as fat pointers are encoded as a typedef
2140 of the fat pointer type. We need the name of the fat pointer type
2141 to do the decoding, so strip the typedef layer. */
2142 if (TYPE_CODE (type) == TYPE_CODE_TYPEDEF)
2143 type = ada_typedef_target_type (type);
2144
2145 raw_name = ada_type_name (ada_check_typedef (type));
2146 if (!raw_name)
2147 raw_name = ada_type_name (desc_base_type (type));
2148
2149 if (!raw_name)
2150 return 0;
2151
2152 tail = strstr (raw_name, "___XP");
2153 gdb_assert (tail != NULL);
2154
2155 if (sscanf (tail + sizeof ("___XP") - 1, "%ld", &bits) != 1)
2156 {
2157 lim_warning
2158 (_("could not understand bit size information on packed array"));
2159 return 0;
2160 }
2161
2162 return bits;
2163 }
2164
2165 /* Given that TYPE is a standard GDB array type with all bounds filled
2166 in, and that the element size of its ultimate scalar constituents
2167 (that is, either its elements, or, if it is an array of arrays, its
2168 elements' elements, etc.) is *ELT_BITS, return an identical type,
2169 but with the bit sizes of its elements (and those of any
2170 constituent arrays) recorded in the BITSIZE components of its
2171 TYPE_FIELD_BITSIZE values, and with *ELT_BITS set to its total size
2172 in bits.
2173
2174 Note that, for arrays whose index type has an XA encoding where
2175 a bound references a record discriminant, getting that discriminant,
2176 and therefore the actual value of that bound, is not possible
2177 because none of the given parameters gives us access to the record.
2178 This function assumes that it is OK in the context where it is being
2179 used to return an array whose bounds are still dynamic and where
2180 the length is arbitrary. */
2181
2182 static struct type *
2183 constrained_packed_array_type (struct type *type, long *elt_bits)
2184 {
2185 struct type *new_elt_type;
2186 struct type *new_type;
2187 struct type *index_type_desc;
2188 struct type *index_type;
2189 LONGEST low_bound, high_bound;
2190
2191 type = ada_check_typedef (type);
2192 if (TYPE_CODE (type) != TYPE_CODE_ARRAY)
2193 return type;
2194
2195 index_type_desc = ada_find_parallel_type (type, "___XA");
2196 if (index_type_desc)
2197 index_type = to_fixed_range_type (TYPE_FIELD_TYPE (index_type_desc, 0),
2198 NULL);
2199 else
2200 index_type = TYPE_INDEX_TYPE (type);
2201
2202 new_type = alloc_type_copy (type);
2203 new_elt_type =
2204 constrained_packed_array_type (ada_check_typedef (TYPE_TARGET_TYPE (type)),
2205 elt_bits);
2206 create_array_type (new_type, new_elt_type, index_type);
2207 TYPE_FIELD_BITSIZE (new_type, 0) = *elt_bits;
2208 TYPE_NAME (new_type) = ada_type_name (type);
2209
2210 if ((TYPE_CODE (check_typedef (index_type)) == TYPE_CODE_RANGE
2211 && is_dynamic_type (check_typedef (index_type)))
2212 || get_discrete_bounds (index_type, &low_bound, &high_bound) < 0)
2213 low_bound = high_bound = 0;
2214 if (high_bound < low_bound)
2215 *elt_bits = TYPE_LENGTH (new_type) = 0;
2216 else
2217 {
2218 *elt_bits *= (high_bound - low_bound + 1);
2219 TYPE_LENGTH (new_type) =
2220 (*elt_bits + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT;
2221 }
2222
2223 TYPE_FIXED_INSTANCE (new_type) = 1;
2224 return new_type;
2225 }
2226
2227 /* The array type encoded by TYPE, where
2228 ada_is_constrained_packed_array_type (TYPE). */
2229
2230 static struct type *
2231 decode_constrained_packed_array_type (struct type *type)
2232 {
2233 const char *raw_name = ada_type_name (ada_check_typedef (type));
2234 char *name;
2235 const char *tail;
2236 struct type *shadow_type;
2237 long bits;
2238
2239 if (!raw_name)
2240 raw_name = ada_type_name (desc_base_type (type));
2241
2242 if (!raw_name)
2243 return NULL;
2244
2245 name = (char *) alloca (strlen (raw_name) + 1);
2246 tail = strstr (raw_name, "___XP");
2247 type = desc_base_type (type);
2248
2249 memcpy (name, raw_name, tail - raw_name);
2250 name[tail - raw_name] = '\000';
2251
2252 shadow_type = ada_find_parallel_type_with_name (type, name);
2253
2254 if (shadow_type == NULL)
2255 {
2256 lim_warning (_("could not find bounds information on packed array"));
2257 return NULL;
2258 }
2259 shadow_type = check_typedef (shadow_type);
2260
2261 if (TYPE_CODE (shadow_type) != TYPE_CODE_ARRAY)
2262 {
2263 lim_warning (_("could not understand bounds "
2264 "information on packed array"));
2265 return NULL;
2266 }
2267
2268 bits = decode_packed_array_bitsize (type);
2269 return constrained_packed_array_type (shadow_type, &bits);
2270 }
2271
2272 /* Given that ARR is a struct value *indicating a GNAT constrained packed
2273 array, returns a simple array that denotes that array. Its type is a
2274 standard GDB array type except that the BITSIZEs of the array
2275 target types are set to the number of bits in each element, and the
2276 type length is set appropriately. */
2277
2278 static struct value *
2279 decode_constrained_packed_array (struct value *arr)
2280 {
2281 struct type *type;
2282
2283 /* If our value is a pointer, then dereference it. Likewise if
2284 the value is a reference. Make sure that this operation does not
2285 cause the target type to be fixed, as this would indirectly cause
2286 this array to be decoded. The rest of the routine assumes that
2287 the array hasn't been decoded yet, so we use the basic "coerce_ref"
2288 and "value_ind" routines to perform the dereferencing, as opposed
2289 to using "ada_coerce_ref" or "ada_value_ind". */
2290 arr = coerce_ref (arr);
2291 if (TYPE_CODE (ada_check_typedef (value_type (arr))) == TYPE_CODE_PTR)
2292 arr = value_ind (arr);
2293
2294 type = decode_constrained_packed_array_type (value_type (arr));
2295 if (type == NULL)
2296 {
2297 error (_("can't unpack array"));
2298 return NULL;
2299 }
2300
2301 if (gdbarch_bits_big_endian (get_type_arch (value_type (arr)))
2302 && ada_is_modular_type (value_type (arr)))
2303 {
2304 /* This is a (right-justified) modular type representing a packed
2305 array with no wrapper. In order to interpret the value through
2306 the (left-justified) packed array type we just built, we must
2307 first left-justify it. */
2308 int bit_size, bit_pos;
2309 ULONGEST mod;
2310
2311 mod = ada_modulus (value_type (arr)) - 1;
2312 bit_size = 0;
2313 while (mod > 0)
2314 {
2315 bit_size += 1;
2316 mod >>= 1;
2317 }
2318 bit_pos = HOST_CHAR_BIT * TYPE_LENGTH (value_type (arr)) - bit_size;
2319 arr = ada_value_primitive_packed_val (arr, NULL,
2320 bit_pos / HOST_CHAR_BIT,
2321 bit_pos % HOST_CHAR_BIT,
2322 bit_size,
2323 type);
2324 }
2325
2326 return coerce_unspec_val_to_type (arr, type);
2327 }
2328
2329
2330 /* The value of the element of packed array ARR at the ARITY indices
2331 given in IND. ARR must be a simple array. */
2332
2333 static struct value *
2334 value_subscript_packed (struct value *arr, int arity, struct value **ind)
2335 {
2336 int i;
2337 int bits, elt_off, bit_off;
2338 long elt_total_bit_offset;
2339 struct type *elt_type;
2340 struct value *v;
2341
2342 bits = 0;
2343 elt_total_bit_offset = 0;
2344 elt_type = ada_check_typedef (value_type (arr));
2345 for (i = 0; i < arity; i += 1)
2346 {
2347 if (TYPE_CODE (elt_type) != TYPE_CODE_ARRAY
2348 || TYPE_FIELD_BITSIZE (elt_type, 0) == 0)
2349 error
2350 (_("attempt to do packed indexing of "
2351 "something other than a packed array"));
2352 else
2353 {
2354 struct type *range_type = TYPE_INDEX_TYPE (elt_type);
2355 LONGEST lowerbound, upperbound;
2356 LONGEST idx;
2357
2358 if (get_discrete_bounds (range_type, &lowerbound, &upperbound) < 0)
2359 {
2360 lim_warning (_("don't know bounds of array"));
2361 lowerbound = upperbound = 0;
2362 }
2363
2364 idx = pos_atr (ind[i]);
2365 if (idx < lowerbound || idx > upperbound)
2366 lim_warning (_("packed array index %ld out of bounds"),
2367 (long) idx);
2368 bits = TYPE_FIELD_BITSIZE (elt_type, 0);
2369 elt_total_bit_offset += (idx - lowerbound) * bits;
2370 elt_type = ada_check_typedef (TYPE_TARGET_TYPE (elt_type));
2371 }
2372 }
2373 elt_off = elt_total_bit_offset / HOST_CHAR_BIT;
2374 bit_off = elt_total_bit_offset % HOST_CHAR_BIT;
2375
2376 v = ada_value_primitive_packed_val (arr, NULL, elt_off, bit_off,
2377 bits, elt_type);
2378 return v;
2379 }
2380
2381 /* Non-zero iff TYPE includes negative integer values. */
2382
2383 static int
2384 has_negatives (struct type *type)
2385 {
2386 switch (TYPE_CODE (type))
2387 {
2388 default:
2389 return 0;
2390 case TYPE_CODE_INT:
2391 return !TYPE_UNSIGNED (type);
2392 case TYPE_CODE_RANGE:
2393 return TYPE_LOW_BOUND (type) < 0;
2394 }
2395 }
2396
2397 /* With SRC being a buffer containing BIT_SIZE bits of data at BIT_OFFSET,
2398 unpack that data into UNPACKED. UNPACKED_LEN is the size in bytes of
2399 the unpacked buffer.
2400
2401 The size of the unpacked buffer (UNPACKED_LEN) is expected to be large
2402 enough to contain at least BIT_OFFSET bits. If not, an error is raised.
2403
2404 IS_BIG_ENDIAN is nonzero if the data is stored in big endian mode,
2405 zero otherwise.
2406
2407 IS_SIGNED_TYPE is nonzero if the data corresponds to a signed type.
2408
2409 IS_SCALAR is nonzero if the data corresponds to a signed type. */
2410
2411 static void
2412 ada_unpack_from_contents (const gdb_byte *src, int bit_offset, int bit_size,
2413 gdb_byte *unpacked, int unpacked_len,
2414 int is_big_endian, int is_signed_type,
2415 int is_scalar)
2416 {
2417 int src_len = (bit_size + bit_offset + HOST_CHAR_BIT - 1) / 8;
2418 int src_idx; /* Index into the source area */
2419 int src_bytes_left; /* Number of source bytes left to process. */
2420 int srcBitsLeft; /* Number of source bits left to move */
2421 int unusedLS; /* Number of bits in next significant
2422 byte of source that are unused */
2423
2424 int unpacked_idx; /* Index into the unpacked buffer */
2425 int unpacked_bytes_left; /* Number of bytes left to set in unpacked. */
2426
2427 unsigned long accum; /* Staging area for bits being transferred */
2428 int accumSize; /* Number of meaningful bits in accum */
2429 unsigned char sign;
2430
2431 /* Transmit bytes from least to most significant; delta is the direction
2432 the indices move. */
2433 int delta = is_big_endian ? -1 : 1;
2434
2435 /* Make sure that unpacked is large enough to receive the BIT_SIZE
2436 bits from SRC. .*/
2437 if ((bit_size + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT > unpacked_len)
2438 error (_("Cannot unpack %d bits into buffer of %d bytes"),
2439 bit_size, unpacked_len);
2440
2441 srcBitsLeft = bit_size;
2442 src_bytes_left = src_len;
2443 unpacked_bytes_left = unpacked_len;
2444 sign = 0;
2445
2446 if (is_big_endian)
2447 {
2448 src_idx = src_len - 1;
2449 if (is_signed_type
2450 && ((src[0] << bit_offset) & (1 << (HOST_CHAR_BIT - 1))))
2451 sign = ~0;
2452
2453 unusedLS =
2454 (HOST_CHAR_BIT - (bit_size + bit_offset) % HOST_CHAR_BIT)
2455 % HOST_CHAR_BIT;
2456
2457 if (is_scalar)
2458 {
2459 accumSize = 0;
2460 unpacked_idx = unpacked_len - 1;
2461 }
2462 else
2463 {
2464 /* Non-scalar values must be aligned at a byte boundary... */
2465 accumSize =
2466 (HOST_CHAR_BIT - bit_size % HOST_CHAR_BIT) % HOST_CHAR_BIT;
2467 /* ... And are placed at the beginning (most-significant) bytes
2468 of the target. */
2469 unpacked_idx = (bit_size + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT - 1;
2470 unpacked_bytes_left = unpacked_idx + 1;
2471 }
2472 }
2473 else
2474 {
2475 int sign_bit_offset = (bit_size + bit_offset - 1) % 8;
2476
2477 src_idx = unpacked_idx = 0;
2478 unusedLS = bit_offset;
2479 accumSize = 0;
2480
2481 if (is_signed_type && (src[src_len - 1] & (1 << sign_bit_offset)))
2482 sign = ~0;
2483 }
2484
2485 accum = 0;
2486 while (src_bytes_left > 0)
2487 {
2488 /* Mask for removing bits of the next source byte that are not
2489 part of the value. */
2490 unsigned int unusedMSMask =
2491 (1 << (srcBitsLeft >= HOST_CHAR_BIT ? HOST_CHAR_BIT : srcBitsLeft)) -
2492 1;
2493 /* Sign-extend bits for this byte. */
2494 unsigned int signMask = sign & ~unusedMSMask;
2495
2496 accum |=
2497 (((src[src_idx] >> unusedLS) & unusedMSMask) | signMask) << accumSize;
2498 accumSize += HOST_CHAR_BIT - unusedLS;
2499 if (accumSize >= HOST_CHAR_BIT)
2500 {
2501 unpacked[unpacked_idx] = accum & ~(~0UL << HOST_CHAR_BIT);
2502 accumSize -= HOST_CHAR_BIT;
2503 accum >>= HOST_CHAR_BIT;
2504 unpacked_bytes_left -= 1;
2505 unpacked_idx += delta;
2506 }
2507 srcBitsLeft -= HOST_CHAR_BIT - unusedLS;
2508 unusedLS = 0;
2509 src_bytes_left -= 1;
2510 src_idx += delta;
2511 }
2512 while (unpacked_bytes_left > 0)
2513 {
2514 accum |= sign << accumSize;
2515 unpacked[unpacked_idx] = accum & ~(~0UL << HOST_CHAR_BIT);
2516 accumSize -= HOST_CHAR_BIT;
2517 if (accumSize < 0)
2518 accumSize = 0;
2519 accum >>= HOST_CHAR_BIT;
2520 unpacked_bytes_left -= 1;
2521 unpacked_idx += delta;
2522 }
2523 }
2524
2525 /* Create a new value of type TYPE from the contents of OBJ starting
2526 at byte OFFSET, and bit offset BIT_OFFSET within that byte,
2527 proceeding for BIT_SIZE bits. If OBJ is an lval in memory, then
2528 assigning through the result will set the field fetched from.
2529 VALADDR is ignored unless OBJ is NULL, in which case,
2530 VALADDR+OFFSET must address the start of storage containing the
2531 packed value. The value returned in this case is never an lval.
2532 Assumes 0 <= BIT_OFFSET < HOST_CHAR_BIT. */
2533
2534 struct value *
2535 ada_value_primitive_packed_val (struct value *obj, const gdb_byte *valaddr,
2536 long offset, int bit_offset, int bit_size,
2537 struct type *type)
2538 {
2539 struct value *v;
2540 const gdb_byte *src; /* First byte containing data to unpack */
2541 gdb_byte *unpacked;
2542 const int is_scalar = is_scalar_type (type);
2543 const int is_big_endian = gdbarch_bits_big_endian (get_type_arch (type));
2544 gdb::byte_vector staging;
2545
2546 type = ada_check_typedef (type);
2547
2548 if (obj == NULL)
2549 src = valaddr + offset;
2550 else
2551 src = value_contents (obj) + offset;
2552
2553 if (is_dynamic_type (type))
2554 {
2555 /* The length of TYPE might by dynamic, so we need to resolve
2556 TYPE in order to know its actual size, which we then use
2557 to create the contents buffer of the value we return.
2558 The difficulty is that the data containing our object is
2559 packed, and therefore maybe not at a byte boundary. So, what
2560 we do, is unpack the data into a byte-aligned buffer, and then
2561 use that buffer as our object's value for resolving the type. */
2562 int staging_len = (bit_size + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT;
2563 staging.resize (staging_len);
2564
2565 ada_unpack_from_contents (src, bit_offset, bit_size,
2566 staging.data (), staging.size (),
2567 is_big_endian, has_negatives (type),
2568 is_scalar);
2569 type = resolve_dynamic_type (type, staging.data (), 0);
2570 if (TYPE_LENGTH (type) < (bit_size + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT)
2571 {
2572 /* This happens when the length of the object is dynamic,
2573 and is actually smaller than the space reserved for it.
2574 For instance, in an array of variant records, the bit_size
2575 we're given is the array stride, which is constant and
2576 normally equal to the maximum size of its element.
2577 But, in reality, each element only actually spans a portion
2578 of that stride. */
2579 bit_size = TYPE_LENGTH (type) * HOST_CHAR_BIT;
2580 }
2581 }
2582
2583 if (obj == NULL)
2584 {
2585 v = allocate_value (type);
2586 src = valaddr + offset;
2587 }
2588 else if (VALUE_LVAL (obj) == lval_memory && value_lazy (obj))
2589 {
2590 int src_len = (bit_size + bit_offset + HOST_CHAR_BIT - 1) / 8;
2591 gdb_byte *buf;
2592
2593 v = value_at (type, value_address (obj) + offset);
2594 buf = (gdb_byte *) alloca (src_len);
2595 read_memory (value_address (v), buf, src_len);
2596 src = buf;
2597 }
2598 else
2599 {
2600 v = allocate_value (type);
2601 src = value_contents (obj) + offset;
2602 }
2603
2604 if (obj != NULL)
2605 {
2606 long new_offset = offset;
2607
2608 set_value_component_location (v, obj);
2609 set_value_bitpos (v, bit_offset + value_bitpos (obj));
2610 set_value_bitsize (v, bit_size);
2611 if (value_bitpos (v) >= HOST_CHAR_BIT)
2612 {
2613 ++new_offset;
2614 set_value_bitpos (v, value_bitpos (v) - HOST_CHAR_BIT);
2615 }
2616 set_value_offset (v, new_offset);
2617
2618 /* Also set the parent value. This is needed when trying to
2619 assign a new value (in inferior memory). */
2620 set_value_parent (v, obj);
2621 }
2622 else
2623 set_value_bitsize (v, bit_size);
2624 unpacked = value_contents_writeable (v);
2625
2626 if (bit_size == 0)
2627 {
2628 memset (unpacked, 0, TYPE_LENGTH (type));
2629 return v;
2630 }
2631
2632 if (staging.size () == TYPE_LENGTH (type))
2633 {
2634 /* Small short-cut: If we've unpacked the data into a buffer
2635 of the same size as TYPE's length, then we can reuse that,
2636 instead of doing the unpacking again. */
2637 memcpy (unpacked, staging.data (), staging.size ());
2638 }
2639 else
2640 ada_unpack_from_contents (src, bit_offset, bit_size,
2641 unpacked, TYPE_LENGTH (type),
2642 is_big_endian, has_negatives (type), is_scalar);
2643
2644 return v;
2645 }
2646
2647 /* Store the contents of FROMVAL into the location of TOVAL.
2648 Return a new value with the location of TOVAL and contents of
2649 FROMVAL. Handles assignment into packed fields that have
2650 floating-point or non-scalar types. */
2651
2652 static struct value *
2653 ada_value_assign (struct value *toval, struct value *fromval)
2654 {
2655 struct type *type = value_type (toval);
2656 int bits = value_bitsize (toval);
2657
2658 toval = ada_coerce_ref (toval);
2659 fromval = ada_coerce_ref (fromval);
2660
2661 if (ada_is_direct_array_type (value_type (toval)))
2662 toval = ada_coerce_to_simple_array (toval);
2663 if (ada_is_direct_array_type (value_type (fromval)))
2664 fromval = ada_coerce_to_simple_array (fromval);
2665
2666 if (!deprecated_value_modifiable (toval))
2667 error (_("Left operand of assignment is not a modifiable lvalue."));
2668
2669 if (VALUE_LVAL (toval) == lval_memory
2670 && bits > 0
2671 && (TYPE_CODE (type) == TYPE_CODE_FLT
2672 || TYPE_CODE (type) == TYPE_CODE_STRUCT))
2673 {
2674 int len = (value_bitpos (toval)
2675 + bits + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT;
2676 int from_size;
2677 gdb_byte *buffer = (gdb_byte *) alloca (len);
2678 struct value *val;
2679 CORE_ADDR to_addr = value_address (toval);
2680
2681 if (TYPE_CODE (type) == TYPE_CODE_FLT)
2682 fromval = value_cast (type, fromval);
2683
2684 read_memory (to_addr, buffer, len);
2685 from_size = value_bitsize (fromval);
2686 if (from_size == 0)
2687 from_size = TYPE_LENGTH (value_type (fromval)) * TARGET_CHAR_BIT;
2688
2689 const int is_big_endian = gdbarch_bits_big_endian (get_type_arch (type));
2690 ULONGEST from_offset = 0;
2691 if (is_big_endian && is_scalar_type (value_type (fromval)))
2692 from_offset = from_size - bits;
2693 copy_bitwise (buffer, value_bitpos (toval),
2694 value_contents (fromval), from_offset,
2695 bits, is_big_endian);
2696 write_memory_with_notification (to_addr, buffer, len);
2697
2698 val = value_copy (toval);
2699 memcpy (value_contents_raw (val), value_contents (fromval),
2700 TYPE_LENGTH (type));
2701 deprecated_set_value_type (val, type);
2702
2703 return val;
2704 }
2705
2706 return value_assign (toval, fromval);
2707 }
2708
2709
2710 /* Given that COMPONENT is a memory lvalue that is part of the lvalue
2711 CONTAINER, assign the contents of VAL to COMPONENTS's place in
2712 CONTAINER. Modifies the VALUE_CONTENTS of CONTAINER only, not
2713 COMPONENT, and not the inferior's memory. The current contents
2714 of COMPONENT are ignored.
2715
2716 Although not part of the initial design, this function also works
2717 when CONTAINER and COMPONENT are not_lval's: it works as if CONTAINER
2718 had a null address, and COMPONENT had an address which is equal to
2719 its offset inside CONTAINER. */
2720
2721 static void
2722 value_assign_to_component (struct value *container, struct value *component,
2723 struct value *val)
2724 {
2725 LONGEST offset_in_container =
2726 (LONGEST) (value_address (component) - value_address (container));
2727 int bit_offset_in_container =
2728 value_bitpos (component) - value_bitpos (container);
2729 int bits;
2730
2731 val = value_cast (value_type (component), val);
2732
2733 if (value_bitsize (component) == 0)
2734 bits = TARGET_CHAR_BIT * TYPE_LENGTH (value_type (component));
2735 else
2736 bits = value_bitsize (component);
2737
2738 if (gdbarch_bits_big_endian (get_type_arch (value_type (container))))
2739 {
2740 int src_offset;
2741
2742 if (is_scalar_type (check_typedef (value_type (component))))
2743 src_offset
2744 = TYPE_LENGTH (value_type (component)) * TARGET_CHAR_BIT - bits;
2745 else
2746 src_offset = 0;
2747 copy_bitwise (value_contents_writeable (container) + offset_in_container,
2748 value_bitpos (container) + bit_offset_in_container,
2749 value_contents (val), src_offset, bits, 1);
2750 }
2751 else
2752 copy_bitwise (value_contents_writeable (container) + offset_in_container,
2753 value_bitpos (container) + bit_offset_in_container,
2754 value_contents (val), 0, bits, 0);
2755 }
2756
2757 /* Determine if TYPE is an access to an unconstrained array. */
2758
2759 bool
2760 ada_is_access_to_unconstrained_array (struct type *type)
2761 {
2762 return (TYPE_CODE (type) == TYPE_CODE_TYPEDEF
2763 && is_thick_pntr (ada_typedef_target_type (type)));
2764 }
2765
2766 /* The value of the element of array ARR at the ARITY indices given in IND.
2767 ARR may be either a simple array, GNAT array descriptor, or pointer
2768 thereto. */
2769
2770 struct value *
2771 ada_value_subscript (struct value *arr, int arity, struct value **ind)
2772 {
2773 int k;
2774 struct value *elt;
2775 struct type *elt_type;
2776
2777 elt = ada_coerce_to_simple_array (arr);
2778
2779 elt_type = ada_check_typedef (value_type (elt));
2780 if (TYPE_CODE (elt_type) == TYPE_CODE_ARRAY
2781 && TYPE_FIELD_BITSIZE (elt_type, 0) > 0)
2782 return value_subscript_packed (elt, arity, ind);
2783
2784 for (k = 0; k < arity; k += 1)
2785 {
2786 struct type *saved_elt_type = TYPE_TARGET_TYPE (elt_type);
2787
2788 if (TYPE_CODE (elt_type) != TYPE_CODE_ARRAY)
2789 error (_("too many subscripts (%d expected)"), k);
2790
2791 elt = value_subscript (elt, pos_atr (ind[k]));
2792
2793 if (ada_is_access_to_unconstrained_array (saved_elt_type)
2794 && TYPE_CODE (value_type (elt)) != TYPE_CODE_TYPEDEF)
2795 {
2796 /* The element is a typedef to an unconstrained array,
2797 except that the value_subscript call stripped the
2798 typedef layer. The typedef layer is GNAT's way to
2799 specify that the element is, at the source level, an
2800 access to the unconstrained array, rather than the
2801 unconstrained array. So, we need to restore that
2802 typedef layer, which we can do by forcing the element's
2803 type back to its original type. Otherwise, the returned
2804 value is going to be printed as the array, rather
2805 than as an access. Another symptom of the same issue
2806 would be that an expression trying to dereference the
2807 element would also be improperly rejected. */
2808 deprecated_set_value_type (elt, saved_elt_type);
2809 }
2810
2811 elt_type = ada_check_typedef (value_type (elt));
2812 }
2813
2814 return elt;
2815 }
2816
2817 /* Assuming ARR is a pointer to a GDB array, the value of the element
2818 of *ARR at the ARITY indices given in IND.
2819 Does not read the entire array into memory.
2820
2821 Note: Unlike what one would expect, this function is used instead of
2822 ada_value_subscript for basically all non-packed array types. The reason
2823 for this is that a side effect of doing our own pointer arithmetics instead
2824 of relying on value_subscript is that there is no implicit typedef peeling.
2825 This is important for arrays of array accesses, where it allows us to
2826 preserve the fact that the array's element is an array access, where the
2827 access part os encoded in a typedef layer. */
2828
2829 static struct value *
2830 ada_value_ptr_subscript (struct value *arr, int arity, struct value **ind)
2831 {
2832 int k;
2833 struct value *array_ind = ada_value_ind (arr);
2834 struct type *type
2835 = check_typedef (value_enclosing_type (array_ind));
2836
2837 if (TYPE_CODE (type) == TYPE_CODE_ARRAY
2838 && TYPE_FIELD_BITSIZE (type, 0) > 0)
2839 return value_subscript_packed (array_ind, arity, ind);
2840
2841 for (k = 0; k < arity; k += 1)
2842 {
2843 LONGEST lwb, upb;
2844 struct value *lwb_value;
2845
2846 if (TYPE_CODE (type) != TYPE_CODE_ARRAY)
2847 error (_("too many subscripts (%d expected)"), k);
2848 arr = value_cast (lookup_pointer_type (TYPE_TARGET_TYPE (type)),
2849 value_copy (arr));
2850 get_discrete_bounds (TYPE_INDEX_TYPE (type), &lwb, &upb);
2851 lwb_value = value_from_longest (value_type(ind[k]), lwb);
2852 arr = value_ptradd (arr, pos_atr (ind[k]) - pos_atr (lwb_value));
2853 type = TYPE_TARGET_TYPE (type);
2854 }
2855
2856 return value_ind (arr);
2857 }
2858
2859 /* Given that ARRAY_PTR is a pointer or reference to an array of type TYPE (the
2860 actual type of ARRAY_PTR is ignored), returns the Ada slice of
2861 HIGH'Pos-LOW'Pos+1 elements starting at index LOW. The lower bound of
2862 this array is LOW, as per Ada rules. */
2863 static struct value *
2864 ada_value_slice_from_ptr (struct value *array_ptr, struct type *type,
2865 int low, int high)
2866 {
2867 struct type *type0 = ada_check_typedef (type);
2868 struct type *base_index_type = TYPE_TARGET_TYPE (TYPE_INDEX_TYPE (type0));
2869 struct type *index_type
2870 = create_static_range_type (NULL, base_index_type, low, high);
2871 struct type *slice_type = create_array_type_with_stride
2872 (NULL, TYPE_TARGET_TYPE (type0), index_type,
2873 get_dyn_prop (DYN_PROP_BYTE_STRIDE, type0),
2874 TYPE_FIELD_BITSIZE (type0, 0));
2875 int base_low = ada_discrete_type_low_bound (TYPE_INDEX_TYPE (type0));
2876 LONGEST base_low_pos, low_pos;
2877 CORE_ADDR base;
2878
2879 if (!discrete_position (base_index_type, low, &low_pos)
2880 || !discrete_position (base_index_type, base_low, &base_low_pos))
2881 {
2882 warning (_("unable to get positions in slice, use bounds instead"));
2883 low_pos = low;
2884 base_low_pos = base_low;
2885 }
2886
2887 base = value_as_address (array_ptr)
2888 + ((low_pos - base_low_pos)
2889 * TYPE_LENGTH (TYPE_TARGET_TYPE (type0)));
2890 return value_at_lazy (slice_type, base);
2891 }
2892
2893
2894 static struct value *
2895 ada_value_slice (struct value *array, int low, int high)
2896 {
2897 struct type *type = ada_check_typedef (value_type (array));
2898 struct type *base_index_type = TYPE_TARGET_TYPE (TYPE_INDEX_TYPE (type));
2899 struct type *index_type
2900 = create_static_range_type (NULL, TYPE_INDEX_TYPE (type), low, high);
2901 struct type *slice_type = create_array_type_with_stride
2902 (NULL, TYPE_TARGET_TYPE (type), index_type,
2903 get_dyn_prop (DYN_PROP_BYTE_STRIDE, type),
2904 TYPE_FIELD_BITSIZE (type, 0));
2905 LONGEST low_pos, high_pos;
2906
2907 if (!discrete_position (base_index_type, low, &low_pos)
2908 || !discrete_position (base_index_type, high, &high_pos))
2909 {
2910 warning (_("unable to get positions in slice, use bounds instead"));
2911 low_pos = low;
2912 high_pos = high;
2913 }
2914
2915 return value_cast (slice_type,
2916 value_slice (array, low, high_pos - low_pos + 1));
2917 }
2918
2919 /* If type is a record type in the form of a standard GNAT array
2920 descriptor, returns the number of dimensions for type. If arr is a
2921 simple array, returns the number of "array of"s that prefix its
2922 type designation. Otherwise, returns 0. */
2923
2924 int
2925 ada_array_arity (struct type *type)
2926 {
2927 int arity;
2928
2929 if (type == NULL)
2930 return 0;
2931
2932 type = desc_base_type (type);
2933
2934 arity = 0;
2935 if (TYPE_CODE (type) == TYPE_CODE_STRUCT)
2936 return desc_arity (desc_bounds_type (type));
2937 else
2938 while (TYPE_CODE (type) == TYPE_CODE_ARRAY)
2939 {
2940 arity += 1;
2941 type = ada_check_typedef (TYPE_TARGET_TYPE (type));
2942 }
2943
2944 return arity;
2945 }
2946
2947 /* If TYPE is a record type in the form of a standard GNAT array
2948 descriptor or a simple array type, returns the element type for
2949 TYPE after indexing by NINDICES indices, or by all indices if
2950 NINDICES is -1. Otherwise, returns NULL. */
2951
2952 struct type *
2953 ada_array_element_type (struct type *type, int nindices)
2954 {
2955 type = desc_base_type (type);
2956
2957 if (TYPE_CODE (type) == TYPE_CODE_STRUCT)
2958 {
2959 int k;
2960 struct type *p_array_type;
2961
2962 p_array_type = desc_data_target_type (type);
2963
2964 k = ada_array_arity (type);
2965 if (k == 0)
2966 return NULL;
2967
2968 /* Initially p_array_type = elt_type(*)[]...(k times)...[]. */
2969 if (nindices >= 0 && k > nindices)
2970 k = nindices;
2971 while (k > 0 && p_array_type != NULL)
2972 {
2973 p_array_type = ada_check_typedef (TYPE_TARGET_TYPE (p_array_type));
2974 k -= 1;
2975 }
2976 return p_array_type;
2977 }
2978 else if (TYPE_CODE (type) == TYPE_CODE_ARRAY)
2979 {
2980 while (nindices != 0 && TYPE_CODE (type) == TYPE_CODE_ARRAY)
2981 {
2982 type = TYPE_TARGET_TYPE (type);
2983 nindices -= 1;
2984 }
2985 return type;
2986 }
2987
2988 return NULL;
2989 }
2990
2991 /* The type of nth index in arrays of given type (n numbering from 1).
2992 Does not examine memory. Throws an error if N is invalid or TYPE
2993 is not an array type. NAME is the name of the Ada attribute being
2994 evaluated ('range, 'first, 'last, or 'length); it is used in building
2995 the error message. */
2996
2997 static struct type *
2998 ada_index_type (struct type *type, int n, const char *name)
2999 {
3000 struct type *result_type;
3001
3002 type = desc_base_type (type);
3003
3004 if (n < 0 || n > ada_array_arity (type))
3005 error (_("invalid dimension number to '%s"), name);
3006
3007 if (ada_is_simple_array_type (type))
3008 {
3009 int i;
3010
3011 for (i = 1; i < n; i += 1)
3012 type = TYPE_TARGET_TYPE (type);
3013 result_type = TYPE_TARGET_TYPE (TYPE_INDEX_TYPE (type));
3014 /* FIXME: The stabs type r(0,0);bound;bound in an array type
3015 has a target type of TYPE_CODE_UNDEF. We compensate here, but
3016 perhaps stabsread.c would make more sense. */
3017 if (result_type && TYPE_CODE (result_type) == TYPE_CODE_UNDEF)
3018 result_type = NULL;
3019 }
3020 else
3021 {
3022 result_type = desc_index_type (desc_bounds_type (type), n);
3023 if (result_type == NULL)
3024 error (_("attempt to take bound of something that is not an array"));
3025 }
3026
3027 return result_type;
3028 }
3029
3030 /* Given that arr is an array type, returns the lower bound of the
3031 Nth index (numbering from 1) if WHICH is 0, and the upper bound if
3032 WHICH is 1. This returns bounds 0 .. -1 if ARR_TYPE is an
3033 array-descriptor type. It works for other arrays with bounds supplied
3034 by run-time quantities other than discriminants. */
3035
3036 static LONGEST
3037 ada_array_bound_from_type (struct type *arr_type, int n, int which)
3038 {
3039 struct type *type, *index_type_desc, *index_type;
3040 int i;
3041
3042 gdb_assert (which == 0 || which == 1);
3043
3044 if (ada_is_constrained_packed_array_type (arr_type))
3045 arr_type = decode_constrained_packed_array_type (arr_type);
3046
3047 if (arr_type == NULL || !ada_is_simple_array_type (arr_type))
3048 return (LONGEST) - which;
3049
3050 if (TYPE_CODE (arr_type) == TYPE_CODE_PTR)
3051 type = TYPE_TARGET_TYPE (arr_type);
3052 else
3053 type = arr_type;
3054
3055 if (TYPE_FIXED_INSTANCE (type))
3056 {
3057 /* The array has already been fixed, so we do not need to
3058 check the parallel ___XA type again. That encoding has
3059 already been applied, so ignore it now. */
3060 index_type_desc = NULL;
3061 }
3062 else
3063 {
3064 index_type_desc = ada_find_parallel_type (type, "___XA");
3065 ada_fixup_array_indexes_type (index_type_desc);
3066 }
3067
3068 if (index_type_desc != NULL)
3069 index_type = to_fixed_range_type (TYPE_FIELD_TYPE (index_type_desc, n - 1),
3070 NULL);
3071 else
3072 {
3073 struct type *elt_type = check_typedef (type);
3074
3075 for (i = 1; i < n; i++)
3076 elt_type = check_typedef (TYPE_TARGET_TYPE (elt_type));
3077
3078 index_type = TYPE_INDEX_TYPE (elt_type);
3079 }
3080
3081 return
3082 (LONGEST) (which == 0
3083 ? ada_discrete_type_low_bound (index_type)
3084 : ada_discrete_type_high_bound (index_type));
3085 }
3086
3087 /* Given that arr is an array value, returns the lower bound of the
3088 nth index (numbering from 1) if WHICH is 0, and the upper bound if
3089 WHICH is 1. This routine will also work for arrays with bounds
3090 supplied by run-time quantities other than discriminants. */
3091
3092 static LONGEST
3093 ada_array_bound (struct value *arr, int n, int which)
3094 {
3095 struct type *arr_type;
3096
3097 if (TYPE_CODE (check_typedef (value_type (arr))) == TYPE_CODE_PTR)
3098 arr = value_ind (arr);
3099 arr_type = value_enclosing_type (arr);
3100
3101 if (ada_is_constrained_packed_array_type (arr_type))
3102 return ada_array_bound (decode_constrained_packed_array (arr), n, which);
3103 else if (ada_is_simple_array_type (arr_type))
3104 return ada_array_bound_from_type (arr_type, n, which);
3105 else
3106 return value_as_long (desc_one_bound (desc_bounds (arr), n, which));
3107 }
3108
3109 /* Given that arr is an array value, returns the length of the
3110 nth index. This routine will also work for arrays with bounds
3111 supplied by run-time quantities other than discriminants.
3112 Does not work for arrays indexed by enumeration types with representation
3113 clauses at the moment. */
3114
3115 static LONGEST
3116 ada_array_length (struct value *arr, int n)
3117 {
3118 struct type *arr_type, *index_type;
3119 int low, high;
3120
3121 if (TYPE_CODE (check_typedef (value_type (arr))) == TYPE_CODE_PTR)
3122 arr = value_ind (arr);
3123 arr_type = value_enclosing_type (arr);
3124
3125 if (ada_is_constrained_packed_array_type (arr_type))
3126 return ada_array_length (decode_constrained_packed_array (arr), n);
3127
3128 if (ada_is_simple_array_type (arr_type))
3129 {
3130 low = ada_array_bound_from_type (arr_type, n, 0);
3131 high = ada_array_bound_from_type (arr_type, n, 1);
3132 }
3133 else
3134 {
3135 low = value_as_long (desc_one_bound (desc_bounds (arr), n, 0));
3136 high = value_as_long (desc_one_bound (desc_bounds (arr), n, 1));
3137 }
3138
3139 arr_type = check_typedef (arr_type);
3140 index_type = ada_index_type (arr_type, n, "length");
3141 if (index_type != NULL)
3142 {
3143 struct type *base_type;
3144 if (TYPE_CODE (index_type) == TYPE_CODE_RANGE)
3145 base_type = TYPE_TARGET_TYPE (index_type);
3146 else
3147 base_type = index_type;
3148
3149 low = pos_atr (value_from_longest (base_type, low));
3150 high = pos_atr (value_from_longest (base_type, high));
3151 }
3152 return high - low + 1;
3153 }
3154
3155 /* An array whose type is that of ARR_TYPE (an array type), with
3156 bounds LOW to HIGH, but whose contents are unimportant. If HIGH is
3157 less than LOW, then LOW-1 is used. */
3158
3159 static struct value *
3160 empty_array (struct type *arr_type, int low, int high)
3161 {
3162 struct type *arr_type0 = ada_check_typedef (arr_type);
3163 struct type *index_type
3164 = create_static_range_type
3165 (NULL, TYPE_TARGET_TYPE (TYPE_INDEX_TYPE (arr_type0)), low,
3166 high < low ? low - 1 : high);
3167 struct type *elt_type = ada_array_element_type (arr_type0, 1);
3168
3169 return allocate_value (create_array_type (NULL, elt_type, index_type));
3170 }
3171 \f
3172
3173 /* Name resolution */
3174
3175 /* The "decoded" name for the user-definable Ada operator corresponding
3176 to OP. */
3177
3178 static const char *
3179 ada_decoded_op_name (enum exp_opcode op)
3180 {
3181 int i;
3182
3183 for (i = 0; ada_opname_table[i].encoded != NULL; i += 1)
3184 {
3185 if (ada_opname_table[i].op == op)
3186 return ada_opname_table[i].decoded;
3187 }
3188 error (_("Could not find operator name for opcode"));
3189 }
3190
3191
3192 /* Same as evaluate_type (*EXP), but resolves ambiguous symbol
3193 references (marked by OP_VAR_VALUE nodes in which the symbol has an
3194 undefined namespace) and converts operators that are
3195 user-defined into appropriate function calls. If CONTEXT_TYPE is
3196 non-null, it provides a preferred result type [at the moment, only
3197 type void has any effect---causing procedures to be preferred over
3198 functions in calls]. A null CONTEXT_TYPE indicates that a non-void
3199 return type is preferred. May change (expand) *EXP. */
3200
3201 static void
3202 resolve (expression_up *expp, int void_context_p, int parse_completion,
3203 innermost_block_tracker *tracker)
3204 {
3205 struct type *context_type = NULL;
3206 int pc = 0;
3207
3208 if (void_context_p)
3209 context_type = builtin_type ((*expp)->gdbarch)->builtin_void;
3210
3211 resolve_subexp (expp, &pc, 1, context_type, parse_completion, tracker);
3212 }
3213
3214 /* Resolve the operator of the subexpression beginning at
3215 position *POS of *EXPP. "Resolving" consists of replacing
3216 the symbols that have undefined namespaces in OP_VAR_VALUE nodes
3217 with their resolutions, replacing built-in operators with
3218 function calls to user-defined operators, where appropriate, and,
3219 when DEPROCEDURE_P is non-zero, converting function-valued variables
3220 into parameterless calls. May expand *EXPP. The CONTEXT_TYPE functions
3221 are as in ada_resolve, above. */
3222
3223 static struct value *
3224 resolve_subexp (expression_up *expp, int *pos, int deprocedure_p,
3225 struct type *context_type, int parse_completion,
3226 innermost_block_tracker *tracker)
3227 {
3228 int pc = *pos;
3229 int i;
3230 struct expression *exp; /* Convenience: == *expp. */
3231 enum exp_opcode op = (*expp)->elts[pc].opcode;
3232 struct value **argvec; /* Vector of operand types (alloca'ed). */
3233 int nargs; /* Number of operands. */
3234 int oplen;
3235
3236 argvec = NULL;
3237 nargs = 0;
3238 exp = expp->get ();
3239
3240 /* Pass one: resolve operands, saving their types and updating *pos,
3241 if needed. */
3242 switch (op)
3243 {
3244 case OP_FUNCALL:
3245 if (exp->elts[pc + 3].opcode == OP_VAR_VALUE
3246 && SYMBOL_DOMAIN (exp->elts[pc + 5].symbol) == UNDEF_DOMAIN)
3247 *pos += 7;
3248 else
3249 {
3250 *pos += 3;
3251 resolve_subexp (expp, pos, 0, NULL, parse_completion, tracker);
3252 }
3253 nargs = longest_to_int (exp->elts[pc + 1].longconst);
3254 break;
3255
3256 case UNOP_ADDR:
3257 *pos += 1;
3258 resolve_subexp (expp, pos, 0, NULL, parse_completion, tracker);
3259 break;
3260
3261 case UNOP_QUAL:
3262 *pos += 3;
3263 resolve_subexp (expp, pos, 1, check_typedef (exp->elts[pc + 1].type),
3264 parse_completion, tracker);
3265 break;
3266
3267 case OP_ATR_MODULUS:
3268 case OP_ATR_SIZE:
3269 case OP_ATR_TAG:
3270 case OP_ATR_FIRST:
3271 case OP_ATR_LAST:
3272 case OP_ATR_LENGTH:
3273 case OP_ATR_POS:
3274 case OP_ATR_VAL:
3275 case OP_ATR_MIN:
3276 case OP_ATR_MAX:
3277 case TERNOP_IN_RANGE:
3278 case BINOP_IN_BOUNDS:
3279 case UNOP_IN_RANGE:
3280 case OP_AGGREGATE:
3281 case OP_OTHERS:
3282 case OP_CHOICES:
3283 case OP_POSITIONAL:
3284 case OP_DISCRETE_RANGE:
3285 case OP_NAME:
3286 ada_forward_operator_length (exp, pc, &oplen, &nargs);
3287 *pos += oplen;
3288 break;
3289
3290 case BINOP_ASSIGN:
3291 {
3292 struct value *arg1;
3293
3294 *pos += 1;
3295 arg1 = resolve_subexp (expp, pos, 0, NULL, parse_completion, tracker);
3296 if (arg1 == NULL)
3297 resolve_subexp (expp, pos, 1, NULL, parse_completion, tracker);
3298 else
3299 resolve_subexp (expp, pos, 1, value_type (arg1), parse_completion,
3300 tracker);
3301 break;
3302 }
3303
3304 case UNOP_CAST:
3305 *pos += 3;
3306 nargs = 1;
3307 break;
3308
3309 case BINOP_ADD:
3310 case BINOP_SUB:
3311 case BINOP_MUL:
3312 case BINOP_DIV:
3313 case BINOP_REM:
3314 case BINOP_MOD:
3315 case BINOP_EXP:
3316 case BINOP_CONCAT:
3317 case BINOP_LOGICAL_AND:
3318 case BINOP_LOGICAL_OR:
3319 case BINOP_BITWISE_AND:
3320 case BINOP_BITWISE_IOR:
3321 case BINOP_BITWISE_XOR:
3322
3323 case BINOP_EQUAL:
3324 case BINOP_NOTEQUAL:
3325 case BINOP_LESS:
3326 case BINOP_GTR:
3327 case BINOP_LEQ:
3328 case BINOP_GEQ:
3329
3330 case BINOP_REPEAT:
3331 case BINOP_SUBSCRIPT:
3332 case BINOP_COMMA:
3333 *pos += 1;
3334 nargs = 2;
3335 break;
3336
3337 case UNOP_NEG:
3338 case UNOP_PLUS:
3339 case UNOP_LOGICAL_NOT:
3340 case UNOP_ABS:
3341 case UNOP_IND:
3342 *pos += 1;
3343 nargs = 1;
3344 break;
3345
3346 case OP_LONG:
3347 case OP_FLOAT:
3348 case OP_VAR_VALUE:
3349 case OP_VAR_MSYM_VALUE:
3350 *pos += 4;
3351 break;
3352
3353 case OP_TYPE:
3354 case OP_BOOL:
3355 case OP_LAST:
3356 case OP_INTERNALVAR:
3357 *pos += 3;
3358 break;
3359
3360 case UNOP_MEMVAL:
3361 *pos += 3;
3362 nargs = 1;
3363 break;
3364
3365 case OP_REGISTER:
3366 *pos += 4 + BYTES_TO_EXP_ELEM (exp->elts[pc + 1].longconst + 1);
3367 break;
3368
3369 case STRUCTOP_STRUCT:
3370 *pos += 4 + BYTES_TO_EXP_ELEM (exp->elts[pc + 1].longconst + 1);
3371 nargs = 1;
3372 break;
3373
3374 case TERNOP_SLICE:
3375 *pos += 1;
3376 nargs = 3;
3377 break;
3378
3379 case OP_STRING:
3380 break;
3381
3382 default:
3383 error (_("Unexpected operator during name resolution"));
3384 }
3385
3386 argvec = XALLOCAVEC (struct value *, nargs + 1);
3387 for (i = 0; i < nargs; i += 1)
3388 argvec[i] = resolve_subexp (expp, pos, 1, NULL, parse_completion,
3389 tracker);
3390 argvec[i] = NULL;
3391 exp = expp->get ();
3392
3393 /* Pass two: perform any resolution on principal operator. */
3394 switch (op)
3395 {
3396 default:
3397 break;
3398
3399 case OP_VAR_VALUE:
3400 if (SYMBOL_DOMAIN (exp->elts[pc + 2].symbol) == UNDEF_DOMAIN)
3401 {
3402 std::vector<struct block_symbol> candidates;
3403 int n_candidates;
3404
3405 n_candidates =
3406 ada_lookup_symbol_list (SYMBOL_LINKAGE_NAME
3407 (exp->elts[pc + 2].symbol),
3408 exp->elts[pc + 1].block, VAR_DOMAIN,
3409 &candidates);
3410
3411 if (n_candidates > 1)
3412 {
3413 /* Types tend to get re-introduced locally, so if there
3414 are any local symbols that are not types, first filter
3415 out all types. */
3416 int j;
3417 for (j = 0; j < n_candidates; j += 1)
3418 switch (SYMBOL_CLASS (candidates[j].symbol))
3419 {
3420 case LOC_REGISTER:
3421 case LOC_ARG:
3422 case LOC_REF_ARG:
3423 case LOC_REGPARM_ADDR:
3424 case LOC_LOCAL:
3425 case LOC_COMPUTED:
3426 goto FoundNonType;
3427 default:
3428 break;
3429 }
3430 FoundNonType:
3431 if (j < n_candidates)
3432 {
3433 j = 0;
3434 while (j < n_candidates)
3435 {
3436 if (SYMBOL_CLASS (candidates[j].symbol) == LOC_TYPEDEF)
3437 {
3438 candidates[j] = candidates[n_candidates - 1];
3439 n_candidates -= 1;
3440 }
3441 else
3442 j += 1;
3443 }
3444 }
3445 }
3446
3447 if (n_candidates == 0)
3448 error (_("No definition found for %s"),
3449 SYMBOL_PRINT_NAME (exp->elts[pc + 2].symbol));
3450 else if (n_candidates == 1)
3451 i = 0;
3452 else if (deprocedure_p
3453 && !is_nonfunction (candidates.data (), n_candidates))
3454 {
3455 i = ada_resolve_function
3456 (candidates.data (), n_candidates, NULL, 0,
3457 SYMBOL_LINKAGE_NAME (exp->elts[pc + 2].symbol),
3458 context_type, parse_completion);
3459 if (i < 0)
3460 error (_("Could not find a match for %s"),
3461 SYMBOL_PRINT_NAME (exp->elts[pc + 2].symbol));
3462 }
3463 else
3464 {
3465 printf_filtered (_("Multiple matches for %s\n"),
3466 SYMBOL_PRINT_NAME (exp->elts[pc + 2].symbol));
3467 user_select_syms (candidates.data (), n_candidates, 1);
3468 i = 0;
3469 }
3470
3471 exp->elts[pc + 1].block = candidates[i].block;
3472 exp->elts[pc + 2].symbol = candidates[i].symbol;
3473 tracker->update (candidates[i]);
3474 }
3475
3476 if (deprocedure_p
3477 && (TYPE_CODE (SYMBOL_TYPE (exp->elts[pc + 2].symbol))
3478 == TYPE_CODE_FUNC))
3479 {
3480 replace_operator_with_call (expp, pc, 0, 4,
3481 exp->elts[pc + 2].symbol,
3482 exp->elts[pc + 1].block);
3483 exp = expp->get ();
3484 }
3485 break;
3486
3487 case OP_FUNCALL:
3488 {
3489 if (exp->elts[pc + 3].opcode == OP_VAR_VALUE
3490 && SYMBOL_DOMAIN (exp->elts[pc + 5].symbol) == UNDEF_DOMAIN)
3491 {
3492 std::vector<struct block_symbol> candidates;
3493 int n_candidates;
3494
3495 n_candidates =
3496 ada_lookup_symbol_list (SYMBOL_LINKAGE_NAME
3497 (exp->elts[pc + 5].symbol),
3498 exp->elts[pc + 4].block, VAR_DOMAIN,
3499 &candidates);
3500
3501 if (n_candidates == 1)
3502 i = 0;
3503 else
3504 {
3505 i = ada_resolve_function
3506 (candidates.data (), n_candidates,
3507 argvec, nargs,
3508 SYMBOL_LINKAGE_NAME (exp->elts[pc + 5].symbol),
3509 context_type, parse_completion);
3510 if (i < 0)
3511 error (_("Could not find a match for %s"),
3512 SYMBOL_PRINT_NAME (exp->elts[pc + 5].symbol));
3513 }
3514
3515 exp->elts[pc + 4].block = candidates[i].block;
3516 exp->elts[pc + 5].symbol = candidates[i].symbol;
3517 tracker->update (candidates[i]);
3518 }
3519 }
3520 break;
3521 case BINOP_ADD:
3522 case BINOP_SUB:
3523 case BINOP_MUL:
3524 case BINOP_DIV:
3525 case BINOP_REM:
3526 case BINOP_MOD:
3527 case BINOP_CONCAT:
3528 case BINOP_BITWISE_AND:
3529 case BINOP_BITWISE_IOR:
3530 case BINOP_BITWISE_XOR:
3531 case BINOP_EQUAL:
3532 case BINOP_NOTEQUAL:
3533 case BINOP_LESS:
3534 case BINOP_GTR:
3535 case BINOP_LEQ:
3536 case BINOP_GEQ:
3537 case BINOP_EXP:
3538 case UNOP_NEG:
3539 case UNOP_PLUS:
3540 case UNOP_LOGICAL_NOT:
3541 case UNOP_ABS:
3542 if (possible_user_operator_p (op, argvec))
3543 {
3544 std::vector<struct block_symbol> candidates;
3545 int n_candidates;
3546
3547 n_candidates =
3548 ada_lookup_symbol_list (ada_decoded_op_name (op),
3549 NULL, VAR_DOMAIN,
3550 &candidates);
3551
3552 i = ada_resolve_function (candidates.data (), n_candidates, argvec,
3553 nargs, ada_decoded_op_name (op), NULL,
3554 parse_completion);
3555 if (i < 0)
3556 break;
3557
3558 replace_operator_with_call (expp, pc, nargs, 1,
3559 candidates[i].symbol,
3560 candidates[i].block);
3561 exp = expp->get ();
3562 }
3563 break;
3564
3565 case OP_TYPE:
3566 case OP_REGISTER:
3567 return NULL;
3568 }
3569
3570 *pos = pc;
3571 if (exp->elts[pc].opcode == OP_VAR_MSYM_VALUE)
3572 return evaluate_var_msym_value (EVAL_AVOID_SIDE_EFFECTS,
3573 exp->elts[pc + 1].objfile,
3574 exp->elts[pc + 2].msymbol);
3575 else
3576 return evaluate_subexp_type (exp, pos);
3577 }
3578
3579 /* Return non-zero if formal type FTYPE matches actual type ATYPE. If
3580 MAY_DEREF is non-zero, the formal may be a pointer and the actual
3581 a non-pointer. */
3582 /* The term "match" here is rather loose. The match is heuristic and
3583 liberal. */
3584
3585 static int
3586 ada_type_match (struct type *ftype, struct type *atype, int may_deref)
3587 {
3588 ftype = ada_check_typedef (ftype);
3589 atype = ada_check_typedef (atype);
3590
3591 if (TYPE_CODE (ftype) == TYPE_CODE_REF)
3592 ftype = TYPE_TARGET_TYPE (ftype);
3593 if (TYPE_CODE (atype) == TYPE_CODE_REF)
3594 atype = TYPE_TARGET_TYPE (atype);
3595
3596 switch (TYPE_CODE (ftype))
3597 {
3598 default:
3599 return TYPE_CODE (ftype) == TYPE_CODE (atype);
3600 case TYPE_CODE_PTR:
3601 if (TYPE_CODE (atype) == TYPE_CODE_PTR)
3602 return ada_type_match (TYPE_TARGET_TYPE (ftype),
3603 TYPE_TARGET_TYPE (atype), 0);
3604 else
3605 return (may_deref
3606 && ada_type_match (TYPE_TARGET_TYPE (ftype), atype, 0));
3607 case TYPE_CODE_INT:
3608 case TYPE_CODE_ENUM:
3609 case TYPE_CODE_RANGE:
3610 switch (TYPE_CODE (atype))
3611 {
3612 case TYPE_CODE_INT:
3613 case TYPE_CODE_ENUM:
3614 case TYPE_CODE_RANGE:
3615 return 1;
3616 default:
3617 return 0;
3618 }
3619
3620 case TYPE_CODE_ARRAY:
3621 return (TYPE_CODE (atype) == TYPE_CODE_ARRAY
3622 || ada_is_array_descriptor_type (atype));
3623
3624 case TYPE_CODE_STRUCT:
3625 if (ada_is_array_descriptor_type (ftype))
3626 return (TYPE_CODE (atype) == TYPE_CODE_ARRAY
3627 || ada_is_array_descriptor_type (atype));
3628 else
3629 return (TYPE_CODE (atype) == TYPE_CODE_STRUCT
3630 && !ada_is_array_descriptor_type (atype));
3631
3632 case TYPE_CODE_UNION:
3633 case TYPE_CODE_FLT:
3634 return (TYPE_CODE (atype) == TYPE_CODE (ftype));
3635 }
3636 }
3637
3638 /* Return non-zero if the formals of FUNC "sufficiently match" the
3639 vector of actual argument types ACTUALS of size N_ACTUALS. FUNC
3640 may also be an enumeral, in which case it is treated as a 0-
3641 argument function. */
3642
3643 static int
3644 ada_args_match (struct symbol *func, struct value **actuals, int n_actuals)
3645 {
3646 int i;
3647 struct type *func_type = SYMBOL_TYPE (func);
3648
3649 if (SYMBOL_CLASS (func) == LOC_CONST
3650 && TYPE_CODE (func_type) == TYPE_CODE_ENUM)
3651 return (n_actuals == 0);
3652 else if (func_type == NULL || TYPE_CODE (func_type) != TYPE_CODE_FUNC)
3653 return 0;
3654
3655 if (TYPE_NFIELDS (func_type) != n_actuals)
3656 return 0;
3657
3658 for (i = 0; i < n_actuals; i += 1)
3659 {
3660 if (actuals[i] == NULL)
3661 return 0;
3662 else
3663 {
3664 struct type *ftype = ada_check_typedef (TYPE_FIELD_TYPE (func_type,
3665 i));
3666 struct type *atype = ada_check_typedef (value_type (actuals[i]));
3667
3668 if (!ada_type_match (ftype, atype, 1))
3669 return 0;
3670 }
3671 }
3672 return 1;
3673 }
3674
3675 /* False iff function type FUNC_TYPE definitely does not produce a value
3676 compatible with type CONTEXT_TYPE. Conservatively returns 1 if
3677 FUNC_TYPE is not a valid function type with a non-null return type
3678 or an enumerated type. A null CONTEXT_TYPE indicates any non-void type. */
3679
3680 static int
3681 return_match (struct type *func_type, struct type *context_type)
3682 {
3683 struct type *return_type;
3684
3685 if (func_type == NULL)
3686 return 1;
3687
3688 if (TYPE_CODE (func_type) == TYPE_CODE_FUNC)
3689 return_type = get_base_type (TYPE_TARGET_TYPE (func_type));
3690 else
3691 return_type = get_base_type (func_type);
3692 if (return_type == NULL)
3693 return 1;
3694
3695 context_type = get_base_type (context_type);
3696
3697 if (TYPE_CODE (return_type) == TYPE_CODE_ENUM)
3698 return context_type == NULL || return_type == context_type;
3699 else if (context_type == NULL)
3700 return TYPE_CODE (return_type) != TYPE_CODE_VOID;
3701 else
3702 return TYPE_CODE (return_type) == TYPE_CODE (context_type);
3703 }
3704
3705
3706 /* Returns the index in SYMS[0..NSYMS-1] that contains the symbol for the
3707 function (if any) that matches the types of the NARGS arguments in
3708 ARGS. If CONTEXT_TYPE is non-null and there is at least one match
3709 that returns that type, then eliminate matches that don't. If
3710 CONTEXT_TYPE is void and there is at least one match that does not
3711 return void, eliminate all matches that do.
3712
3713 Asks the user if there is more than one match remaining. Returns -1
3714 if there is no such symbol or none is selected. NAME is used
3715 solely for messages. May re-arrange and modify SYMS in
3716 the process; the index returned is for the modified vector. */
3717
3718 static int
3719 ada_resolve_function (struct block_symbol syms[],
3720 int nsyms, struct value **args, int nargs,
3721 const char *name, struct type *context_type,
3722 int parse_completion)
3723 {
3724 int fallback;
3725 int k;
3726 int m; /* Number of hits */
3727
3728 m = 0;
3729 /* In the first pass of the loop, we only accept functions matching
3730 context_type. If none are found, we add a second pass of the loop
3731 where every function is accepted. */
3732 for (fallback = 0; m == 0 && fallback < 2; fallback++)
3733 {
3734 for (k = 0; k < nsyms; k += 1)
3735 {
3736 struct type *type = ada_check_typedef (SYMBOL_TYPE (syms[k].symbol));
3737
3738 if (ada_args_match (syms[k].symbol, args, nargs)
3739 && (fallback || return_match (type, context_type)))
3740 {
3741 syms[m] = syms[k];
3742 m += 1;
3743 }
3744 }
3745 }
3746
3747 /* If we got multiple matches, ask the user which one to use. Don't do this
3748 interactive thing during completion, though, as the purpose of the
3749 completion is providing a list of all possible matches. Prompting the
3750 user to filter it down would be completely unexpected in this case. */
3751 if (m == 0)
3752 return -1;
3753 else if (m > 1 && !parse_completion)
3754 {
3755 printf_filtered (_("Multiple matches for %s\n"), name);
3756 user_select_syms (syms, m, 1);
3757 return 0;
3758 }
3759 return 0;
3760 }
3761
3762 /* Returns true (non-zero) iff decoded name N0 should appear before N1
3763 in a listing of choices during disambiguation (see sort_choices, below).
3764 The idea is that overloadings of a subprogram name from the
3765 same package should sort in their source order. We settle for ordering
3766 such symbols by their trailing number (__N or $N). */
3767
3768 static int
3769 encoded_ordered_before (const char *N0, const char *N1)
3770 {
3771 if (N1 == NULL)
3772 return 0;
3773 else if (N0 == NULL)
3774 return 1;
3775 else
3776 {
3777 int k0, k1;
3778
3779 for (k0 = strlen (N0) - 1; k0 > 0 && isdigit (N0[k0]); k0 -= 1)
3780 ;
3781 for (k1 = strlen (N1) - 1; k1 > 0 && isdigit (N1[k1]); k1 -= 1)
3782 ;
3783 if ((N0[k0] == '_' || N0[k0] == '$') && N0[k0 + 1] != '\000'
3784 && (N1[k1] == '_' || N1[k1] == '$') && N1[k1 + 1] != '\000')
3785 {
3786 int n0, n1;
3787
3788 n0 = k0;
3789 while (N0[n0] == '_' && n0 > 0 && N0[n0 - 1] == '_')
3790 n0 -= 1;
3791 n1 = k1;
3792 while (N1[n1] == '_' && n1 > 0 && N1[n1 - 1] == '_')
3793 n1 -= 1;
3794 if (n0 == n1 && strncmp (N0, N1, n0) == 0)
3795 return (atoi (N0 + k0 + 1) < atoi (N1 + k1 + 1));
3796 }
3797 return (strcmp (N0, N1) < 0);
3798 }
3799 }
3800
3801 /* Sort SYMS[0..NSYMS-1] to put the choices in a canonical order by the
3802 encoded names. */
3803
3804 static void
3805 sort_choices (struct block_symbol syms[], int nsyms)
3806 {
3807 int i;
3808
3809 for (i = 1; i < nsyms; i += 1)
3810 {
3811 struct block_symbol sym = syms[i];
3812 int j;
3813
3814 for (j = i - 1; j >= 0; j -= 1)
3815 {
3816 if (encoded_ordered_before (SYMBOL_LINKAGE_NAME (syms[j].symbol),
3817 SYMBOL_LINKAGE_NAME (sym.symbol)))
3818 break;
3819 syms[j + 1] = syms[j];
3820 }
3821 syms[j + 1] = sym;
3822 }
3823 }
3824
3825 /* Whether GDB should display formals and return types for functions in the
3826 overloads selection menu. */
3827 static int print_signatures = 1;
3828
3829 /* Print the signature for SYM on STREAM according to the FLAGS options. For
3830 all but functions, the signature is just the name of the symbol. For
3831 functions, this is the name of the function, the list of types for formals
3832 and the return type (if any). */
3833
3834 static void
3835 ada_print_symbol_signature (struct ui_file *stream, struct symbol *sym,
3836 const struct type_print_options *flags)
3837 {
3838 struct type *type = SYMBOL_TYPE (sym);
3839
3840 fprintf_filtered (stream, "%s", SYMBOL_PRINT_NAME (sym));
3841 if (!print_signatures
3842 || type == NULL
3843 || TYPE_CODE (type) != TYPE_CODE_FUNC)
3844 return;
3845
3846 if (TYPE_NFIELDS (type) > 0)
3847 {
3848 int i;
3849
3850 fprintf_filtered (stream, " (");
3851 for (i = 0; i < TYPE_NFIELDS (type); ++i)
3852 {
3853 if (i > 0)
3854 fprintf_filtered (stream, "; ");
3855 ada_print_type (TYPE_FIELD_TYPE (type, i), NULL, stream, -1, 0,
3856 flags);
3857 }
3858 fprintf_filtered (stream, ")");
3859 }
3860 if (TYPE_TARGET_TYPE (type) != NULL
3861 && TYPE_CODE (TYPE_TARGET_TYPE (type)) != TYPE_CODE_VOID)
3862 {
3863 fprintf_filtered (stream, " return ");
3864 ada_print_type (TYPE_TARGET_TYPE (type), NULL, stream, -1, 0, flags);
3865 }
3866 }
3867
3868 /* Given a list of NSYMS symbols in SYMS, select up to MAX_RESULTS>0
3869 by asking the user (if necessary), returning the number selected,
3870 and setting the first elements of SYMS items. Error if no symbols
3871 selected. */
3872
3873 /* NOTE: Adapted from decode_line_2 in symtab.c, with which it ought
3874 to be re-integrated one of these days. */
3875
3876 int
3877 user_select_syms (struct block_symbol *syms, int nsyms, int max_results)
3878 {
3879 int i;
3880 int *chosen = XALLOCAVEC (int , nsyms);
3881 int n_chosen;
3882 int first_choice = (max_results == 1) ? 1 : 2;
3883 const char *select_mode = multiple_symbols_select_mode ();
3884
3885 if (max_results < 1)
3886 error (_("Request to select 0 symbols!"));
3887 if (nsyms <= 1)
3888 return nsyms;
3889
3890 if (select_mode == multiple_symbols_cancel)
3891 error (_("\
3892 canceled because the command is ambiguous\n\
3893 See set/show multiple-symbol."));
3894
3895 /* If select_mode is "all", then return all possible symbols.
3896 Only do that if more than one symbol can be selected, of course.
3897 Otherwise, display the menu as usual. */
3898 if (select_mode == multiple_symbols_all && max_results > 1)
3899 return nsyms;
3900
3901 printf_filtered (_("[0] cancel\n"));
3902 if (max_results > 1)
3903 printf_filtered (_("[1] all\n"));
3904
3905 sort_choices (syms, nsyms);
3906
3907 for (i = 0; i < nsyms; i += 1)
3908 {
3909 if (syms[i].symbol == NULL)
3910 continue;
3911
3912 if (SYMBOL_CLASS (syms[i].symbol) == LOC_BLOCK)
3913 {
3914 struct symtab_and_line sal =
3915 find_function_start_sal (syms[i].symbol, 1);
3916
3917 printf_filtered ("[%d] ", i + first_choice);
3918 ada_print_symbol_signature (gdb_stdout, syms[i].symbol,
3919 &type_print_raw_options);
3920 if (sal.symtab == NULL)
3921 printf_filtered (_(" at <no source file available>:%d\n"),
3922 sal.line);
3923 else
3924 printf_filtered (_(" at %s:%d\n"),
3925 symtab_to_filename_for_display (sal.symtab),
3926 sal.line);
3927 continue;
3928 }
3929 else
3930 {
3931 int is_enumeral =
3932 (SYMBOL_CLASS (syms[i].symbol) == LOC_CONST
3933 && SYMBOL_TYPE (syms[i].symbol) != NULL
3934 && TYPE_CODE (SYMBOL_TYPE (syms[i].symbol)) == TYPE_CODE_ENUM);
3935 struct symtab *symtab = NULL;
3936
3937 if (SYMBOL_OBJFILE_OWNED (syms[i].symbol))
3938 symtab = symbol_symtab (syms[i].symbol);
3939
3940 if (SYMBOL_LINE (syms[i].symbol) != 0 && symtab != NULL)
3941 {
3942 printf_filtered ("[%d] ", i + first_choice);
3943 ada_print_symbol_signature (gdb_stdout, syms[i].symbol,
3944 &type_print_raw_options);
3945 printf_filtered (_(" at %s:%d\n"),
3946 symtab_to_filename_for_display (symtab),
3947 SYMBOL_LINE (syms[i].symbol));
3948 }
3949 else if (is_enumeral
3950 && TYPE_NAME (SYMBOL_TYPE (syms[i].symbol)) != NULL)
3951 {
3952 printf_filtered (("[%d] "), i + first_choice);
3953 ada_print_type (SYMBOL_TYPE (syms[i].symbol), NULL,
3954 gdb_stdout, -1, 0, &type_print_raw_options);
3955 printf_filtered (_("'(%s) (enumeral)\n"),
3956 SYMBOL_PRINT_NAME (syms[i].symbol));
3957 }
3958 else
3959 {
3960 printf_filtered ("[%d] ", i + first_choice);
3961 ada_print_symbol_signature (gdb_stdout, syms[i].symbol,
3962 &type_print_raw_options);
3963
3964 if (symtab != NULL)
3965 printf_filtered (is_enumeral
3966 ? _(" in %s (enumeral)\n")
3967 : _(" at %s:?\n"),
3968 symtab_to_filename_for_display (symtab));
3969 else
3970 printf_filtered (is_enumeral
3971 ? _(" (enumeral)\n")
3972 : _(" at ?\n"));
3973 }
3974 }
3975 }
3976
3977 n_chosen = get_selections (chosen, nsyms, max_results, max_results > 1,
3978 "overload-choice");
3979
3980 for (i = 0; i < n_chosen; i += 1)
3981 syms[i] = syms[chosen[i]];
3982
3983 return n_chosen;
3984 }
3985
3986 /* Read and validate a set of numeric choices from the user in the
3987 range 0 .. N_CHOICES-1. Place the results in increasing
3988 order in CHOICES[0 .. N-1], and return N.
3989
3990 The user types choices as a sequence of numbers on one line
3991 separated by blanks, encoding them as follows:
3992
3993 + A choice of 0 means to cancel the selection, throwing an error.
3994 + If IS_ALL_CHOICE, a choice of 1 selects the entire set 0 .. N_CHOICES-1.
3995 + The user chooses k by typing k+IS_ALL_CHOICE+1.
3996
3997 The user is not allowed to choose more than MAX_RESULTS values.
3998
3999 ANNOTATION_SUFFIX, if present, is used to annotate the input
4000 prompts (for use with the -f switch). */
4001
4002 int
4003 get_selections (int *choices, int n_choices, int max_results,
4004 int is_all_choice, const char *annotation_suffix)
4005 {
4006 char *args;
4007 const char *prompt;
4008 int n_chosen;
4009 int first_choice = is_all_choice ? 2 : 1;
4010
4011 prompt = getenv ("PS2");
4012 if (prompt == NULL)
4013 prompt = "> ";
4014
4015 args = command_line_input (prompt, annotation_suffix);
4016
4017 if (args == NULL)
4018 error_no_arg (_("one or more choice numbers"));
4019
4020 n_chosen = 0;
4021
4022 /* Set choices[0 .. n_chosen-1] to the users' choices in ascending
4023 order, as given in args. Choices are validated. */
4024 while (1)
4025 {
4026 char *args2;
4027 int choice, j;
4028
4029 args = skip_spaces (args);
4030 if (*args == '\0' && n_chosen == 0)
4031 error_no_arg (_("one or more choice numbers"));
4032 else if (*args == '\0')
4033 break;
4034
4035 choice = strtol (args, &args2, 10);
4036 if (args == args2 || choice < 0
4037 || choice > n_choices + first_choice - 1)
4038 error (_("Argument must be choice number"));
4039 args = args2;
4040
4041 if (choice == 0)
4042 error (_("cancelled"));
4043
4044 if (choice < first_choice)
4045 {
4046 n_chosen = n_choices;
4047 for (j = 0; j < n_choices; j += 1)
4048 choices[j] = j;
4049 break;
4050 }
4051 choice -= first_choice;
4052
4053 for (j = n_chosen - 1; j >= 0 && choice < choices[j]; j -= 1)
4054 {
4055 }
4056
4057 if (j < 0 || choice != choices[j])
4058 {
4059 int k;
4060
4061 for (k = n_chosen - 1; k > j; k -= 1)
4062 choices[k + 1] = choices[k];
4063 choices[j + 1] = choice;
4064 n_chosen += 1;
4065 }
4066 }
4067
4068 if (n_chosen > max_results)
4069 error (_("Select no more than %d of the above"), max_results);
4070
4071 return n_chosen;
4072 }
4073
4074 /* Replace the operator of length OPLEN at position PC in *EXPP with a call
4075 on the function identified by SYM and BLOCK, and taking NARGS
4076 arguments. Update *EXPP as needed to hold more space. */
4077
4078 static void
4079 replace_operator_with_call (expression_up *expp, int pc, int nargs,
4080 int oplen, struct symbol *sym,
4081 const struct block *block)
4082 {
4083 /* A new expression, with 6 more elements (3 for funcall, 4 for function
4084 symbol, -oplen for operator being replaced). */
4085 struct expression *newexp = (struct expression *)
4086 xzalloc (sizeof (struct expression)
4087 + EXP_ELEM_TO_BYTES ((*expp)->nelts + 7 - oplen));
4088 struct expression *exp = expp->get ();
4089
4090 newexp->nelts = exp->nelts + 7 - oplen;
4091 newexp->language_defn = exp->language_defn;
4092 newexp->gdbarch = exp->gdbarch;
4093 memcpy (newexp->elts, exp->elts, EXP_ELEM_TO_BYTES (pc));
4094 memcpy (newexp->elts + pc + 7, exp->elts + pc + oplen,
4095 EXP_ELEM_TO_BYTES (exp->nelts - pc - oplen));
4096
4097 newexp->elts[pc].opcode = newexp->elts[pc + 2].opcode = OP_FUNCALL;
4098 newexp->elts[pc + 1].longconst = (LONGEST) nargs;
4099
4100 newexp->elts[pc + 3].opcode = newexp->elts[pc + 6].opcode = OP_VAR_VALUE;
4101 newexp->elts[pc + 4].block = block;
4102 newexp->elts[pc + 5].symbol = sym;
4103
4104 expp->reset (newexp);
4105 }
4106
4107 /* Type-class predicates */
4108
4109 /* True iff TYPE is numeric (i.e., an INT, RANGE (of numeric type),
4110 or FLOAT). */
4111
4112 static int
4113 numeric_type_p (struct type *type)
4114 {
4115 if (type == NULL)
4116 return 0;
4117 else
4118 {
4119 switch (TYPE_CODE (type))
4120 {
4121 case TYPE_CODE_INT:
4122 case TYPE_CODE_FLT:
4123 return 1;
4124 case TYPE_CODE_RANGE:
4125 return (type == TYPE_TARGET_TYPE (type)
4126 || numeric_type_p (TYPE_TARGET_TYPE (type)));
4127 default:
4128 return 0;
4129 }
4130 }
4131 }
4132
4133 /* True iff TYPE is integral (an INT or RANGE of INTs). */
4134
4135 static int
4136 integer_type_p (struct type *type)
4137 {
4138 if (type == NULL)
4139 return 0;
4140 else
4141 {
4142 switch (TYPE_CODE (type))
4143 {
4144 case TYPE_CODE_INT:
4145 return 1;
4146 case TYPE_CODE_RANGE:
4147 return (type == TYPE_TARGET_TYPE (type)
4148 || integer_type_p (TYPE_TARGET_TYPE (type)));
4149 default:
4150 return 0;
4151 }
4152 }
4153 }
4154
4155 /* True iff TYPE is scalar (INT, RANGE, FLOAT, ENUM). */
4156
4157 static int
4158 scalar_type_p (struct type *type)
4159 {
4160 if (type == NULL)
4161 return 0;
4162 else
4163 {
4164 switch (TYPE_CODE (type))
4165 {
4166 case TYPE_CODE_INT:
4167 case TYPE_CODE_RANGE:
4168 case TYPE_CODE_ENUM:
4169 case TYPE_CODE_FLT:
4170 return 1;
4171 default:
4172 return 0;
4173 }
4174 }
4175 }
4176
4177 /* True iff TYPE is discrete (INT, RANGE, ENUM). */
4178
4179 static int
4180 discrete_type_p (struct type *type)
4181 {
4182 if (type == NULL)
4183 return 0;
4184 else
4185 {
4186 switch (TYPE_CODE (type))
4187 {
4188 case TYPE_CODE_INT:
4189 case TYPE_CODE_RANGE:
4190 case TYPE_CODE_ENUM:
4191 case TYPE_CODE_BOOL:
4192 return 1;
4193 default:
4194 return 0;
4195 }
4196 }
4197 }
4198
4199 /* Returns non-zero if OP with operands in the vector ARGS could be
4200 a user-defined function. Errs on the side of pre-defined operators
4201 (i.e., result 0). */
4202
4203 static int
4204 possible_user_operator_p (enum exp_opcode op, struct value *args[])
4205 {
4206 struct type *type0 =
4207 (args[0] == NULL) ? NULL : ada_check_typedef (value_type (args[0]));
4208 struct type *type1 =
4209 (args[1] == NULL) ? NULL : ada_check_typedef (value_type (args[1]));
4210
4211 if (type0 == NULL)
4212 return 0;
4213
4214 switch (op)
4215 {
4216 default:
4217 return 0;
4218
4219 case BINOP_ADD:
4220 case BINOP_SUB:
4221 case BINOP_MUL:
4222 case BINOP_DIV:
4223 return (!(numeric_type_p (type0) && numeric_type_p (type1)));
4224
4225 case BINOP_REM:
4226 case BINOP_MOD:
4227 case BINOP_BITWISE_AND:
4228 case BINOP_BITWISE_IOR:
4229 case BINOP_BITWISE_XOR:
4230 return (!(integer_type_p (type0) && integer_type_p (type1)));
4231
4232 case BINOP_EQUAL:
4233 case BINOP_NOTEQUAL:
4234 case BINOP_LESS:
4235 case BINOP_GTR:
4236 case BINOP_LEQ:
4237 case BINOP_GEQ:
4238 return (!(scalar_type_p (type0) && scalar_type_p (type1)));
4239
4240 case BINOP_CONCAT:
4241 return !ada_is_array_type (type0) || !ada_is_array_type (type1);
4242
4243 case BINOP_EXP:
4244 return (!(numeric_type_p (type0) && integer_type_p (type1)));
4245
4246 case UNOP_NEG:
4247 case UNOP_PLUS:
4248 case UNOP_LOGICAL_NOT:
4249 case UNOP_ABS:
4250 return (!numeric_type_p (type0));
4251
4252 }
4253 }
4254 \f
4255 /* Renaming */
4256
4257 /* NOTES:
4258
4259 1. In the following, we assume that a renaming type's name may
4260 have an ___XD suffix. It would be nice if this went away at some
4261 point.
4262 2. We handle both the (old) purely type-based representation of
4263 renamings and the (new) variable-based encoding. At some point,
4264 it is devoutly to be hoped that the former goes away
4265 (FIXME: hilfinger-2007-07-09).
4266 3. Subprogram renamings are not implemented, although the XRS
4267 suffix is recognized (FIXME: hilfinger-2007-07-09). */
4268
4269 /* If SYM encodes a renaming,
4270
4271 <renaming> renames <renamed entity>,
4272
4273 sets *LEN to the length of the renamed entity's name,
4274 *RENAMED_ENTITY to that name (not null-terminated), and *RENAMING_EXPR to
4275 the string describing the subcomponent selected from the renamed
4276 entity. Returns ADA_NOT_RENAMING if SYM does not encode a renaming
4277 (in which case, the values of *RENAMED_ENTITY, *LEN, and *RENAMING_EXPR
4278 are undefined). Otherwise, returns a value indicating the category
4279 of entity renamed: an object (ADA_OBJECT_RENAMING), exception
4280 (ADA_EXCEPTION_RENAMING), package (ADA_PACKAGE_RENAMING), or
4281 subprogram (ADA_SUBPROGRAM_RENAMING). Does no allocation; the
4282 strings returned in *RENAMED_ENTITY and *RENAMING_EXPR should not be
4283 deallocated. The values of RENAMED_ENTITY, LEN, or RENAMING_EXPR
4284 may be NULL, in which case they are not assigned.
4285
4286 [Currently, however, GCC does not generate subprogram renamings.] */
4287
4288 enum ada_renaming_category
4289 ada_parse_renaming (struct symbol *sym,
4290 const char **renamed_entity, int *len,
4291 const char **renaming_expr)
4292 {
4293 enum ada_renaming_category kind;
4294 const char *info;
4295 const char *suffix;
4296
4297 if (sym == NULL)
4298 return ADA_NOT_RENAMING;
4299 switch (SYMBOL_CLASS (sym))
4300 {
4301 default:
4302 return ADA_NOT_RENAMING;
4303 case LOC_TYPEDEF:
4304 return parse_old_style_renaming (SYMBOL_TYPE (sym),
4305 renamed_entity, len, renaming_expr);
4306 case LOC_LOCAL:
4307 case LOC_STATIC:
4308 case LOC_COMPUTED:
4309 case LOC_OPTIMIZED_OUT:
4310 info = strstr (SYMBOL_LINKAGE_NAME (sym), "___XR");
4311 if (info == NULL)
4312 return ADA_NOT_RENAMING;
4313 switch (info[5])
4314 {
4315 case '_':
4316 kind = ADA_OBJECT_RENAMING;
4317 info += 6;
4318 break;
4319 case 'E':
4320 kind = ADA_EXCEPTION_RENAMING;
4321 info += 7;
4322 break;
4323 case 'P':
4324 kind = ADA_PACKAGE_RENAMING;
4325 info += 7;
4326 break;
4327 case 'S':
4328 kind = ADA_SUBPROGRAM_RENAMING;
4329 info += 7;
4330 break;
4331 default:
4332 return ADA_NOT_RENAMING;
4333 }
4334 }
4335
4336 if (renamed_entity != NULL)
4337 *renamed_entity = info;
4338 suffix = strstr (info, "___XE");
4339 if (suffix == NULL || suffix == info)
4340 return ADA_NOT_RENAMING;
4341 if (len != NULL)
4342 *len = strlen (info) - strlen (suffix);
4343 suffix += 5;
4344 if (renaming_expr != NULL)
4345 *renaming_expr = suffix;
4346 return kind;
4347 }
4348
4349 /* Assuming TYPE encodes a renaming according to the old encoding in
4350 exp_dbug.ads, returns details of that renaming in *RENAMED_ENTITY,
4351 *LEN, and *RENAMING_EXPR, as for ada_parse_renaming, above. Returns
4352 ADA_NOT_RENAMING otherwise. */
4353 static enum ada_renaming_category
4354 parse_old_style_renaming (struct type *type,
4355 const char **renamed_entity, int *len,
4356 const char **renaming_expr)
4357 {
4358 enum ada_renaming_category kind;
4359 const char *name;
4360 const char *info;
4361 const char *suffix;
4362
4363 if (type == NULL || TYPE_CODE (type) != TYPE_CODE_ENUM
4364 || TYPE_NFIELDS (type) != 1)
4365 return ADA_NOT_RENAMING;
4366
4367 name = TYPE_NAME (type);
4368 if (name == NULL)
4369 return ADA_NOT_RENAMING;
4370
4371 name = strstr (name, "___XR");
4372 if (name == NULL)
4373 return ADA_NOT_RENAMING;
4374 switch (name[5])
4375 {
4376 case '\0':
4377 case '_':
4378 kind = ADA_OBJECT_RENAMING;
4379 break;
4380 case 'E':
4381 kind = ADA_EXCEPTION_RENAMING;
4382 break;
4383 case 'P':
4384 kind = ADA_PACKAGE_RENAMING;
4385 break;
4386 case 'S':
4387 kind = ADA_SUBPROGRAM_RENAMING;
4388 break;
4389 default:
4390 return ADA_NOT_RENAMING;
4391 }
4392
4393 info = TYPE_FIELD_NAME (type, 0);
4394 if (info == NULL)
4395 return ADA_NOT_RENAMING;
4396 if (renamed_entity != NULL)
4397 *renamed_entity = info;
4398 suffix = strstr (info, "___XE");
4399 if (renaming_expr != NULL)
4400 *renaming_expr = suffix + 5;
4401 if (suffix == NULL || suffix == info)
4402 return ADA_NOT_RENAMING;
4403 if (len != NULL)
4404 *len = suffix - info;
4405 return kind;
4406 }
4407
4408 /* Compute the value of the given RENAMING_SYM, which is expected to
4409 be a symbol encoding a renaming expression. BLOCK is the block
4410 used to evaluate the renaming. */
4411
4412 static struct value *
4413 ada_read_renaming_var_value (struct symbol *renaming_sym,
4414 const struct block *block)
4415 {
4416 const char *sym_name;
4417
4418 sym_name = SYMBOL_LINKAGE_NAME (renaming_sym);
4419 expression_up expr = parse_exp_1 (&sym_name, 0, block, 0);
4420 return evaluate_expression (expr.get ());
4421 }
4422 \f
4423
4424 /* Evaluation: Function Calls */
4425
4426 /* Return an lvalue containing the value VAL. This is the identity on
4427 lvalues, and otherwise has the side-effect of allocating memory
4428 in the inferior where a copy of the value contents is copied. */
4429
4430 static struct value *
4431 ensure_lval (struct value *val)
4432 {
4433 if (VALUE_LVAL (val) == not_lval
4434 || VALUE_LVAL (val) == lval_internalvar)
4435 {
4436 int len = TYPE_LENGTH (ada_check_typedef (value_type (val)));
4437 const CORE_ADDR addr =
4438 value_as_long (value_allocate_space_in_inferior (len));
4439
4440 VALUE_LVAL (val) = lval_memory;
4441 set_value_address (val, addr);
4442 write_memory (addr, value_contents (val), len);
4443 }
4444
4445 return val;
4446 }
4447
4448 /* Return the value ACTUAL, converted to be an appropriate value for a
4449 formal of type FORMAL_TYPE. Use *SP as a stack pointer for
4450 allocating any necessary descriptors (fat pointers), or copies of
4451 values not residing in memory, updating it as needed. */
4452
4453 struct value *
4454 ada_convert_actual (struct value *actual, struct type *formal_type0)
4455 {
4456 struct type *actual_type = ada_check_typedef (value_type (actual));
4457 struct type *formal_type = ada_check_typedef (formal_type0);
4458 struct type *formal_target =
4459 TYPE_CODE (formal_type) == TYPE_CODE_PTR
4460 ? ada_check_typedef (TYPE_TARGET_TYPE (formal_type)) : formal_type;
4461 struct type *actual_target =
4462 TYPE_CODE (actual_type) == TYPE_CODE_PTR
4463 ? ada_check_typedef (TYPE_TARGET_TYPE (actual_type)) : actual_type;
4464
4465 if (ada_is_array_descriptor_type (formal_target)
4466 && TYPE_CODE (actual_target) == TYPE_CODE_ARRAY)
4467 return make_array_descriptor (formal_type, actual);
4468 else if (TYPE_CODE (formal_type) == TYPE_CODE_PTR
4469 || TYPE_CODE (formal_type) == TYPE_CODE_REF)
4470 {
4471 struct value *result;
4472
4473 if (TYPE_CODE (formal_target) == TYPE_CODE_ARRAY
4474 && ada_is_array_descriptor_type (actual_target))
4475 result = desc_data (actual);
4476 else if (TYPE_CODE (formal_type) != TYPE_CODE_PTR)
4477 {
4478 if (VALUE_LVAL (actual) != lval_memory)
4479 {
4480 struct value *val;
4481
4482 actual_type = ada_check_typedef (value_type (actual));
4483 val = allocate_value (actual_type);
4484 memcpy ((char *) value_contents_raw (val),
4485 (char *) value_contents (actual),
4486 TYPE_LENGTH (actual_type));
4487 actual = ensure_lval (val);
4488 }
4489 result = value_addr (actual);
4490 }
4491 else
4492 return actual;
4493 return value_cast_pointers (formal_type, result, 0);
4494 }
4495 else if (TYPE_CODE (actual_type) == TYPE_CODE_PTR)
4496 return ada_value_ind (actual);
4497 else if (ada_is_aligner_type (formal_type))
4498 {
4499 /* We need to turn this parameter into an aligner type
4500 as well. */
4501 struct value *aligner = allocate_value (formal_type);
4502 struct value *component = ada_value_struct_elt (aligner, "F", 0);
4503
4504 value_assign_to_component (aligner, component, actual);
4505 return aligner;
4506 }
4507
4508 return actual;
4509 }
4510
4511 /* Convert VALUE (which must be an address) to a CORE_ADDR that is a pointer of
4512 type TYPE. This is usually an inefficient no-op except on some targets
4513 (such as AVR) where the representation of a pointer and an address
4514 differs. */
4515
4516 static CORE_ADDR
4517 value_pointer (struct value *value, struct type *type)
4518 {
4519 struct gdbarch *gdbarch = get_type_arch (type);
4520 unsigned len = TYPE_LENGTH (type);
4521 gdb_byte *buf = (gdb_byte *) alloca (len);
4522 CORE_ADDR addr;
4523
4524 addr = value_address (value);
4525 gdbarch_address_to_pointer (gdbarch, type, buf, addr);
4526 addr = extract_unsigned_integer (buf, len, gdbarch_byte_order (gdbarch));
4527 return addr;
4528 }
4529
4530
4531 /* Push a descriptor of type TYPE for array value ARR on the stack at
4532 *SP, updating *SP to reflect the new descriptor. Return either
4533 an lvalue representing the new descriptor, or (if TYPE is a pointer-
4534 to-descriptor type rather than a descriptor type), a struct value *
4535 representing a pointer to this descriptor. */
4536
4537 static struct value *
4538 make_array_descriptor (struct type *type, struct value *arr)
4539 {
4540 struct type *bounds_type = desc_bounds_type (type);
4541 struct type *desc_type = desc_base_type (type);
4542 struct value *descriptor = allocate_value (desc_type);
4543 struct value *bounds = allocate_value (bounds_type);
4544 int i;
4545
4546 for (i = ada_array_arity (ada_check_typedef (value_type (arr)));
4547 i > 0; i -= 1)
4548 {
4549 modify_field (value_type (bounds), value_contents_writeable (bounds),
4550 ada_array_bound (arr, i, 0),
4551 desc_bound_bitpos (bounds_type, i, 0),
4552 desc_bound_bitsize (bounds_type, i, 0));
4553 modify_field (value_type (bounds), value_contents_writeable (bounds),
4554 ada_array_bound (arr, i, 1),
4555 desc_bound_bitpos (bounds_type, i, 1),
4556 desc_bound_bitsize (bounds_type, i, 1));
4557 }
4558
4559 bounds = ensure_lval (bounds);
4560
4561 modify_field (value_type (descriptor),
4562 value_contents_writeable (descriptor),
4563 value_pointer (ensure_lval (arr),
4564 TYPE_FIELD_TYPE (desc_type, 0)),
4565 fat_pntr_data_bitpos (desc_type),
4566 fat_pntr_data_bitsize (desc_type));
4567
4568 modify_field (value_type (descriptor),
4569 value_contents_writeable (descriptor),
4570 value_pointer (bounds,
4571 TYPE_FIELD_TYPE (desc_type, 1)),
4572 fat_pntr_bounds_bitpos (desc_type),
4573 fat_pntr_bounds_bitsize (desc_type));
4574
4575 descriptor = ensure_lval (descriptor);
4576
4577 if (TYPE_CODE (type) == TYPE_CODE_PTR)
4578 return value_addr (descriptor);
4579 else
4580 return descriptor;
4581 }
4582 \f
4583 /* Symbol Cache Module */
4584
4585 /* Performance measurements made as of 2010-01-15 indicate that
4586 this cache does bring some noticeable improvements. Depending
4587 on the type of entity being printed, the cache can make it as much
4588 as an order of magnitude faster than without it.
4589
4590 The descriptive type DWARF extension has significantly reduced
4591 the need for this cache, at least when DWARF is being used. However,
4592 even in this case, some expensive name-based symbol searches are still
4593 sometimes necessary - to find an XVZ variable, mostly. */
4594
4595 /* Initialize the contents of SYM_CACHE. */
4596
4597 static void
4598 ada_init_symbol_cache (struct ada_symbol_cache *sym_cache)
4599 {
4600 obstack_init (&sym_cache->cache_space);
4601 memset (sym_cache->root, '\000', sizeof (sym_cache->root));
4602 }
4603
4604 /* Free the memory used by SYM_CACHE. */
4605
4606 static void
4607 ada_free_symbol_cache (struct ada_symbol_cache *sym_cache)
4608 {
4609 obstack_free (&sym_cache->cache_space, NULL);
4610 xfree (sym_cache);
4611 }
4612
4613 /* Return the symbol cache associated to the given program space PSPACE.
4614 If not allocated for this PSPACE yet, allocate and initialize one. */
4615
4616 static struct ada_symbol_cache *
4617 ada_get_symbol_cache (struct program_space *pspace)
4618 {
4619 struct ada_pspace_data *pspace_data = get_ada_pspace_data (pspace);
4620
4621 if (pspace_data->sym_cache == NULL)
4622 {
4623 pspace_data->sym_cache = XCNEW (struct ada_symbol_cache);
4624 ada_init_symbol_cache (pspace_data->sym_cache);
4625 }
4626
4627 return pspace_data->sym_cache;
4628 }
4629
4630 /* Clear all entries from the symbol cache. */
4631
4632 static void
4633 ada_clear_symbol_cache (void)
4634 {
4635 struct ada_symbol_cache *sym_cache
4636 = ada_get_symbol_cache (current_program_space);
4637
4638 obstack_free (&sym_cache->cache_space, NULL);
4639 ada_init_symbol_cache (sym_cache);
4640 }
4641
4642 /* Search our cache for an entry matching NAME and DOMAIN.
4643 Return it if found, or NULL otherwise. */
4644
4645 static struct cache_entry **
4646 find_entry (const char *name, domain_enum domain)
4647 {
4648 struct ada_symbol_cache *sym_cache
4649 = ada_get_symbol_cache (current_program_space);
4650 int h = msymbol_hash (name) % HASH_SIZE;
4651 struct cache_entry **e;
4652
4653 for (e = &sym_cache->root[h]; *e != NULL; e = &(*e)->next)
4654 {
4655 if (domain == (*e)->domain && strcmp (name, (*e)->name) == 0)
4656 return e;
4657 }
4658 return NULL;
4659 }
4660
4661 /* Search the symbol cache for an entry matching NAME and DOMAIN.
4662 Return 1 if found, 0 otherwise.
4663
4664 If an entry was found and SYM is not NULL, set *SYM to the entry's
4665 SYM. Same principle for BLOCK if not NULL. */
4666
4667 static int
4668 lookup_cached_symbol (const char *name, domain_enum domain,
4669 struct symbol **sym, const struct block **block)
4670 {
4671 struct cache_entry **e = find_entry (name, domain);
4672
4673 if (e == NULL)
4674 return 0;
4675 if (sym != NULL)
4676 *sym = (*e)->sym;
4677 if (block != NULL)
4678 *block = (*e)->block;
4679 return 1;
4680 }
4681
4682 /* Assuming that (SYM, BLOCK) is the result of the lookup of NAME
4683 in domain DOMAIN, save this result in our symbol cache. */
4684
4685 static void
4686 cache_symbol (const char *name, domain_enum domain, struct symbol *sym,
4687 const struct block *block)
4688 {
4689 struct ada_symbol_cache *sym_cache
4690 = ada_get_symbol_cache (current_program_space);
4691 int h;
4692 char *copy;
4693 struct cache_entry *e;
4694
4695 /* Symbols for builtin types don't have a block.
4696 For now don't cache such symbols. */
4697 if (sym != NULL && !SYMBOL_OBJFILE_OWNED (sym))
4698 return;
4699
4700 /* If the symbol is a local symbol, then do not cache it, as a search
4701 for that symbol depends on the context. To determine whether
4702 the symbol is local or not, we check the block where we found it
4703 against the global and static blocks of its associated symtab. */
4704 if (sym
4705 && BLOCKVECTOR_BLOCK (SYMTAB_BLOCKVECTOR (symbol_symtab (sym)),
4706 GLOBAL_BLOCK) != block
4707 && BLOCKVECTOR_BLOCK (SYMTAB_BLOCKVECTOR (symbol_symtab (sym)),
4708 STATIC_BLOCK) != block)
4709 return;
4710
4711 h = msymbol_hash (name) % HASH_SIZE;
4712 e = XOBNEW (&sym_cache->cache_space, cache_entry);
4713 e->next = sym_cache->root[h];
4714 sym_cache->root[h] = e;
4715 e->name = copy
4716 = (char *) obstack_alloc (&sym_cache->cache_space, strlen (name) + 1);
4717 strcpy (copy, name);
4718 e->sym = sym;
4719 e->domain = domain;
4720 e->block = block;
4721 }
4722 \f
4723 /* Symbol Lookup */
4724
4725 /* Return the symbol name match type that should be used used when
4726 searching for all symbols matching LOOKUP_NAME.
4727
4728 LOOKUP_NAME is expected to be a symbol name after transformation
4729 for Ada lookups. */
4730
4731 static symbol_name_match_type
4732 name_match_type_from_name (const char *lookup_name)
4733 {
4734 return (strstr (lookup_name, "__") == NULL
4735 ? symbol_name_match_type::WILD
4736 : symbol_name_match_type::FULL);
4737 }
4738
4739 /* Return the result of a standard (literal, C-like) lookup of NAME in
4740 given DOMAIN, visible from lexical block BLOCK. */
4741
4742 static struct symbol *
4743 standard_lookup (const char *name, const struct block *block,
4744 domain_enum domain)
4745 {
4746 /* Initialize it just to avoid a GCC false warning. */
4747 struct block_symbol sym = {};
4748
4749 if (lookup_cached_symbol (name, domain, &sym.symbol, NULL))
4750 return sym.symbol;
4751 ada_lookup_encoded_symbol (name, block, domain, &sym);
4752 cache_symbol (name, domain, sym.symbol, sym.block);
4753 return sym.symbol;
4754 }
4755
4756
4757 /* Non-zero iff there is at least one non-function/non-enumeral symbol
4758 in the symbol fields of SYMS[0..N-1]. We treat enumerals as functions,
4759 since they contend in overloading in the same way. */
4760 static int
4761 is_nonfunction (struct block_symbol syms[], int n)
4762 {
4763 int i;
4764
4765 for (i = 0; i < n; i += 1)
4766 if (TYPE_CODE (SYMBOL_TYPE (syms[i].symbol)) != TYPE_CODE_FUNC
4767 && (TYPE_CODE (SYMBOL_TYPE (syms[i].symbol)) != TYPE_CODE_ENUM
4768 || SYMBOL_CLASS (syms[i].symbol) != LOC_CONST))
4769 return 1;
4770
4771 return 0;
4772 }
4773
4774 /* If true (non-zero), then TYPE0 and TYPE1 represent equivalent
4775 struct types. Otherwise, they may not. */
4776
4777 static int
4778 equiv_types (struct type *type0, struct type *type1)
4779 {
4780 if (type0 == type1)
4781 return 1;
4782 if (type0 == NULL || type1 == NULL
4783 || TYPE_CODE (type0) != TYPE_CODE (type1))
4784 return 0;
4785 if ((TYPE_CODE (type0) == TYPE_CODE_STRUCT
4786 || TYPE_CODE (type0) == TYPE_CODE_ENUM)
4787 && ada_type_name (type0) != NULL && ada_type_name (type1) != NULL
4788 && strcmp (ada_type_name (type0), ada_type_name (type1)) == 0)
4789 return 1;
4790
4791 return 0;
4792 }
4793
4794 /* True iff SYM0 represents the same entity as SYM1, or one that is
4795 no more defined than that of SYM1. */
4796
4797 static int
4798 lesseq_defined_than (struct symbol *sym0, struct symbol *sym1)
4799 {
4800 if (sym0 == sym1)
4801 return 1;
4802 if (SYMBOL_DOMAIN (sym0) != SYMBOL_DOMAIN (sym1)
4803 || SYMBOL_CLASS (sym0) != SYMBOL_CLASS (sym1))
4804 return 0;
4805
4806 switch (SYMBOL_CLASS (sym0))
4807 {
4808 case LOC_UNDEF:
4809 return 1;
4810 case LOC_TYPEDEF:
4811 {
4812 struct type *type0 = SYMBOL_TYPE (sym0);
4813 struct type *type1 = SYMBOL_TYPE (sym1);
4814 const char *name0 = SYMBOL_LINKAGE_NAME (sym0);
4815 const char *name1 = SYMBOL_LINKAGE_NAME (sym1);
4816 int len0 = strlen (name0);
4817
4818 return
4819 TYPE_CODE (type0) == TYPE_CODE (type1)
4820 && (equiv_types (type0, type1)
4821 || (len0 < strlen (name1) && strncmp (name0, name1, len0) == 0
4822 && startswith (name1 + len0, "___XV")));
4823 }
4824 case LOC_CONST:
4825 return SYMBOL_VALUE (sym0) == SYMBOL_VALUE (sym1)
4826 && equiv_types (SYMBOL_TYPE (sym0), SYMBOL_TYPE (sym1));
4827 default:
4828 return 0;
4829 }
4830 }
4831
4832 /* Append (SYM,BLOCK,SYMTAB) to the end of the array of struct block_symbol
4833 records in OBSTACKP. Do nothing if SYM is a duplicate. */
4834
4835 static void
4836 add_defn_to_vec (struct obstack *obstackp,
4837 struct symbol *sym,
4838 const struct block *block)
4839 {
4840 int i;
4841 struct block_symbol *prevDefns = defns_collected (obstackp, 0);
4842
4843 /* Do not try to complete stub types, as the debugger is probably
4844 already scanning all symbols matching a certain name at the
4845 time when this function is called. Trying to replace the stub
4846 type by its associated full type will cause us to restart a scan
4847 which may lead to an infinite recursion. Instead, the client
4848 collecting the matching symbols will end up collecting several
4849 matches, with at least one of them complete. It can then filter
4850 out the stub ones if needed. */
4851
4852 for (i = num_defns_collected (obstackp) - 1; i >= 0; i -= 1)
4853 {
4854 if (lesseq_defined_than (sym, prevDefns[i].symbol))
4855 return;
4856 else if (lesseq_defined_than (prevDefns[i].symbol, sym))
4857 {
4858 prevDefns[i].symbol = sym;
4859 prevDefns[i].block = block;
4860 return;
4861 }
4862 }
4863
4864 {
4865 struct block_symbol info;
4866
4867 info.symbol = sym;
4868 info.block = block;
4869 obstack_grow (obstackp, &info, sizeof (struct block_symbol));
4870 }
4871 }
4872
4873 /* Number of block_symbol structures currently collected in current vector in
4874 OBSTACKP. */
4875
4876 static int
4877 num_defns_collected (struct obstack *obstackp)
4878 {
4879 return obstack_object_size (obstackp) / sizeof (struct block_symbol);
4880 }
4881
4882 /* Vector of block_symbol structures currently collected in current vector in
4883 OBSTACKP. If FINISH, close off the vector and return its final address. */
4884
4885 static struct block_symbol *
4886 defns_collected (struct obstack *obstackp, int finish)
4887 {
4888 if (finish)
4889 return (struct block_symbol *) obstack_finish (obstackp);
4890 else
4891 return (struct block_symbol *) obstack_base (obstackp);
4892 }
4893
4894 /* Return a bound minimal symbol matching NAME according to Ada
4895 decoding rules. Returns an invalid symbol if there is no such
4896 minimal symbol. Names prefixed with "standard__" are handled
4897 specially: "standard__" is first stripped off, and only static and
4898 global symbols are searched. */
4899
4900 struct bound_minimal_symbol
4901 ada_lookup_simple_minsym (const char *name)
4902 {
4903 struct bound_minimal_symbol result;
4904
4905 memset (&result, 0, sizeof (result));
4906
4907 symbol_name_match_type match_type = name_match_type_from_name (name);
4908 lookup_name_info lookup_name (name, match_type);
4909
4910 symbol_name_matcher_ftype *match_name
4911 = ada_get_symbol_name_matcher (lookup_name);
4912
4913 for (objfile *objfile : current_program_space->objfiles ())
4914 {
4915 for (minimal_symbol *msymbol : objfile->msymbols ())
4916 {
4917 if (match_name (MSYMBOL_LINKAGE_NAME (msymbol), lookup_name, NULL)
4918 && MSYMBOL_TYPE (msymbol) != mst_solib_trampoline)
4919 {
4920 result.minsym = msymbol;
4921 result.objfile = objfile;
4922 break;
4923 }
4924 }
4925 }
4926
4927 return result;
4928 }
4929
4930 /* Return all the bound minimal symbols matching NAME according to Ada
4931 decoding rules. Returns an empty vector if there is no such
4932 minimal symbol. Names prefixed with "standard__" are handled
4933 specially: "standard__" is first stripped off, and only static and
4934 global symbols are searched. */
4935
4936 static std::vector<struct bound_minimal_symbol>
4937 ada_lookup_simple_minsyms (const char *name)
4938 {
4939 std::vector<struct bound_minimal_symbol> result;
4940
4941 symbol_name_match_type match_type = name_match_type_from_name (name);
4942 lookup_name_info lookup_name (name, match_type);
4943
4944 symbol_name_matcher_ftype *match_name
4945 = ada_get_symbol_name_matcher (lookup_name);
4946
4947 for (objfile *objfile : current_program_space->objfiles ())
4948 {
4949 for (minimal_symbol *msymbol : objfile->msymbols ())
4950 {
4951 if (match_name (MSYMBOL_LINKAGE_NAME (msymbol), lookup_name, NULL)
4952 && MSYMBOL_TYPE (msymbol) != mst_solib_trampoline)
4953 result.push_back ({msymbol, objfile});
4954 }
4955 }
4956
4957 return result;
4958 }
4959
4960 /* For all subprograms that statically enclose the subprogram of the
4961 selected frame, add symbols matching identifier NAME in DOMAIN
4962 and their blocks to the list of data in OBSTACKP, as for
4963 ada_add_block_symbols (q.v.). If WILD_MATCH_P, treat as NAME
4964 with a wildcard prefix. */
4965
4966 static void
4967 add_symbols_from_enclosing_procs (struct obstack *obstackp,
4968 const lookup_name_info &lookup_name,
4969 domain_enum domain)
4970 {
4971 }
4972
4973 /* True if TYPE is definitely an artificial type supplied to a symbol
4974 for which no debugging information was given in the symbol file. */
4975
4976 static int
4977 is_nondebugging_type (struct type *type)
4978 {
4979 const char *name = ada_type_name (type);
4980
4981 return (name != NULL && strcmp (name, "<variable, no debug info>") == 0);
4982 }
4983
4984 /* Return nonzero if TYPE1 and TYPE2 are two enumeration types
4985 that are deemed "identical" for practical purposes.
4986
4987 This function assumes that TYPE1 and TYPE2 are both TYPE_CODE_ENUM
4988 types and that their number of enumerals is identical (in other
4989 words, TYPE_NFIELDS (type1) == TYPE_NFIELDS (type2)). */
4990
4991 static int
4992 ada_identical_enum_types_p (struct type *type1, struct type *type2)
4993 {
4994 int i;
4995
4996 /* The heuristic we use here is fairly conservative. We consider
4997 that 2 enumerate types are identical if they have the same
4998 number of enumerals and that all enumerals have the same
4999 underlying value and name. */
5000
5001 /* All enums in the type should have an identical underlying value. */
5002 for (i = 0; i < TYPE_NFIELDS (type1); i++)
5003 if (TYPE_FIELD_ENUMVAL (type1, i) != TYPE_FIELD_ENUMVAL (type2, i))
5004 return 0;
5005
5006 /* All enumerals should also have the same name (modulo any numerical
5007 suffix). */
5008 for (i = 0; i < TYPE_NFIELDS (type1); i++)
5009 {
5010 const char *name_1 = TYPE_FIELD_NAME (type1, i);
5011 const char *name_2 = TYPE_FIELD_NAME (type2, i);
5012 int len_1 = strlen (name_1);
5013 int len_2 = strlen (name_2);
5014
5015 ada_remove_trailing_digits (TYPE_FIELD_NAME (type1, i), &len_1);
5016 ada_remove_trailing_digits (TYPE_FIELD_NAME (type2, i), &len_2);
5017 if (len_1 != len_2
5018 || strncmp (TYPE_FIELD_NAME (type1, i),
5019 TYPE_FIELD_NAME (type2, i),
5020 len_1) != 0)
5021 return 0;
5022 }
5023
5024 return 1;
5025 }
5026
5027 /* Return nonzero if all the symbols in SYMS are all enumeral symbols
5028 that are deemed "identical" for practical purposes. Sometimes,
5029 enumerals are not strictly identical, but their types are so similar
5030 that they can be considered identical.
5031
5032 For instance, consider the following code:
5033
5034 type Color is (Black, Red, Green, Blue, White);
5035 type RGB_Color is new Color range Red .. Blue;
5036
5037 Type RGB_Color is a subrange of an implicit type which is a copy
5038 of type Color. If we call that implicit type RGB_ColorB ("B" is
5039 for "Base Type"), then type RGB_ColorB is a copy of type Color.
5040 As a result, when an expression references any of the enumeral
5041 by name (Eg. "print green"), the expression is technically
5042 ambiguous and the user should be asked to disambiguate. But
5043 doing so would only hinder the user, since it wouldn't matter
5044 what choice he makes, the outcome would always be the same.
5045 So, for practical purposes, we consider them as the same. */
5046
5047 static int
5048 symbols_are_identical_enums (const std::vector<struct block_symbol> &syms)
5049 {
5050 int i;
5051
5052 /* Before performing a thorough comparison check of each type,
5053 we perform a series of inexpensive checks. We expect that these
5054 checks will quickly fail in the vast majority of cases, and thus
5055 help prevent the unnecessary use of a more expensive comparison.
5056 Said comparison also expects us to make some of these checks
5057 (see ada_identical_enum_types_p). */
5058
5059 /* Quick check: All symbols should have an enum type. */
5060 for (i = 0; i < syms.size (); i++)
5061 if (TYPE_CODE (SYMBOL_TYPE (syms[i].symbol)) != TYPE_CODE_ENUM)
5062 return 0;
5063
5064 /* Quick check: They should all have the same value. */
5065 for (i = 1; i < syms.size (); i++)
5066 if (SYMBOL_VALUE (syms[i].symbol) != SYMBOL_VALUE (syms[0].symbol))
5067 return 0;
5068
5069 /* Quick check: They should all have the same number of enumerals. */
5070 for (i = 1; i < syms.size (); i++)
5071 if (TYPE_NFIELDS (SYMBOL_TYPE (syms[i].symbol))
5072 != TYPE_NFIELDS (SYMBOL_TYPE (syms[0].symbol)))
5073 return 0;
5074
5075 /* All the sanity checks passed, so we might have a set of
5076 identical enumeration types. Perform a more complete
5077 comparison of the type of each symbol. */
5078 for (i = 1; i < syms.size (); i++)
5079 if (!ada_identical_enum_types_p (SYMBOL_TYPE (syms[i].symbol),
5080 SYMBOL_TYPE (syms[0].symbol)))
5081 return 0;
5082
5083 return 1;
5084 }
5085
5086 /* Remove any non-debugging symbols in SYMS that definitely
5087 duplicate other symbols in the list (The only case I know of where
5088 this happens is when object files containing stabs-in-ecoff are
5089 linked with files containing ordinary ecoff debugging symbols (or no
5090 debugging symbols)). Modifies SYMS to squeeze out deleted entries.
5091 Returns the number of items in the modified list. */
5092
5093 static int
5094 remove_extra_symbols (std::vector<struct block_symbol> *syms)
5095 {
5096 int i, j;
5097
5098 /* We should never be called with less than 2 symbols, as there
5099 cannot be any extra symbol in that case. But it's easy to
5100 handle, since we have nothing to do in that case. */
5101 if (syms->size () < 2)
5102 return syms->size ();
5103
5104 i = 0;
5105 while (i < syms->size ())
5106 {
5107 int remove_p = 0;
5108
5109 /* If two symbols have the same name and one of them is a stub type,
5110 the get rid of the stub. */
5111
5112 if (TYPE_STUB (SYMBOL_TYPE ((*syms)[i].symbol))
5113 && SYMBOL_LINKAGE_NAME ((*syms)[i].symbol) != NULL)
5114 {
5115 for (j = 0; j < syms->size (); j++)
5116 {
5117 if (j != i
5118 && !TYPE_STUB (SYMBOL_TYPE ((*syms)[j].symbol))
5119 && SYMBOL_LINKAGE_NAME ((*syms)[j].symbol) != NULL
5120 && strcmp (SYMBOL_LINKAGE_NAME ((*syms)[i].symbol),
5121 SYMBOL_LINKAGE_NAME ((*syms)[j].symbol)) == 0)
5122 remove_p = 1;
5123 }
5124 }
5125
5126 /* Two symbols with the same name, same class and same address
5127 should be identical. */
5128
5129 else if (SYMBOL_LINKAGE_NAME ((*syms)[i].symbol) != NULL
5130 && SYMBOL_CLASS ((*syms)[i].symbol) == LOC_STATIC
5131 && is_nondebugging_type (SYMBOL_TYPE ((*syms)[i].symbol)))
5132 {
5133 for (j = 0; j < syms->size (); j += 1)
5134 {
5135 if (i != j
5136 && SYMBOL_LINKAGE_NAME ((*syms)[j].symbol) != NULL
5137 && strcmp (SYMBOL_LINKAGE_NAME ((*syms)[i].symbol),
5138 SYMBOL_LINKAGE_NAME ((*syms)[j].symbol)) == 0
5139 && SYMBOL_CLASS ((*syms)[i].symbol)
5140 == SYMBOL_CLASS ((*syms)[j].symbol)
5141 && SYMBOL_VALUE_ADDRESS ((*syms)[i].symbol)
5142 == SYMBOL_VALUE_ADDRESS ((*syms)[j].symbol))
5143 remove_p = 1;
5144 }
5145 }
5146
5147 if (remove_p)
5148 syms->erase (syms->begin () + i);
5149
5150 i += 1;
5151 }
5152
5153 /* If all the remaining symbols are identical enumerals, then
5154 just keep the first one and discard the rest.
5155
5156 Unlike what we did previously, we do not discard any entry
5157 unless they are ALL identical. This is because the symbol
5158 comparison is not a strict comparison, but rather a practical
5159 comparison. If all symbols are considered identical, then
5160 we can just go ahead and use the first one and discard the rest.
5161 But if we cannot reduce the list to a single element, we have
5162 to ask the user to disambiguate anyways. And if we have to
5163 present a multiple-choice menu, it's less confusing if the list
5164 isn't missing some choices that were identical and yet distinct. */
5165 if (symbols_are_identical_enums (*syms))
5166 syms->resize (1);
5167
5168 return syms->size ();
5169 }
5170
5171 /* Given a type that corresponds to a renaming entity, use the type name
5172 to extract the scope (package name or function name, fully qualified,
5173 and following the GNAT encoding convention) where this renaming has been
5174 defined. */
5175
5176 static std::string
5177 xget_renaming_scope (struct type *renaming_type)
5178 {
5179 /* The renaming types adhere to the following convention:
5180 <scope>__<rename>___<XR extension>.
5181 So, to extract the scope, we search for the "___XR" extension,
5182 and then backtrack until we find the first "__". */
5183
5184 const char *name = TYPE_NAME (renaming_type);
5185 const char *suffix = strstr (name, "___XR");
5186 const char *last;
5187
5188 /* Now, backtrack a bit until we find the first "__". Start looking
5189 at suffix - 3, as the <rename> part is at least one character long. */
5190
5191 for (last = suffix - 3; last > name; last--)
5192 if (last[0] == '_' && last[1] == '_')
5193 break;
5194
5195 /* Make a copy of scope and return it. */
5196 return std::string (name, last);
5197 }
5198
5199 /* Return nonzero if NAME corresponds to a package name. */
5200
5201 static int
5202 is_package_name (const char *name)
5203 {
5204 /* Here, We take advantage of the fact that no symbols are generated
5205 for packages, while symbols are generated for each function.
5206 So the condition for NAME represent a package becomes equivalent
5207 to NAME not existing in our list of symbols. There is only one
5208 small complication with library-level functions (see below). */
5209
5210 /* If it is a function that has not been defined at library level,
5211 then we should be able to look it up in the symbols. */
5212 if (standard_lookup (name, NULL, VAR_DOMAIN) != NULL)
5213 return 0;
5214
5215 /* Library-level function names start with "_ada_". See if function
5216 "_ada_" followed by NAME can be found. */
5217
5218 /* Do a quick check that NAME does not contain "__", since library-level
5219 functions names cannot contain "__" in them. */
5220 if (strstr (name, "__") != NULL)
5221 return 0;
5222
5223 std::string fun_name = string_printf ("_ada_%s", name);
5224
5225 return (standard_lookup (fun_name.c_str (), NULL, VAR_DOMAIN) == NULL);
5226 }
5227
5228 /* Return nonzero if SYM corresponds to a renaming entity that is
5229 not visible from FUNCTION_NAME. */
5230
5231 static int
5232 old_renaming_is_invisible (const struct symbol *sym, const char *function_name)
5233 {
5234 if (SYMBOL_CLASS (sym) != LOC_TYPEDEF)
5235 return 0;
5236
5237 std::string scope = xget_renaming_scope (SYMBOL_TYPE (sym));
5238
5239 /* If the rename has been defined in a package, then it is visible. */
5240 if (is_package_name (scope.c_str ()))
5241 return 0;
5242
5243 /* Check that the rename is in the current function scope by checking
5244 that its name starts with SCOPE. */
5245
5246 /* If the function name starts with "_ada_", it means that it is
5247 a library-level function. Strip this prefix before doing the
5248 comparison, as the encoding for the renaming does not contain
5249 this prefix. */
5250 if (startswith (function_name, "_ada_"))
5251 function_name += 5;
5252
5253 return !startswith (function_name, scope.c_str ());
5254 }
5255
5256 /* Remove entries from SYMS that corresponds to a renaming entity that
5257 is not visible from the function associated with CURRENT_BLOCK or
5258 that is superfluous due to the presence of more specific renaming
5259 information. Places surviving symbols in the initial entries of
5260 SYMS and returns the number of surviving symbols.
5261
5262 Rationale:
5263 First, in cases where an object renaming is implemented as a
5264 reference variable, GNAT may produce both the actual reference
5265 variable and the renaming encoding. In this case, we discard the
5266 latter.
5267
5268 Second, GNAT emits a type following a specified encoding for each renaming
5269 entity. Unfortunately, STABS currently does not support the definition
5270 of types that are local to a given lexical block, so all renamings types
5271 are emitted at library level. As a consequence, if an application
5272 contains two renaming entities using the same name, and a user tries to
5273 print the value of one of these entities, the result of the ada symbol
5274 lookup will also contain the wrong renaming type.
5275
5276 This function partially covers for this limitation by attempting to
5277 remove from the SYMS list renaming symbols that should be visible
5278 from CURRENT_BLOCK. However, there does not seem be a 100% reliable
5279 method with the current information available. The implementation
5280 below has a couple of limitations (FIXME: brobecker-2003-05-12):
5281
5282 - When the user tries to print a rename in a function while there
5283 is another rename entity defined in a package: Normally, the
5284 rename in the function has precedence over the rename in the
5285 package, so the latter should be removed from the list. This is
5286 currently not the case.
5287
5288 - This function will incorrectly remove valid renames if
5289 the CURRENT_BLOCK corresponds to a function which symbol name
5290 has been changed by an "Export" pragma. As a consequence,
5291 the user will be unable to print such rename entities. */
5292
5293 static int
5294 remove_irrelevant_renamings (std::vector<struct block_symbol> *syms,
5295 const struct block *current_block)
5296 {
5297 struct symbol *current_function;
5298 const char *current_function_name;
5299 int i;
5300 int is_new_style_renaming;
5301
5302 /* If there is both a renaming foo___XR... encoded as a variable and
5303 a simple variable foo in the same block, discard the latter.
5304 First, zero out such symbols, then compress. */
5305 is_new_style_renaming = 0;
5306 for (i = 0; i < syms->size (); i += 1)
5307 {
5308 struct symbol *sym = (*syms)[i].symbol;
5309 const struct block *block = (*syms)[i].block;
5310 const char *name;
5311 const char *suffix;
5312
5313 if (sym == NULL || SYMBOL_CLASS (sym) == LOC_TYPEDEF)
5314 continue;
5315 name = SYMBOL_LINKAGE_NAME (sym);
5316 suffix = strstr (name, "___XR");
5317
5318 if (suffix != NULL)
5319 {
5320 int name_len = suffix - name;
5321 int j;
5322
5323 is_new_style_renaming = 1;
5324 for (j = 0; j < syms->size (); j += 1)
5325 if (i != j && (*syms)[j].symbol != NULL
5326 && strncmp (name, SYMBOL_LINKAGE_NAME ((*syms)[j].symbol),
5327 name_len) == 0
5328 && block == (*syms)[j].block)
5329 (*syms)[j].symbol = NULL;
5330 }
5331 }
5332 if (is_new_style_renaming)
5333 {
5334 int j, k;
5335
5336 for (j = k = 0; j < syms->size (); j += 1)
5337 if ((*syms)[j].symbol != NULL)
5338 {
5339 (*syms)[k] = (*syms)[j];
5340 k += 1;
5341 }
5342 return k;
5343 }
5344
5345 /* Extract the function name associated to CURRENT_BLOCK.
5346 Abort if unable to do so. */
5347
5348 if (current_block == NULL)
5349 return syms->size ();
5350
5351 current_function = block_linkage_function (current_block);
5352 if (current_function == NULL)
5353 return syms->size ();
5354
5355 current_function_name = SYMBOL_LINKAGE_NAME (current_function);
5356 if (current_function_name == NULL)
5357 return syms->size ();
5358
5359 /* Check each of the symbols, and remove it from the list if it is
5360 a type corresponding to a renaming that is out of the scope of
5361 the current block. */
5362
5363 i = 0;
5364 while (i < syms->size ())
5365 {
5366 if (ada_parse_renaming ((*syms)[i].symbol, NULL, NULL, NULL)
5367 == ADA_OBJECT_RENAMING
5368 && old_renaming_is_invisible ((*syms)[i].symbol,
5369 current_function_name))
5370 syms->erase (syms->begin () + i);
5371 else
5372 i += 1;
5373 }
5374
5375 return syms->size ();
5376 }
5377
5378 /* Add to OBSTACKP all symbols from BLOCK (and its super-blocks)
5379 whose name and domain match NAME and DOMAIN respectively.
5380 If no match was found, then extend the search to "enclosing"
5381 routines (in other words, if we're inside a nested function,
5382 search the symbols defined inside the enclosing functions).
5383 If WILD_MATCH_P is nonzero, perform the naming matching in
5384 "wild" mode (see function "wild_match" for more info).
5385
5386 Note: This function assumes that OBSTACKP has 0 (zero) element in it. */
5387
5388 static void
5389 ada_add_local_symbols (struct obstack *obstackp,
5390 const lookup_name_info &lookup_name,
5391 const struct block *block, domain_enum domain)
5392 {
5393 int block_depth = 0;
5394
5395 while (block != NULL)
5396 {
5397 block_depth += 1;
5398 ada_add_block_symbols (obstackp, block, lookup_name, domain, NULL);
5399
5400 /* If we found a non-function match, assume that's the one. */
5401 if (is_nonfunction (defns_collected (obstackp, 0),
5402 num_defns_collected (obstackp)))
5403 return;
5404
5405 block = BLOCK_SUPERBLOCK (block);
5406 }
5407
5408 /* If no luck so far, try to find NAME as a local symbol in some lexically
5409 enclosing subprogram. */
5410 if (num_defns_collected (obstackp) == 0 && block_depth > 2)
5411 add_symbols_from_enclosing_procs (obstackp, lookup_name, domain);
5412 }
5413
5414 /* An object of this type is used as the user_data argument when
5415 calling the map_matching_symbols method. */
5416
5417 struct match_data
5418 {
5419 struct objfile *objfile;
5420 struct obstack *obstackp;
5421 struct symbol *arg_sym;
5422 int found_sym;
5423 };
5424
5425 /* A callback for add_nonlocal_symbols that adds SYM, found in BLOCK,
5426 to a list of symbols. DATA0 is a pointer to a struct match_data *
5427 containing the obstack that collects the symbol list, the file that SYM
5428 must come from, a flag indicating whether a non-argument symbol has
5429 been found in the current block, and the last argument symbol
5430 passed in SYM within the current block (if any). When SYM is null,
5431 marking the end of a block, the argument symbol is added if no
5432 other has been found. */
5433
5434 static int
5435 aux_add_nonlocal_symbols (const struct block *block, struct symbol *sym,
5436 void *data0)
5437 {
5438 struct match_data *data = (struct match_data *) data0;
5439
5440 if (sym == NULL)
5441 {
5442 if (!data->found_sym && data->arg_sym != NULL)
5443 add_defn_to_vec (data->obstackp,
5444 fixup_symbol_section (data->arg_sym, data->objfile),
5445 block);
5446 data->found_sym = 0;
5447 data->arg_sym = NULL;
5448 }
5449 else
5450 {
5451 if (SYMBOL_CLASS (sym) == LOC_UNRESOLVED)
5452 return 0;
5453 else if (SYMBOL_IS_ARGUMENT (sym))
5454 data->arg_sym = sym;
5455 else
5456 {
5457 data->found_sym = 1;
5458 add_defn_to_vec (data->obstackp,
5459 fixup_symbol_section (sym, data->objfile),
5460 block);
5461 }
5462 }
5463 return 0;
5464 }
5465
5466 /* Helper for add_nonlocal_symbols. Find symbols in DOMAIN which are
5467 targeted by renamings matching LOOKUP_NAME in BLOCK. Add these
5468 symbols to OBSTACKP. Return whether we found such symbols. */
5469
5470 static int
5471 ada_add_block_renamings (struct obstack *obstackp,
5472 const struct block *block,
5473 const lookup_name_info &lookup_name,
5474 domain_enum domain)
5475 {
5476 struct using_direct *renaming;
5477 int defns_mark = num_defns_collected (obstackp);
5478
5479 symbol_name_matcher_ftype *name_match
5480 = ada_get_symbol_name_matcher (lookup_name);
5481
5482 for (renaming = block_using (block);
5483 renaming != NULL;
5484 renaming = renaming->next)
5485 {
5486 const char *r_name;
5487
5488 /* Avoid infinite recursions: skip this renaming if we are actually
5489 already traversing it.
5490
5491 Currently, symbol lookup in Ada don't use the namespace machinery from
5492 C++/Fortran support: skip namespace imports that use them. */
5493 if (renaming->searched
5494 || (renaming->import_src != NULL
5495 && renaming->import_src[0] != '\0')
5496 || (renaming->import_dest != NULL
5497 && renaming->import_dest[0] != '\0'))
5498 continue;
5499 renaming->searched = 1;
5500
5501 /* TODO: here, we perform another name-based symbol lookup, which can
5502 pull its own multiple overloads. In theory, we should be able to do
5503 better in this case since, in DWARF, DW_AT_import is a DIE reference,
5504 not a simple name. But in order to do this, we would need to enhance
5505 the DWARF reader to associate a symbol to this renaming, instead of a
5506 name. So, for now, we do something simpler: re-use the C++/Fortran
5507 namespace machinery. */
5508 r_name = (renaming->alias != NULL
5509 ? renaming->alias
5510 : renaming->declaration);
5511 if (name_match (r_name, lookup_name, NULL))
5512 {
5513 lookup_name_info decl_lookup_name (renaming->declaration,
5514 lookup_name.match_type ());
5515 ada_add_all_symbols (obstackp, block, decl_lookup_name, domain,
5516 1, NULL);
5517 }
5518 renaming->searched = 0;
5519 }
5520 return num_defns_collected (obstackp) != defns_mark;
5521 }
5522
5523 /* Implements compare_names, but only applying the comparision using
5524 the given CASING. */
5525
5526 static int
5527 compare_names_with_case (const char *string1, const char *string2,
5528 enum case_sensitivity casing)
5529 {
5530 while (*string1 != '\0' && *string2 != '\0')
5531 {
5532 char c1, c2;
5533
5534 if (isspace (*string1) || isspace (*string2))
5535 return strcmp_iw_ordered (string1, string2);
5536
5537 if (casing == case_sensitive_off)
5538 {
5539 c1 = tolower (*string1);
5540 c2 = tolower (*string2);
5541 }
5542 else
5543 {
5544 c1 = *string1;
5545 c2 = *string2;
5546 }
5547 if (c1 != c2)
5548 break;
5549
5550 string1 += 1;
5551 string2 += 1;
5552 }
5553
5554 switch (*string1)
5555 {
5556 case '(':
5557 return strcmp_iw_ordered (string1, string2);
5558 case '_':
5559 if (*string2 == '\0')
5560 {
5561 if (is_name_suffix (string1))
5562 return 0;
5563 else
5564 return 1;
5565 }
5566 /* FALLTHROUGH */
5567 default:
5568 if (*string2 == '(')
5569 return strcmp_iw_ordered (string1, string2);
5570 else
5571 {
5572 if (casing == case_sensitive_off)
5573 return tolower (*string1) - tolower (*string2);
5574 else
5575 return *string1 - *string2;
5576 }
5577 }
5578 }
5579
5580 /* Compare STRING1 to STRING2, with results as for strcmp.
5581 Compatible with strcmp_iw_ordered in that...
5582
5583 strcmp_iw_ordered (STRING1, STRING2) <= 0
5584
5585 ... implies...
5586
5587 compare_names (STRING1, STRING2) <= 0
5588
5589 (they may differ as to what symbols compare equal). */
5590
5591 static int
5592 compare_names (const char *string1, const char *string2)
5593 {
5594 int result;
5595
5596 /* Similar to what strcmp_iw_ordered does, we need to perform
5597 a case-insensitive comparison first, and only resort to
5598 a second, case-sensitive, comparison if the first one was
5599 not sufficient to differentiate the two strings. */
5600
5601 result = compare_names_with_case (string1, string2, case_sensitive_off);
5602 if (result == 0)
5603 result = compare_names_with_case (string1, string2, case_sensitive_on);
5604
5605 return result;
5606 }
5607
5608 /* Convenience function to get at the Ada encoded lookup name for
5609 LOOKUP_NAME, as a C string. */
5610
5611 static const char *
5612 ada_lookup_name (const lookup_name_info &lookup_name)
5613 {
5614 return lookup_name.ada ().lookup_name ().c_str ();
5615 }
5616
5617 /* Add to OBSTACKP all non-local symbols whose name and domain match
5618 LOOKUP_NAME and DOMAIN respectively. The search is performed on
5619 GLOBAL_BLOCK symbols if GLOBAL is non-zero, or on STATIC_BLOCK
5620 symbols otherwise. */
5621
5622 static void
5623 add_nonlocal_symbols (struct obstack *obstackp,
5624 const lookup_name_info &lookup_name,
5625 domain_enum domain, int global)
5626 {
5627 struct match_data data;
5628
5629 memset (&data, 0, sizeof data);
5630 data.obstackp = obstackp;
5631
5632 bool is_wild_match = lookup_name.ada ().wild_match_p ();
5633
5634 for (objfile *objfile : current_program_space->objfiles ())
5635 {
5636 data.objfile = objfile;
5637
5638 if (is_wild_match)
5639 objfile->sf->qf->map_matching_symbols (objfile, lookup_name.name ().c_str (),
5640 domain, global,
5641 aux_add_nonlocal_symbols, &data,
5642 symbol_name_match_type::WILD,
5643 NULL);
5644 else
5645 objfile->sf->qf->map_matching_symbols (objfile, lookup_name.name ().c_str (),
5646 domain, global,
5647 aux_add_nonlocal_symbols, &data,
5648 symbol_name_match_type::FULL,
5649 compare_names);
5650
5651 for (compunit_symtab *cu : objfile->compunits ())
5652 {
5653 const struct block *global_block
5654 = BLOCKVECTOR_BLOCK (COMPUNIT_BLOCKVECTOR (cu), GLOBAL_BLOCK);
5655
5656 if (ada_add_block_renamings (obstackp, global_block, lookup_name,
5657 domain))
5658 data.found_sym = 1;
5659 }
5660 }
5661
5662 if (num_defns_collected (obstackp) == 0 && global && !is_wild_match)
5663 {
5664 const char *name = ada_lookup_name (lookup_name);
5665 std::string name1 = std::string ("<_ada_") + name + '>';
5666
5667 for (objfile *objfile : current_program_space->objfiles ())
5668 {
5669 data.objfile = objfile;
5670 objfile->sf->qf->map_matching_symbols (objfile, name1.c_str (),
5671 domain, global,
5672 aux_add_nonlocal_symbols,
5673 &data,
5674 symbol_name_match_type::FULL,
5675 compare_names);
5676 }
5677 }
5678 }
5679
5680 /* Find symbols in DOMAIN matching LOOKUP_NAME, in BLOCK and, if
5681 FULL_SEARCH is non-zero, enclosing scope and in global scopes,
5682 returning the number of matches. Add these to OBSTACKP.
5683
5684 When FULL_SEARCH is non-zero, any non-function/non-enumeral
5685 symbol match within the nest of blocks whose innermost member is BLOCK,
5686 is the one match returned (no other matches in that or
5687 enclosing blocks is returned). If there are any matches in or
5688 surrounding BLOCK, then these alone are returned.
5689
5690 Names prefixed with "standard__" are handled specially:
5691 "standard__" is first stripped off (by the lookup_name
5692 constructor), and only static and global symbols are searched.
5693
5694 If MADE_GLOBAL_LOOKUP_P is non-null, set it before return to whether we had
5695 to lookup global symbols. */
5696
5697 static void
5698 ada_add_all_symbols (struct obstack *obstackp,
5699 const struct block *block,
5700 const lookup_name_info &lookup_name,
5701 domain_enum domain,
5702 int full_search,
5703 int *made_global_lookup_p)
5704 {
5705 struct symbol *sym;
5706
5707 if (made_global_lookup_p)
5708 *made_global_lookup_p = 0;
5709
5710 /* Special case: If the user specifies a symbol name inside package
5711 Standard, do a non-wild matching of the symbol name without
5712 the "standard__" prefix. This was primarily introduced in order
5713 to allow the user to specifically access the standard exceptions
5714 using, for instance, Standard.Constraint_Error when Constraint_Error
5715 is ambiguous (due to the user defining its own Constraint_Error
5716 entity inside its program). */
5717 if (lookup_name.ada ().standard_p ())
5718 block = NULL;
5719
5720 /* Check the non-global symbols. If we have ANY match, then we're done. */
5721
5722 if (block != NULL)
5723 {
5724 if (full_search)
5725 ada_add_local_symbols (obstackp, lookup_name, block, domain);
5726 else
5727 {
5728 /* In the !full_search case we're are being called by
5729 ada_iterate_over_symbols, and we don't want to search
5730 superblocks. */
5731 ada_add_block_symbols (obstackp, block, lookup_name, domain, NULL);
5732 }
5733 if (num_defns_collected (obstackp) > 0 || !full_search)
5734 return;
5735 }
5736
5737 /* No non-global symbols found. Check our cache to see if we have
5738 already performed this search before. If we have, then return
5739 the same result. */
5740
5741 if (lookup_cached_symbol (ada_lookup_name (lookup_name),
5742 domain, &sym, &block))
5743 {
5744 if (sym != NULL)
5745 add_defn_to_vec (obstackp, sym, block);
5746 return;
5747 }
5748
5749 if (made_global_lookup_p)
5750 *made_global_lookup_p = 1;
5751
5752 /* Search symbols from all global blocks. */
5753
5754 add_nonlocal_symbols (obstackp, lookup_name, domain, 1);
5755
5756 /* Now add symbols from all per-file blocks if we've gotten no hits
5757 (not strictly correct, but perhaps better than an error). */
5758
5759 if (num_defns_collected (obstackp) == 0)
5760 add_nonlocal_symbols (obstackp, lookup_name, domain, 0);
5761 }
5762
5763 /* Find symbols in DOMAIN matching LOOKUP_NAME, in BLOCK and, if FULL_SEARCH
5764 is non-zero, enclosing scope and in global scopes, returning the number of
5765 matches.
5766 Fills *RESULTS with (SYM,BLOCK) tuples, indicating the symbols
5767 found and the blocks and symbol tables (if any) in which they were
5768 found.
5769
5770 When full_search is non-zero, any non-function/non-enumeral
5771 symbol match within the nest of blocks whose innermost member is BLOCK,
5772 is the one match returned (no other matches in that or
5773 enclosing blocks is returned). If there are any matches in or
5774 surrounding BLOCK, then these alone are returned.
5775
5776 Names prefixed with "standard__" are handled specially: "standard__"
5777 is first stripped off, and only static and global symbols are searched. */
5778
5779 static int
5780 ada_lookup_symbol_list_worker (const lookup_name_info &lookup_name,
5781 const struct block *block,
5782 domain_enum domain,
5783 std::vector<struct block_symbol> *results,
5784 int full_search)
5785 {
5786 int syms_from_global_search;
5787 int ndefns;
5788 auto_obstack obstack;
5789
5790 ada_add_all_symbols (&obstack, block, lookup_name,
5791 domain, full_search, &syms_from_global_search);
5792
5793 ndefns = num_defns_collected (&obstack);
5794
5795 struct block_symbol *base = defns_collected (&obstack, 1);
5796 for (int i = 0; i < ndefns; ++i)
5797 results->push_back (base[i]);
5798
5799 ndefns = remove_extra_symbols (results);
5800
5801 if (ndefns == 0 && full_search && syms_from_global_search)
5802 cache_symbol (ada_lookup_name (lookup_name), domain, NULL, NULL);
5803
5804 if (ndefns == 1 && full_search && syms_from_global_search)
5805 cache_symbol (ada_lookup_name (lookup_name), domain,
5806 (*results)[0].symbol, (*results)[0].block);
5807
5808 ndefns = remove_irrelevant_renamings (results, block);
5809
5810 return ndefns;
5811 }
5812
5813 /* Find symbols in DOMAIN matching NAME, in BLOCK and enclosing scope and
5814 in global scopes, returning the number of matches, and filling *RESULTS
5815 with (SYM,BLOCK) tuples.
5816
5817 See ada_lookup_symbol_list_worker for further details. */
5818
5819 int
5820 ada_lookup_symbol_list (const char *name, const struct block *block,
5821 domain_enum domain,
5822 std::vector<struct block_symbol> *results)
5823 {
5824 symbol_name_match_type name_match_type = name_match_type_from_name (name);
5825 lookup_name_info lookup_name (name, name_match_type);
5826
5827 return ada_lookup_symbol_list_worker (lookup_name, block, domain, results, 1);
5828 }
5829
5830 /* Implementation of the la_iterate_over_symbols method. */
5831
5832 static void
5833 ada_iterate_over_symbols
5834 (const struct block *block, const lookup_name_info &name,
5835 domain_enum domain,
5836 gdb::function_view<symbol_found_callback_ftype> callback)
5837 {
5838 int ndefs, i;
5839 std::vector<struct block_symbol> results;
5840
5841 ndefs = ada_lookup_symbol_list_worker (name, block, domain, &results, 0);
5842
5843 for (i = 0; i < ndefs; ++i)
5844 {
5845 if (!callback (&results[i]))
5846 break;
5847 }
5848 }
5849
5850 /* The result is as for ada_lookup_symbol_list with FULL_SEARCH set
5851 to 1, but choosing the first symbol found if there are multiple
5852 choices.
5853
5854 The result is stored in *INFO, which must be non-NULL.
5855 If no match is found, INFO->SYM is set to NULL. */
5856
5857 void
5858 ada_lookup_encoded_symbol (const char *name, const struct block *block,
5859 domain_enum domain,
5860 struct block_symbol *info)
5861 {
5862 /* Since we already have an encoded name, wrap it in '<>' to force a
5863 verbatim match. Otherwise, if the name happens to not look like
5864 an encoded name (because it doesn't include a "__"),
5865 ada_lookup_name_info would re-encode/fold it again, and that
5866 would e.g., incorrectly lowercase object renaming names like
5867 "R28b" -> "r28b". */
5868 std::string verbatim = std::string ("<") + name + '>';
5869
5870 gdb_assert (info != NULL);
5871 *info = ada_lookup_symbol (verbatim.c_str (), block, domain, NULL);
5872 }
5873
5874 /* Return a symbol in DOMAIN matching NAME, in BLOCK0 and enclosing
5875 scope and in global scopes, or NULL if none. NAME is folded and
5876 encoded first. Otherwise, the result is as for ada_lookup_symbol_list,
5877 choosing the first symbol if there are multiple choices.
5878 If IS_A_FIELD_OF_THIS is not NULL, it is set to zero. */
5879
5880 struct block_symbol
5881 ada_lookup_symbol (const char *name, const struct block *block0,
5882 domain_enum domain, int *is_a_field_of_this)
5883 {
5884 if (is_a_field_of_this != NULL)
5885 *is_a_field_of_this = 0;
5886
5887 std::vector<struct block_symbol> candidates;
5888 int n_candidates;
5889
5890 n_candidates = ada_lookup_symbol_list (name, block0, domain, &candidates);
5891
5892 if (n_candidates == 0)
5893 return {};
5894
5895 block_symbol info = candidates[0];
5896 info.symbol = fixup_symbol_section (info.symbol, NULL);
5897 return info;
5898 }
5899
5900 static struct block_symbol
5901 ada_lookup_symbol_nonlocal (const struct language_defn *langdef,
5902 const char *name,
5903 const struct block *block,
5904 const domain_enum domain)
5905 {
5906 struct block_symbol sym;
5907
5908 sym = ada_lookup_symbol (name, block_static_block (block), domain, NULL);
5909 if (sym.symbol != NULL)
5910 return sym;
5911
5912 /* If we haven't found a match at this point, try the primitive
5913 types. In other languages, this search is performed before
5914 searching for global symbols in order to short-circuit that
5915 global-symbol search if it happens that the name corresponds
5916 to a primitive type. But we cannot do the same in Ada, because
5917 it is perfectly legitimate for a program to declare a type which
5918 has the same name as a standard type. If looking up a type in
5919 that situation, we have traditionally ignored the primitive type
5920 in favor of user-defined types. This is why, unlike most other
5921 languages, we search the primitive types this late and only after
5922 having searched the global symbols without success. */
5923
5924 if (domain == VAR_DOMAIN)
5925 {
5926 struct gdbarch *gdbarch;
5927
5928 if (block == NULL)
5929 gdbarch = target_gdbarch ();
5930 else
5931 gdbarch = block_gdbarch (block);
5932 sym.symbol = language_lookup_primitive_type_as_symbol (langdef, gdbarch, name);
5933 if (sym.symbol != NULL)
5934 return sym;
5935 }
5936
5937 return {};
5938 }
5939
5940
5941 /* True iff STR is a possible encoded suffix of a normal Ada name
5942 that is to be ignored for matching purposes. Suffixes of parallel
5943 names (e.g., XVE) are not included here. Currently, the possible suffixes
5944 are given by any of the regular expressions:
5945
5946 [.$][0-9]+ [nested subprogram suffix, on platforms such as GNU/Linux]
5947 ___[0-9]+ [nested subprogram suffix, on platforms such as HP/UX]
5948 TKB [subprogram suffix for task bodies]
5949 _E[0-9]+[bs]$ [protected object entry suffixes]
5950 (X[nb]*)?((\$|__)[0-9](_?[0-9]+)|___(JM|LJM|X([FDBUP].*|R[^T]?)))?$
5951
5952 Also, any leading "__[0-9]+" sequence is skipped before the suffix
5953 match is performed. This sequence is used to differentiate homonyms,
5954 is an optional part of a valid name suffix. */
5955
5956 static int
5957 is_name_suffix (const char *str)
5958 {
5959 int k;
5960 const char *matching;
5961 const int len = strlen (str);
5962
5963 /* Skip optional leading __[0-9]+. */
5964
5965 if (len > 3 && str[0] == '_' && str[1] == '_' && isdigit (str[2]))
5966 {
5967 str += 3;
5968 while (isdigit (str[0]))
5969 str += 1;
5970 }
5971
5972 /* [.$][0-9]+ */
5973
5974 if (str[0] == '.' || str[0] == '$')
5975 {
5976 matching = str + 1;
5977 while (isdigit (matching[0]))
5978 matching += 1;
5979 if (matching[0] == '\0')
5980 return 1;
5981 }
5982
5983 /* ___[0-9]+ */
5984
5985 if (len > 3 && str[0] == '_' && str[1] == '_' && str[2] == '_')
5986 {
5987 matching = str + 3;
5988 while (isdigit (matching[0]))
5989 matching += 1;
5990 if (matching[0] == '\0')
5991 return 1;
5992 }
5993
5994 /* "TKB" suffixes are used for subprograms implementing task bodies. */
5995
5996 if (strcmp (str, "TKB") == 0)
5997 return 1;
5998
5999 #if 0
6000 /* FIXME: brobecker/2005-09-23: Protected Object subprograms end
6001 with a N at the end. Unfortunately, the compiler uses the same
6002 convention for other internal types it creates. So treating
6003 all entity names that end with an "N" as a name suffix causes
6004 some regressions. For instance, consider the case of an enumerated
6005 type. To support the 'Image attribute, it creates an array whose
6006 name ends with N.
6007 Having a single character like this as a suffix carrying some
6008 information is a bit risky. Perhaps we should change the encoding
6009 to be something like "_N" instead. In the meantime, do not do
6010 the following check. */
6011 /* Protected Object Subprograms */
6012 if (len == 1 && str [0] == 'N')
6013 return 1;
6014 #endif
6015
6016 /* _E[0-9]+[bs]$ */
6017 if (len > 3 && str[0] == '_' && str [1] == 'E' && isdigit (str[2]))
6018 {
6019 matching = str + 3;
6020 while (isdigit (matching[0]))
6021 matching += 1;
6022 if ((matching[0] == 'b' || matching[0] == 's')
6023 && matching [1] == '\0')
6024 return 1;
6025 }
6026
6027 /* ??? We should not modify STR directly, as we are doing below. This
6028 is fine in this case, but may become problematic later if we find
6029 that this alternative did not work, and want to try matching
6030 another one from the begining of STR. Since we modified it, we
6031 won't be able to find the begining of the string anymore! */
6032 if (str[0] == 'X')
6033 {
6034 str += 1;
6035 while (str[0] != '_' && str[0] != '\0')
6036 {
6037 if (str[0] != 'n' && str[0] != 'b')
6038 return 0;
6039 str += 1;
6040 }
6041 }
6042
6043 if (str[0] == '\000')
6044 return 1;
6045
6046 if (str[0] == '_')
6047 {
6048 if (str[1] != '_' || str[2] == '\000')
6049 return 0;
6050 if (str[2] == '_')
6051 {
6052 if (strcmp (str + 3, "JM") == 0)
6053 return 1;
6054 /* FIXME: brobecker/2004-09-30: GNAT will soon stop using
6055 the LJM suffix in favor of the JM one. But we will
6056 still accept LJM as a valid suffix for a reasonable
6057 amount of time, just to allow ourselves to debug programs
6058 compiled using an older version of GNAT. */
6059 if (strcmp (str + 3, "LJM") == 0)
6060 return 1;
6061 if (str[3] != 'X')
6062 return 0;
6063 if (str[4] == 'F' || str[4] == 'D' || str[4] == 'B'
6064 || str[4] == 'U' || str[4] == 'P')
6065 return 1;
6066 if (str[4] == 'R' && str[5] != 'T')
6067 return 1;
6068 return 0;
6069 }
6070 if (!isdigit (str[2]))
6071 return 0;
6072 for (k = 3; str[k] != '\0'; k += 1)
6073 if (!isdigit (str[k]) && str[k] != '_')
6074 return 0;
6075 return 1;
6076 }
6077 if (str[0] == '$' && isdigit (str[1]))
6078 {
6079 for (k = 2; str[k] != '\0'; k += 1)
6080 if (!isdigit (str[k]) && str[k] != '_')
6081 return 0;
6082 return 1;
6083 }
6084 return 0;
6085 }
6086
6087 /* Return non-zero if the string starting at NAME and ending before
6088 NAME_END contains no capital letters. */
6089
6090 static int
6091 is_valid_name_for_wild_match (const char *name0)
6092 {
6093 const char *decoded_name = ada_decode (name0);
6094 int i;
6095
6096 /* If the decoded name starts with an angle bracket, it means that
6097 NAME0 does not follow the GNAT encoding format. It should then
6098 not be allowed as a possible wild match. */
6099 if (decoded_name[0] == '<')
6100 return 0;
6101
6102 for (i=0; decoded_name[i] != '\0'; i++)
6103 if (isalpha (decoded_name[i]) && !islower (decoded_name[i]))
6104 return 0;
6105
6106 return 1;
6107 }
6108
6109 /* Advance *NAMEP to next occurrence of TARGET0 in the string NAME0
6110 that could start a simple name. Assumes that *NAMEP points into
6111 the string beginning at NAME0. */
6112
6113 static int
6114 advance_wild_match (const char **namep, const char *name0, int target0)
6115 {
6116 const char *name = *namep;
6117
6118 while (1)
6119 {
6120 int t0, t1;
6121
6122 t0 = *name;
6123 if (t0 == '_')
6124 {
6125 t1 = name[1];
6126 if ((t1 >= 'a' && t1 <= 'z') || (t1 >= '0' && t1 <= '9'))
6127 {
6128 name += 1;
6129 if (name == name0 + 5 && startswith (name0, "_ada"))
6130 break;
6131 else
6132 name += 1;
6133 }
6134 else if (t1 == '_' && ((name[2] >= 'a' && name[2] <= 'z')
6135 || name[2] == target0))
6136 {
6137 name += 2;
6138 break;
6139 }
6140 else
6141 return 0;
6142 }
6143 else if ((t0 >= 'a' && t0 <= 'z') || (t0 >= '0' && t0 <= '9'))
6144 name += 1;
6145 else
6146 return 0;
6147 }
6148
6149 *namep = name;
6150 return 1;
6151 }
6152
6153 /* Return true iff NAME encodes a name of the form prefix.PATN.
6154 Ignores any informational suffixes of NAME (i.e., for which
6155 is_name_suffix is true). Assumes that PATN is a lower-cased Ada
6156 simple name. */
6157
6158 static bool
6159 wild_match (const char *name, const char *patn)
6160 {
6161 const char *p;
6162 const char *name0 = name;
6163
6164 while (1)
6165 {
6166 const char *match = name;
6167
6168 if (*name == *patn)
6169 {
6170 for (name += 1, p = patn + 1; *p != '\0'; name += 1, p += 1)
6171 if (*p != *name)
6172 break;
6173 if (*p == '\0' && is_name_suffix (name))
6174 return match == name0 || is_valid_name_for_wild_match (name0);
6175
6176 if (name[-1] == '_')
6177 name -= 1;
6178 }
6179 if (!advance_wild_match (&name, name0, *patn))
6180 return false;
6181 }
6182 }
6183
6184 /* Returns true iff symbol name SYM_NAME matches SEARCH_NAME, ignoring
6185 any trailing suffixes that encode debugging information or leading
6186 _ada_ on SYM_NAME (see is_name_suffix commentary for the debugging
6187 information that is ignored). */
6188
6189 static bool
6190 full_match (const char *sym_name, const char *search_name)
6191 {
6192 size_t search_name_len = strlen (search_name);
6193
6194 if (strncmp (sym_name, search_name, search_name_len) == 0
6195 && is_name_suffix (sym_name + search_name_len))
6196 return true;
6197
6198 if (startswith (sym_name, "_ada_")
6199 && strncmp (sym_name + 5, search_name, search_name_len) == 0
6200 && is_name_suffix (sym_name + search_name_len + 5))
6201 return true;
6202
6203 return false;
6204 }
6205
6206 /* Add symbols from BLOCK matching LOOKUP_NAME in DOMAIN to vector
6207 *defn_symbols, updating the list of symbols in OBSTACKP (if
6208 necessary). OBJFILE is the section containing BLOCK. */
6209
6210 static void
6211 ada_add_block_symbols (struct obstack *obstackp,
6212 const struct block *block,
6213 const lookup_name_info &lookup_name,
6214 domain_enum domain, struct objfile *objfile)
6215 {
6216 struct block_iterator iter;
6217 /* A matching argument symbol, if any. */
6218 struct symbol *arg_sym;
6219 /* Set true when we find a matching non-argument symbol. */
6220 int found_sym;
6221 struct symbol *sym;
6222
6223 arg_sym = NULL;
6224 found_sym = 0;
6225 for (sym = block_iter_match_first (block, lookup_name, &iter);
6226 sym != NULL;
6227 sym = block_iter_match_next (lookup_name, &iter))
6228 {
6229 if (symbol_matches_domain (SYMBOL_LANGUAGE (sym),
6230 SYMBOL_DOMAIN (sym), domain))
6231 {
6232 if (SYMBOL_CLASS (sym) != LOC_UNRESOLVED)
6233 {
6234 if (SYMBOL_IS_ARGUMENT (sym))
6235 arg_sym = sym;
6236 else
6237 {
6238 found_sym = 1;
6239 add_defn_to_vec (obstackp,
6240 fixup_symbol_section (sym, objfile),
6241 block);
6242 }
6243 }
6244 }
6245 }
6246
6247 /* Handle renamings. */
6248
6249 if (ada_add_block_renamings (obstackp, block, lookup_name, domain))
6250 found_sym = 1;
6251
6252 if (!found_sym && arg_sym != NULL)
6253 {
6254 add_defn_to_vec (obstackp,
6255 fixup_symbol_section (arg_sym, objfile),
6256 block);
6257 }
6258
6259 if (!lookup_name.ada ().wild_match_p ())
6260 {
6261 arg_sym = NULL;
6262 found_sym = 0;
6263 const std::string &ada_lookup_name = lookup_name.ada ().lookup_name ();
6264 const char *name = ada_lookup_name.c_str ();
6265 size_t name_len = ada_lookup_name.size ();
6266
6267 ALL_BLOCK_SYMBOLS (block, iter, sym)
6268 {
6269 if (symbol_matches_domain (SYMBOL_LANGUAGE (sym),
6270 SYMBOL_DOMAIN (sym), domain))
6271 {
6272 int cmp;
6273
6274 cmp = (int) '_' - (int) SYMBOL_LINKAGE_NAME (sym)[0];
6275 if (cmp == 0)
6276 {
6277 cmp = !startswith (SYMBOL_LINKAGE_NAME (sym), "_ada_");
6278 if (cmp == 0)
6279 cmp = strncmp (name, SYMBOL_LINKAGE_NAME (sym) + 5,
6280 name_len);
6281 }
6282
6283 if (cmp == 0
6284 && is_name_suffix (SYMBOL_LINKAGE_NAME (sym) + name_len + 5))
6285 {
6286 if (SYMBOL_CLASS (sym) != LOC_UNRESOLVED)
6287 {
6288 if (SYMBOL_IS_ARGUMENT (sym))
6289 arg_sym = sym;
6290 else
6291 {
6292 found_sym = 1;
6293 add_defn_to_vec (obstackp,
6294 fixup_symbol_section (sym, objfile),
6295 block);
6296 }
6297 }
6298 }
6299 }
6300 }
6301
6302 /* NOTE: This really shouldn't be needed for _ada_ symbols.
6303 They aren't parameters, right? */
6304 if (!found_sym && arg_sym != NULL)
6305 {
6306 add_defn_to_vec (obstackp,
6307 fixup_symbol_section (arg_sym, objfile),
6308 block);
6309 }
6310 }
6311 }
6312 \f
6313
6314 /* Symbol Completion */
6315
6316 /* See symtab.h. */
6317
6318 bool
6319 ada_lookup_name_info::matches
6320 (const char *sym_name,
6321 symbol_name_match_type match_type,
6322 completion_match_result *comp_match_res) const
6323 {
6324 bool match = false;
6325 const char *text = m_encoded_name.c_str ();
6326 size_t text_len = m_encoded_name.size ();
6327
6328 /* First, test against the fully qualified name of the symbol. */
6329
6330 if (strncmp (sym_name, text, text_len) == 0)
6331 match = true;
6332
6333 if (match && !m_encoded_p)
6334 {
6335 /* One needed check before declaring a positive match is to verify
6336 that iff we are doing a verbatim match, the decoded version
6337 of the symbol name starts with '<'. Otherwise, this symbol name
6338 is not a suitable completion. */
6339 const char *sym_name_copy = sym_name;
6340 bool has_angle_bracket;
6341
6342 sym_name = ada_decode (sym_name);
6343 has_angle_bracket = (sym_name[0] == '<');
6344 match = (has_angle_bracket == m_verbatim_p);
6345 sym_name = sym_name_copy;
6346 }
6347
6348 if (match && !m_verbatim_p)
6349 {
6350 /* When doing non-verbatim match, another check that needs to
6351 be done is to verify that the potentially matching symbol name
6352 does not include capital letters, because the ada-mode would
6353 not be able to understand these symbol names without the
6354 angle bracket notation. */
6355 const char *tmp;
6356
6357 for (tmp = sym_name; *tmp != '\0' && !isupper (*tmp); tmp++);
6358 if (*tmp != '\0')
6359 match = false;
6360 }
6361
6362 /* Second: Try wild matching... */
6363
6364 if (!match && m_wild_match_p)
6365 {
6366 /* Since we are doing wild matching, this means that TEXT
6367 may represent an unqualified symbol name. We therefore must
6368 also compare TEXT against the unqualified name of the symbol. */
6369 sym_name = ada_unqualified_name (ada_decode (sym_name));
6370
6371 if (strncmp (sym_name, text, text_len) == 0)
6372 match = true;
6373 }
6374
6375 /* Finally: If we found a match, prepare the result to return. */
6376
6377 if (!match)
6378 return false;
6379
6380 if (comp_match_res != NULL)
6381 {
6382 std::string &match_str = comp_match_res->match.storage ();
6383
6384 if (!m_encoded_p)
6385 match_str = ada_decode (sym_name);
6386 else
6387 {
6388 if (m_verbatim_p)
6389 match_str = add_angle_brackets (sym_name);
6390 else
6391 match_str = sym_name;
6392
6393 }
6394
6395 comp_match_res->set_match (match_str.c_str ());
6396 }
6397
6398 return true;
6399 }
6400
6401 /* Add the list of possible symbol names completing TEXT to TRACKER.
6402 WORD is the entire command on which completion is made. */
6403
6404 static void
6405 ada_collect_symbol_completion_matches (completion_tracker &tracker,
6406 complete_symbol_mode mode,
6407 symbol_name_match_type name_match_type,
6408 const char *text, const char *word,
6409 enum type_code code)
6410 {
6411 struct symbol *sym;
6412 const struct block *b, *surrounding_static_block = 0;
6413 struct block_iterator iter;
6414
6415 gdb_assert (code == TYPE_CODE_UNDEF);
6416
6417 lookup_name_info lookup_name (text, name_match_type, true);
6418
6419 /* First, look at the partial symtab symbols. */
6420 expand_symtabs_matching (NULL,
6421 lookup_name,
6422 NULL,
6423 NULL,
6424 ALL_DOMAIN);
6425
6426 /* At this point scan through the misc symbol vectors and add each
6427 symbol you find to the list. Eventually we want to ignore
6428 anything that isn't a text symbol (everything else will be
6429 handled by the psymtab code above). */
6430
6431 for (objfile *objfile : current_program_space->objfiles ())
6432 {
6433 for (minimal_symbol *msymbol : objfile->msymbols ())
6434 {
6435 QUIT;
6436
6437 if (completion_skip_symbol (mode, msymbol))
6438 continue;
6439
6440 language symbol_language = MSYMBOL_LANGUAGE (msymbol);
6441
6442 /* Ada minimal symbols won't have their language set to Ada. If
6443 we let completion_list_add_name compare using the
6444 default/C-like matcher, then when completing e.g., symbols in a
6445 package named "pck", we'd match internal Ada symbols like
6446 "pckS", which are invalid in an Ada expression, unless you wrap
6447 them in '<' '>' to request a verbatim match.
6448
6449 Unfortunately, some Ada encoded names successfully demangle as
6450 C++ symbols (using an old mangling scheme), such as "name__2Xn"
6451 -> "Xn::name(void)" and thus some Ada minimal symbols end up
6452 with the wrong language set. Paper over that issue here. */
6453 if (symbol_language == language_auto
6454 || symbol_language == language_cplus)
6455 symbol_language = language_ada;
6456
6457 completion_list_add_name (tracker,
6458 symbol_language,
6459 MSYMBOL_LINKAGE_NAME (msymbol),
6460 lookup_name, text, word);
6461 }
6462 }
6463
6464 /* Search upwards from currently selected frame (so that we can
6465 complete on local vars. */
6466
6467 for (b = get_selected_block (0); b != NULL; b = BLOCK_SUPERBLOCK (b))
6468 {
6469 if (!BLOCK_SUPERBLOCK (b))
6470 surrounding_static_block = b; /* For elmin of dups */
6471
6472 ALL_BLOCK_SYMBOLS (b, iter, sym)
6473 {
6474 if (completion_skip_symbol (mode, sym))
6475 continue;
6476
6477 completion_list_add_name (tracker,
6478 SYMBOL_LANGUAGE (sym),
6479 SYMBOL_LINKAGE_NAME (sym),
6480 lookup_name, text, word);
6481 }
6482 }
6483
6484 /* Go through the symtabs and check the externs and statics for
6485 symbols which match. */
6486
6487 for (objfile *objfile : current_program_space->objfiles ())
6488 {
6489 for (compunit_symtab *s : objfile->compunits ())
6490 {
6491 QUIT;
6492 b = BLOCKVECTOR_BLOCK (COMPUNIT_BLOCKVECTOR (s), GLOBAL_BLOCK);
6493 ALL_BLOCK_SYMBOLS (b, iter, sym)
6494 {
6495 if (completion_skip_symbol (mode, sym))
6496 continue;
6497
6498 completion_list_add_name (tracker,
6499 SYMBOL_LANGUAGE (sym),
6500 SYMBOL_LINKAGE_NAME (sym),
6501 lookup_name, text, word);
6502 }
6503 }
6504 }
6505
6506 for (objfile *objfile : current_program_space->objfiles ())
6507 {
6508 for (compunit_symtab *s : objfile->compunits ())
6509 {
6510 QUIT;
6511 b = BLOCKVECTOR_BLOCK (COMPUNIT_BLOCKVECTOR (s), STATIC_BLOCK);
6512 /* Don't do this block twice. */
6513 if (b == surrounding_static_block)
6514 continue;
6515 ALL_BLOCK_SYMBOLS (b, iter, sym)
6516 {
6517 if (completion_skip_symbol (mode, sym))
6518 continue;
6519
6520 completion_list_add_name (tracker,
6521 SYMBOL_LANGUAGE (sym),
6522 SYMBOL_LINKAGE_NAME (sym),
6523 lookup_name, text, word);
6524 }
6525 }
6526 }
6527 }
6528
6529 /* Field Access */
6530
6531 /* Return non-zero if TYPE is a pointer to the GNAT dispatch table used
6532 for tagged types. */
6533
6534 static int
6535 ada_is_dispatch_table_ptr_type (struct type *type)
6536 {
6537 const char *name;
6538
6539 if (TYPE_CODE (type) != TYPE_CODE_PTR)
6540 return 0;
6541
6542 name = TYPE_NAME (TYPE_TARGET_TYPE (type));
6543 if (name == NULL)
6544 return 0;
6545
6546 return (strcmp (name, "ada__tags__dispatch_table") == 0);
6547 }
6548
6549 /* Return non-zero if TYPE is an interface tag. */
6550
6551 static int
6552 ada_is_interface_tag (struct type *type)
6553 {
6554 const char *name = TYPE_NAME (type);
6555
6556 if (name == NULL)
6557 return 0;
6558
6559 return (strcmp (name, "ada__tags__interface_tag") == 0);
6560 }
6561
6562 /* True if field number FIELD_NUM in struct or union type TYPE is supposed
6563 to be invisible to users. */
6564
6565 int
6566 ada_is_ignored_field (struct type *type, int field_num)
6567 {
6568 if (field_num < 0 || field_num > TYPE_NFIELDS (type))
6569 return 1;
6570
6571 /* Check the name of that field. */
6572 {
6573 const char *name = TYPE_FIELD_NAME (type, field_num);
6574
6575 /* Anonymous field names should not be printed.
6576 brobecker/2007-02-20: I don't think this can actually happen
6577 but we don't want to print the value of annonymous fields anyway. */
6578 if (name == NULL)
6579 return 1;
6580
6581 /* Normally, fields whose name start with an underscore ("_")
6582 are fields that have been internally generated by the compiler,
6583 and thus should not be printed. The "_parent" field is special,
6584 however: This is a field internally generated by the compiler
6585 for tagged types, and it contains the components inherited from
6586 the parent type. This field should not be printed as is, but
6587 should not be ignored either. */
6588 if (name[0] == '_' && !startswith (name, "_parent"))
6589 return 1;
6590 }
6591
6592 /* If this is the dispatch table of a tagged type or an interface tag,
6593 then ignore. */
6594 if (ada_is_tagged_type (type, 1)
6595 && (ada_is_dispatch_table_ptr_type (TYPE_FIELD_TYPE (type, field_num))
6596 || ada_is_interface_tag (TYPE_FIELD_TYPE (type, field_num))))
6597 return 1;
6598
6599 /* Not a special field, so it should not be ignored. */
6600 return 0;
6601 }
6602
6603 /* True iff TYPE has a tag field. If REFOK, then TYPE may also be a
6604 pointer or reference type whose ultimate target has a tag field. */
6605
6606 int
6607 ada_is_tagged_type (struct type *type, int refok)
6608 {
6609 return (ada_lookup_struct_elt_type (type, "_tag", refok, 1) != NULL);
6610 }
6611
6612 /* True iff TYPE represents the type of X'Tag */
6613
6614 int
6615 ada_is_tag_type (struct type *type)
6616 {
6617 type = ada_check_typedef (type);
6618
6619 if (type == NULL || TYPE_CODE (type) != TYPE_CODE_PTR)
6620 return 0;
6621 else
6622 {
6623 const char *name = ada_type_name (TYPE_TARGET_TYPE (type));
6624
6625 return (name != NULL
6626 && strcmp (name, "ada__tags__dispatch_table") == 0);
6627 }
6628 }
6629
6630 /* The type of the tag on VAL. */
6631
6632 struct type *
6633 ada_tag_type (struct value *val)
6634 {
6635 return ada_lookup_struct_elt_type (value_type (val), "_tag", 1, 0);
6636 }
6637
6638 /* Return 1 if TAG follows the old scheme for Ada tags (used for Ada 95,
6639 retired at Ada 05). */
6640
6641 static int
6642 is_ada95_tag (struct value *tag)
6643 {
6644 return ada_value_struct_elt (tag, "tsd", 1) != NULL;
6645 }
6646
6647 /* The value of the tag on VAL. */
6648
6649 struct value *
6650 ada_value_tag (struct value *val)
6651 {
6652 return ada_value_struct_elt (val, "_tag", 0);
6653 }
6654
6655 /* The value of the tag on the object of type TYPE whose contents are
6656 saved at VALADDR, if it is non-null, or is at memory address
6657 ADDRESS. */
6658
6659 static struct value *
6660 value_tag_from_contents_and_address (struct type *type,
6661 const gdb_byte *valaddr,
6662 CORE_ADDR address)
6663 {
6664 int tag_byte_offset;
6665 struct type *tag_type;
6666
6667 if (find_struct_field ("_tag", type, 0, &tag_type, &tag_byte_offset,
6668 NULL, NULL, NULL))
6669 {
6670 const gdb_byte *valaddr1 = ((valaddr == NULL)
6671 ? NULL
6672 : valaddr + tag_byte_offset);
6673 CORE_ADDR address1 = (address == 0) ? 0 : address + tag_byte_offset;
6674
6675 return value_from_contents_and_address (tag_type, valaddr1, address1);
6676 }
6677 return NULL;
6678 }
6679
6680 static struct type *
6681 type_from_tag (struct value *tag)
6682 {
6683 const char *type_name = ada_tag_name (tag);
6684
6685 if (type_name != NULL)
6686 return ada_find_any_type (ada_encode (type_name));
6687 return NULL;
6688 }
6689
6690 /* Given a value OBJ of a tagged type, return a value of this
6691 type at the base address of the object. The base address, as
6692 defined in Ada.Tags, it is the address of the primary tag of
6693 the object, and therefore where the field values of its full
6694 view can be fetched. */
6695
6696 struct value *
6697 ada_tag_value_at_base_address (struct value *obj)
6698 {
6699 struct value *val;
6700 LONGEST offset_to_top = 0;
6701 struct type *ptr_type, *obj_type;
6702 struct value *tag;
6703 CORE_ADDR base_address;
6704
6705 obj_type = value_type (obj);
6706
6707 /* It is the responsability of the caller to deref pointers. */
6708
6709 if (TYPE_CODE (obj_type) == TYPE_CODE_PTR
6710 || TYPE_CODE (obj_type) == TYPE_CODE_REF)
6711 return obj;
6712
6713 tag = ada_value_tag (obj);
6714 if (!tag)
6715 return obj;
6716
6717 /* Base addresses only appeared with Ada 05 and multiple inheritance. */
6718
6719 if (is_ada95_tag (tag))
6720 return obj;
6721
6722 ptr_type = language_lookup_primitive_type
6723 (language_def (language_ada), target_gdbarch(), "storage_offset");
6724 ptr_type = lookup_pointer_type (ptr_type);
6725 val = value_cast (ptr_type, tag);
6726 if (!val)
6727 return obj;
6728
6729 /* It is perfectly possible that an exception be raised while
6730 trying to determine the base address, just like for the tag;
6731 see ada_tag_name for more details. We do not print the error
6732 message for the same reason. */
6733
6734 try
6735 {
6736 offset_to_top = value_as_long (value_ind (value_ptradd (val, -2)));
6737 }
6738
6739 catch (const gdb_exception_error &e)
6740 {
6741 return obj;
6742 }
6743
6744 /* If offset is null, nothing to do. */
6745
6746 if (offset_to_top == 0)
6747 return obj;
6748
6749 /* -1 is a special case in Ada.Tags; however, what should be done
6750 is not quite clear from the documentation. So do nothing for
6751 now. */
6752
6753 if (offset_to_top == -1)
6754 return obj;
6755
6756 /* OFFSET_TO_TOP used to be a positive value to be subtracted
6757 from the base address. This was however incompatible with
6758 C++ dispatch table: C++ uses a *negative* value to *add*
6759 to the base address. Ada's convention has therefore been
6760 changed in GNAT 19.0w 20171023: since then, C++ and Ada
6761 use the same convention. Here, we support both cases by
6762 checking the sign of OFFSET_TO_TOP. */
6763
6764 if (offset_to_top > 0)
6765 offset_to_top = -offset_to_top;
6766
6767 base_address = value_address (obj) + offset_to_top;
6768 tag = value_tag_from_contents_and_address (obj_type, NULL, base_address);
6769
6770 /* Make sure that we have a proper tag at the new address.
6771 Otherwise, offset_to_top is bogus (which can happen when
6772 the object is not initialized yet). */
6773
6774 if (!tag)
6775 return obj;
6776
6777 obj_type = type_from_tag (tag);
6778
6779 if (!obj_type)
6780 return obj;
6781
6782 return value_from_contents_and_address (obj_type, NULL, base_address);
6783 }
6784
6785 /* Return the "ada__tags__type_specific_data" type. */
6786
6787 static struct type *
6788 ada_get_tsd_type (struct inferior *inf)
6789 {
6790 struct ada_inferior_data *data = get_ada_inferior_data (inf);
6791
6792 if (data->tsd_type == 0)
6793 data->tsd_type = ada_find_any_type ("ada__tags__type_specific_data");
6794 return data->tsd_type;
6795 }
6796
6797 /* Return the TSD (type-specific data) associated to the given TAG.
6798 TAG is assumed to be the tag of a tagged-type entity.
6799
6800 May return NULL if we are unable to get the TSD. */
6801
6802 static struct value *
6803 ada_get_tsd_from_tag (struct value *tag)
6804 {
6805 struct value *val;
6806 struct type *type;
6807
6808 /* First option: The TSD is simply stored as a field of our TAG.
6809 Only older versions of GNAT would use this format, but we have
6810 to test it first, because there are no visible markers for
6811 the current approach except the absence of that field. */
6812
6813 val = ada_value_struct_elt (tag, "tsd", 1);
6814 if (val)
6815 return val;
6816
6817 /* Try the second representation for the dispatch table (in which
6818 there is no explicit 'tsd' field in the referent of the tag pointer,
6819 and instead the tsd pointer is stored just before the dispatch
6820 table. */
6821
6822 type = ada_get_tsd_type (current_inferior());
6823 if (type == NULL)
6824 return NULL;
6825 type = lookup_pointer_type (lookup_pointer_type (type));
6826 val = value_cast (type, tag);
6827 if (val == NULL)
6828 return NULL;
6829 return value_ind (value_ptradd (val, -1));
6830 }
6831
6832 /* Given the TSD of a tag (type-specific data), return a string
6833 containing the name of the associated type.
6834
6835 The returned value is good until the next call. May return NULL
6836 if we are unable to determine the tag name. */
6837
6838 static char *
6839 ada_tag_name_from_tsd (struct value *tsd)
6840 {
6841 static char name[1024];
6842 char *p;
6843 struct value *val;
6844
6845 val = ada_value_struct_elt (tsd, "expanded_name", 1);
6846 if (val == NULL)
6847 return NULL;
6848 read_memory_string (value_as_address (val), name, sizeof (name) - 1);
6849 for (p = name; *p != '\0'; p += 1)
6850 if (isalpha (*p))
6851 *p = tolower (*p);
6852 return name;
6853 }
6854
6855 /* The type name of the dynamic type denoted by the 'tag value TAG, as
6856 a C string.
6857
6858 Return NULL if the TAG is not an Ada tag, or if we were unable to
6859 determine the name of that tag. The result is good until the next
6860 call. */
6861
6862 const char *
6863 ada_tag_name (struct value *tag)
6864 {
6865 char *name = NULL;
6866
6867 if (!ada_is_tag_type (value_type (tag)))
6868 return NULL;
6869
6870 /* It is perfectly possible that an exception be raised while trying
6871 to determine the TAG's name, even under normal circumstances:
6872 The associated variable may be uninitialized or corrupted, for
6873 instance. We do not let any exception propagate past this point.
6874 instead we return NULL.
6875
6876 We also do not print the error message either (which often is very
6877 low-level (Eg: "Cannot read memory at 0x[...]"), but instead let
6878 the caller print a more meaningful message if necessary. */
6879 try
6880 {
6881 struct value *tsd = ada_get_tsd_from_tag (tag);
6882
6883 if (tsd != NULL)
6884 name = ada_tag_name_from_tsd (tsd);
6885 }
6886 catch (const gdb_exception_error &e)
6887 {
6888 }
6889
6890 return name;
6891 }
6892
6893 /* The parent type of TYPE, or NULL if none. */
6894
6895 struct type *
6896 ada_parent_type (struct type *type)
6897 {
6898 int i;
6899
6900 type = ada_check_typedef (type);
6901
6902 if (type == NULL || TYPE_CODE (type) != TYPE_CODE_STRUCT)
6903 return NULL;
6904
6905 for (i = 0; i < TYPE_NFIELDS (type); i += 1)
6906 if (ada_is_parent_field (type, i))
6907 {
6908 struct type *parent_type = TYPE_FIELD_TYPE (type, i);
6909
6910 /* If the _parent field is a pointer, then dereference it. */
6911 if (TYPE_CODE (parent_type) == TYPE_CODE_PTR)
6912 parent_type = TYPE_TARGET_TYPE (parent_type);
6913 /* If there is a parallel XVS type, get the actual base type. */
6914 parent_type = ada_get_base_type (parent_type);
6915
6916 return ada_check_typedef (parent_type);
6917 }
6918
6919 return NULL;
6920 }
6921
6922 /* True iff field number FIELD_NUM of structure type TYPE contains the
6923 parent-type (inherited) fields of a derived type. Assumes TYPE is
6924 a structure type with at least FIELD_NUM+1 fields. */
6925
6926 int
6927 ada_is_parent_field (struct type *type, int field_num)
6928 {
6929 const char *name = TYPE_FIELD_NAME (ada_check_typedef (type), field_num);
6930
6931 return (name != NULL
6932 && (startswith (name, "PARENT")
6933 || startswith (name, "_parent")));
6934 }
6935
6936 /* True iff field number FIELD_NUM of structure type TYPE is a
6937 transparent wrapper field (which should be silently traversed when doing
6938 field selection and flattened when printing). Assumes TYPE is a
6939 structure type with at least FIELD_NUM+1 fields. Such fields are always
6940 structures. */
6941
6942 int
6943 ada_is_wrapper_field (struct type *type, int field_num)
6944 {
6945 const char *name = TYPE_FIELD_NAME (type, field_num);
6946
6947 if (name != NULL && strcmp (name, "RETVAL") == 0)
6948 {
6949 /* This happens in functions with "out" or "in out" parameters
6950 which are passed by copy. For such functions, GNAT describes
6951 the function's return type as being a struct where the return
6952 value is in a field called RETVAL, and where the other "out"
6953 or "in out" parameters are fields of that struct. This is not
6954 a wrapper. */
6955 return 0;
6956 }
6957
6958 return (name != NULL
6959 && (startswith (name, "PARENT")
6960 || strcmp (name, "REP") == 0
6961 || startswith (name, "_parent")
6962 || name[0] == 'S' || name[0] == 'R' || name[0] == 'O'));
6963 }
6964
6965 /* True iff field number FIELD_NUM of structure or union type TYPE
6966 is a variant wrapper. Assumes TYPE is a structure type with at least
6967 FIELD_NUM+1 fields. */
6968
6969 int
6970 ada_is_variant_part (struct type *type, int field_num)
6971 {
6972 /* Only Ada types are eligible. */
6973 if (!ADA_TYPE_P (type))
6974 return 0;
6975
6976 struct type *field_type = TYPE_FIELD_TYPE (type, field_num);
6977
6978 return (TYPE_CODE (field_type) == TYPE_CODE_UNION
6979 || (is_dynamic_field (type, field_num)
6980 && (TYPE_CODE (TYPE_TARGET_TYPE (field_type))
6981 == TYPE_CODE_UNION)));
6982 }
6983
6984 /* Assuming that VAR_TYPE is a variant wrapper (type of the variant part)
6985 whose discriminants are contained in the record type OUTER_TYPE,
6986 returns the type of the controlling discriminant for the variant.
6987 May return NULL if the type could not be found. */
6988
6989 struct type *
6990 ada_variant_discrim_type (struct type *var_type, struct type *outer_type)
6991 {
6992 const char *name = ada_variant_discrim_name (var_type);
6993
6994 return ada_lookup_struct_elt_type (outer_type, name, 1, 1);
6995 }
6996
6997 /* Assuming that TYPE is the type of a variant wrapper, and FIELD_NUM is a
6998 valid field number within it, returns 1 iff field FIELD_NUM of TYPE
6999 represents a 'when others' clause; otherwise 0. */
7000
7001 int
7002 ada_is_others_clause (struct type *type, int field_num)
7003 {
7004 const char *name = TYPE_FIELD_NAME (type, field_num);
7005
7006 return (name != NULL && name[0] == 'O');
7007 }
7008
7009 /* Assuming that TYPE0 is the type of the variant part of a record,
7010 returns the name of the discriminant controlling the variant.
7011 The value is valid until the next call to ada_variant_discrim_name. */
7012
7013 const char *
7014 ada_variant_discrim_name (struct type *type0)
7015 {
7016 static char *result = NULL;
7017 static size_t result_len = 0;
7018 struct type *type;
7019 const char *name;
7020 const char *discrim_end;
7021 const char *discrim_start;
7022
7023 if (TYPE_CODE (type0) == TYPE_CODE_PTR)
7024 type = TYPE_TARGET_TYPE (type0);
7025 else
7026 type = type0;
7027
7028 name = ada_type_name (type);
7029
7030 if (name == NULL || name[0] == '\000')
7031 return "";
7032
7033 for (discrim_end = name + strlen (name) - 6; discrim_end != name;
7034 discrim_end -= 1)
7035 {
7036 if (startswith (discrim_end, "___XVN"))
7037 break;
7038 }
7039 if (discrim_end == name)
7040 return "";
7041
7042 for (discrim_start = discrim_end; discrim_start != name + 3;
7043 discrim_start -= 1)
7044 {
7045 if (discrim_start == name + 1)
7046 return "";
7047 if ((discrim_start > name + 3
7048 && startswith (discrim_start - 3, "___"))
7049 || discrim_start[-1] == '.')
7050 break;
7051 }
7052
7053 GROW_VECT (result, result_len, discrim_end - discrim_start + 1);
7054 strncpy (result, discrim_start, discrim_end - discrim_start);
7055 result[discrim_end - discrim_start] = '\0';
7056 return result;
7057 }
7058
7059 /* Scan STR for a subtype-encoded number, beginning at position K.
7060 Put the position of the character just past the number scanned in
7061 *NEW_K, if NEW_K!=NULL. Put the scanned number in *R, if R!=NULL.
7062 Return 1 if there was a valid number at the given position, and 0
7063 otherwise. A "subtype-encoded" number consists of the absolute value
7064 in decimal, followed by the letter 'm' to indicate a negative number.
7065 Assumes 0m does not occur. */
7066
7067 int
7068 ada_scan_number (const char str[], int k, LONGEST * R, int *new_k)
7069 {
7070 ULONGEST RU;
7071
7072 if (!isdigit (str[k]))
7073 return 0;
7074
7075 /* Do it the hard way so as not to make any assumption about
7076 the relationship of unsigned long (%lu scan format code) and
7077 LONGEST. */
7078 RU = 0;
7079 while (isdigit (str[k]))
7080 {
7081 RU = RU * 10 + (str[k] - '0');
7082 k += 1;
7083 }
7084
7085 if (str[k] == 'm')
7086 {
7087 if (R != NULL)
7088 *R = (-(LONGEST) (RU - 1)) - 1;
7089 k += 1;
7090 }
7091 else if (R != NULL)
7092 *R = (LONGEST) RU;
7093
7094 /* NOTE on the above: Technically, C does not say what the results of
7095 - (LONGEST) RU or (LONGEST) -RU are for RU == largest positive
7096 number representable as a LONGEST (although either would probably work
7097 in most implementations). When RU>0, the locution in the then branch
7098 above is always equivalent to the negative of RU. */
7099
7100 if (new_k != NULL)
7101 *new_k = k;
7102 return 1;
7103 }
7104
7105 /* Assuming that TYPE is a variant part wrapper type (a VARIANTS field),
7106 and FIELD_NUM is a valid field number within it, returns 1 iff VAL is
7107 in the range encoded by field FIELD_NUM of TYPE; otherwise 0. */
7108
7109 int
7110 ada_in_variant (LONGEST val, struct type *type, int field_num)
7111 {
7112 const char *name = TYPE_FIELD_NAME (type, field_num);
7113 int p;
7114
7115 p = 0;
7116 while (1)
7117 {
7118 switch (name[p])
7119 {
7120 case '\0':
7121 return 0;
7122 case 'S':
7123 {
7124 LONGEST W;
7125
7126 if (!ada_scan_number (name, p + 1, &W, &p))
7127 return 0;
7128 if (val == W)
7129 return 1;
7130 break;
7131 }
7132 case 'R':
7133 {
7134 LONGEST L, U;
7135
7136 if (!ada_scan_number (name, p + 1, &L, &p)
7137 || name[p] != 'T' || !ada_scan_number (name, p + 1, &U, &p))
7138 return 0;
7139 if (val >= L && val <= U)
7140 return 1;
7141 break;
7142 }
7143 case 'O':
7144 return 1;
7145 default:
7146 return 0;
7147 }
7148 }
7149 }
7150
7151 /* FIXME: Lots of redundancy below. Try to consolidate. */
7152
7153 /* Given a value ARG1 (offset by OFFSET bytes) of a struct or union type
7154 ARG_TYPE, extract and return the value of one of its (non-static)
7155 fields. FIELDNO says which field. Differs from value_primitive_field
7156 only in that it can handle packed values of arbitrary type. */
7157
7158 static struct value *
7159 ada_value_primitive_field (struct value *arg1, int offset, int fieldno,
7160 struct type *arg_type)
7161 {
7162 struct type *type;
7163
7164 arg_type = ada_check_typedef (arg_type);
7165 type = TYPE_FIELD_TYPE (arg_type, fieldno);
7166
7167 /* Handle packed fields. It might be that the field is not packed
7168 relative to its containing structure, but the structure itself is
7169 packed; in this case we must take the bit-field path. */
7170 if (TYPE_FIELD_BITSIZE (arg_type, fieldno) != 0 || value_bitpos (arg1) != 0)
7171 {
7172 int bit_pos = TYPE_FIELD_BITPOS (arg_type, fieldno);
7173 int bit_size = TYPE_FIELD_BITSIZE (arg_type, fieldno);
7174
7175 return ada_value_primitive_packed_val (arg1, value_contents (arg1),
7176 offset + bit_pos / 8,
7177 bit_pos % 8, bit_size, type);
7178 }
7179 else
7180 return value_primitive_field (arg1, offset, fieldno, arg_type);
7181 }
7182
7183 /* Find field with name NAME in object of type TYPE. If found,
7184 set the following for each argument that is non-null:
7185 - *FIELD_TYPE_P to the field's type;
7186 - *BYTE_OFFSET_P to OFFSET + the byte offset of the field within
7187 an object of that type;
7188 - *BIT_OFFSET_P to the bit offset modulo byte size of the field;
7189 - *BIT_SIZE_P to its size in bits if the field is packed, and
7190 0 otherwise;
7191 If INDEX_P is non-null, increment *INDEX_P by the number of source-visible
7192 fields up to but not including the desired field, or by the total
7193 number of fields if not found. A NULL value of NAME never
7194 matches; the function just counts visible fields in this case.
7195
7196 Notice that we need to handle when a tagged record hierarchy
7197 has some components with the same name, like in this scenario:
7198
7199 type Top_T is tagged record
7200 N : Integer := 1;
7201 U : Integer := 974;
7202 A : Integer := 48;
7203 end record;
7204
7205 type Middle_T is new Top.Top_T with record
7206 N : Character := 'a';
7207 C : Integer := 3;
7208 end record;
7209
7210 type Bottom_T is new Middle.Middle_T with record
7211 N : Float := 4.0;
7212 C : Character := '5';
7213 X : Integer := 6;
7214 A : Character := 'J';
7215 end record;
7216
7217 Let's say we now have a variable declared and initialized as follow:
7218
7219 TC : Top_A := new Bottom_T;
7220
7221 And then we use this variable to call this function
7222
7223 procedure Assign (Obj: in out Top_T; TV : Integer);
7224
7225 as follow:
7226
7227 Assign (Top_T (B), 12);
7228
7229 Now, we're in the debugger, and we're inside that procedure
7230 then and we want to print the value of obj.c:
7231
7232 Usually, the tagged record or one of the parent type owns the
7233 component to print and there's no issue but in this particular
7234 case, what does it mean to ask for Obj.C? Since the actual
7235 type for object is type Bottom_T, it could mean two things: type
7236 component C from the Middle_T view, but also component C from
7237 Bottom_T. So in that "undefined" case, when the component is
7238 not found in the non-resolved type (which includes all the
7239 components of the parent type), then resolve it and see if we
7240 get better luck once expanded.
7241
7242 In the case of homonyms in the derived tagged type, we don't
7243 guaranty anything, and pick the one that's easiest for us
7244 to program.
7245
7246 Returns 1 if found, 0 otherwise. */
7247
7248 static int
7249 find_struct_field (const char *name, struct type *type, int offset,
7250 struct type **field_type_p,
7251 int *byte_offset_p, int *bit_offset_p, int *bit_size_p,
7252 int *index_p)
7253 {
7254 int i;
7255 int parent_offset = -1;
7256
7257 type = ada_check_typedef (type);
7258
7259 if (field_type_p != NULL)
7260 *field_type_p = NULL;
7261 if (byte_offset_p != NULL)
7262 *byte_offset_p = 0;
7263 if (bit_offset_p != NULL)
7264 *bit_offset_p = 0;
7265 if (bit_size_p != NULL)
7266 *bit_size_p = 0;
7267
7268 for (i = 0; i < TYPE_NFIELDS (type); i += 1)
7269 {
7270 int bit_pos = TYPE_FIELD_BITPOS (type, i);
7271 int fld_offset = offset + bit_pos / 8;
7272 const char *t_field_name = TYPE_FIELD_NAME (type, i);
7273
7274 if (t_field_name == NULL)
7275 continue;
7276
7277 else if (ada_is_parent_field (type, i))
7278 {
7279 /* This is a field pointing us to the parent type of a tagged
7280 type. As hinted in this function's documentation, we give
7281 preference to fields in the current record first, so what
7282 we do here is just record the index of this field before
7283 we skip it. If it turns out we couldn't find our field
7284 in the current record, then we'll get back to it and search
7285 inside it whether the field might exist in the parent. */
7286
7287 parent_offset = i;
7288 continue;
7289 }
7290
7291 else if (name != NULL && field_name_match (t_field_name, name))
7292 {
7293 int bit_size = TYPE_FIELD_BITSIZE (type, i);
7294
7295 if (field_type_p != NULL)
7296 *field_type_p = TYPE_FIELD_TYPE (type, i);
7297 if (byte_offset_p != NULL)
7298 *byte_offset_p = fld_offset;
7299 if (bit_offset_p != NULL)
7300 *bit_offset_p = bit_pos % 8;
7301 if (bit_size_p != NULL)
7302 *bit_size_p = bit_size;
7303 return 1;
7304 }
7305 else if (ada_is_wrapper_field (type, i))
7306 {
7307 if (find_struct_field (name, TYPE_FIELD_TYPE (type, i), fld_offset,
7308 field_type_p, byte_offset_p, bit_offset_p,
7309 bit_size_p, index_p))
7310 return 1;
7311 }
7312 else if (ada_is_variant_part (type, i))
7313 {
7314 /* PNH: Wait. Do we ever execute this section, or is ARG always of
7315 fixed type?? */
7316 int j;
7317 struct type *field_type
7318 = ada_check_typedef (TYPE_FIELD_TYPE (type, i));
7319
7320 for (j = 0; j < TYPE_NFIELDS (field_type); j += 1)
7321 {
7322 if (find_struct_field (name, TYPE_FIELD_TYPE (field_type, j),
7323 fld_offset
7324 + TYPE_FIELD_BITPOS (field_type, j) / 8,
7325 field_type_p, byte_offset_p,
7326 bit_offset_p, bit_size_p, index_p))
7327 return 1;
7328 }
7329 }
7330 else if (index_p != NULL)
7331 *index_p += 1;
7332 }
7333
7334 /* Field not found so far. If this is a tagged type which
7335 has a parent, try finding that field in the parent now. */
7336
7337 if (parent_offset != -1)
7338 {
7339 int bit_pos = TYPE_FIELD_BITPOS (type, parent_offset);
7340 int fld_offset = offset + bit_pos / 8;
7341
7342 if (find_struct_field (name, TYPE_FIELD_TYPE (type, parent_offset),
7343 fld_offset, field_type_p, byte_offset_p,
7344 bit_offset_p, bit_size_p, index_p))
7345 return 1;
7346 }
7347
7348 return 0;
7349 }
7350
7351 /* Number of user-visible fields in record type TYPE. */
7352
7353 static int
7354 num_visible_fields (struct type *type)
7355 {
7356 int n;
7357
7358 n = 0;
7359 find_struct_field (NULL, type, 0, NULL, NULL, NULL, NULL, &n);
7360 return n;
7361 }
7362
7363 /* Look for a field NAME in ARG. Adjust the address of ARG by OFFSET bytes,
7364 and search in it assuming it has (class) type TYPE.
7365 If found, return value, else return NULL.
7366
7367 Searches recursively through wrapper fields (e.g., '_parent').
7368
7369 In the case of homonyms in the tagged types, please refer to the
7370 long explanation in find_struct_field's function documentation. */
7371
7372 static struct value *
7373 ada_search_struct_field (const char *name, struct value *arg, int offset,
7374 struct type *type)
7375 {
7376 int i;
7377 int parent_offset = -1;
7378
7379 type = ada_check_typedef (type);
7380 for (i = 0; i < TYPE_NFIELDS (type); i += 1)
7381 {
7382 const char *t_field_name = TYPE_FIELD_NAME (type, i);
7383
7384 if (t_field_name == NULL)
7385 continue;
7386
7387 else if (ada_is_parent_field (type, i))
7388 {
7389 /* This is a field pointing us to the parent type of a tagged
7390 type. As hinted in this function's documentation, we give
7391 preference to fields in the current record first, so what
7392 we do here is just record the index of this field before
7393 we skip it. If it turns out we couldn't find our field
7394 in the current record, then we'll get back to it and search
7395 inside it whether the field might exist in the parent. */
7396
7397 parent_offset = i;
7398 continue;
7399 }
7400
7401 else if (field_name_match (t_field_name, name))
7402 return ada_value_primitive_field (arg, offset, i, type);
7403
7404 else if (ada_is_wrapper_field (type, i))
7405 {
7406 struct value *v = /* Do not let indent join lines here. */
7407 ada_search_struct_field (name, arg,
7408 offset + TYPE_FIELD_BITPOS (type, i) / 8,
7409 TYPE_FIELD_TYPE (type, i));
7410
7411 if (v != NULL)
7412 return v;
7413 }
7414
7415 else if (ada_is_variant_part (type, i))
7416 {
7417 /* PNH: Do we ever get here? See find_struct_field. */
7418 int j;
7419 struct type *field_type = ada_check_typedef (TYPE_FIELD_TYPE (type,
7420 i));
7421 int var_offset = offset + TYPE_FIELD_BITPOS (type, i) / 8;
7422
7423 for (j = 0; j < TYPE_NFIELDS (field_type); j += 1)
7424 {
7425 struct value *v = ada_search_struct_field /* Force line
7426 break. */
7427 (name, arg,
7428 var_offset + TYPE_FIELD_BITPOS (field_type, j) / 8,
7429 TYPE_FIELD_TYPE (field_type, j));
7430
7431 if (v != NULL)
7432 return v;
7433 }
7434 }
7435 }
7436
7437 /* Field not found so far. If this is a tagged type which
7438 has a parent, try finding that field in the parent now. */
7439
7440 if (parent_offset != -1)
7441 {
7442 struct value *v = ada_search_struct_field (
7443 name, arg, offset + TYPE_FIELD_BITPOS (type, parent_offset) / 8,
7444 TYPE_FIELD_TYPE (type, parent_offset));
7445
7446 if (v != NULL)
7447 return v;
7448 }
7449
7450 return NULL;
7451 }
7452
7453 static struct value *ada_index_struct_field_1 (int *, struct value *,
7454 int, struct type *);
7455
7456
7457 /* Return field #INDEX in ARG, where the index is that returned by
7458 * find_struct_field through its INDEX_P argument. Adjust the address
7459 * of ARG by OFFSET bytes, and search in it assuming it has (class) type TYPE.
7460 * If found, return value, else return NULL. */
7461
7462 static struct value *
7463 ada_index_struct_field (int index, struct value *arg, int offset,
7464 struct type *type)
7465 {
7466 return ada_index_struct_field_1 (&index, arg, offset, type);
7467 }
7468
7469
7470 /* Auxiliary function for ada_index_struct_field. Like
7471 * ada_index_struct_field, but takes index from *INDEX_P and modifies
7472 * *INDEX_P. */
7473
7474 static struct value *
7475 ada_index_struct_field_1 (int *index_p, struct value *arg, int offset,
7476 struct type *type)
7477 {
7478 int i;
7479 type = ada_check_typedef (type);
7480
7481 for (i = 0; i < TYPE_NFIELDS (type); i += 1)
7482 {
7483 if (TYPE_FIELD_NAME (type, i) == NULL)
7484 continue;
7485 else if (ada_is_wrapper_field (type, i))
7486 {
7487 struct value *v = /* Do not let indent join lines here. */
7488 ada_index_struct_field_1 (index_p, arg,
7489 offset + TYPE_FIELD_BITPOS (type, i) / 8,
7490 TYPE_FIELD_TYPE (type, i));
7491
7492 if (v != NULL)
7493 return v;
7494 }
7495
7496 else if (ada_is_variant_part (type, i))
7497 {
7498 /* PNH: Do we ever get here? See ada_search_struct_field,
7499 find_struct_field. */
7500 error (_("Cannot assign this kind of variant record"));
7501 }
7502 else if (*index_p == 0)
7503 return ada_value_primitive_field (arg, offset, i, type);
7504 else
7505 *index_p -= 1;
7506 }
7507 return NULL;
7508 }
7509
7510 /* Given ARG, a value of type (pointer or reference to a)*
7511 structure/union, extract the component named NAME from the ultimate
7512 target structure/union and return it as a value with its
7513 appropriate type.
7514
7515 The routine searches for NAME among all members of the structure itself
7516 and (recursively) among all members of any wrapper members
7517 (e.g., '_parent').
7518
7519 If NO_ERR, then simply return NULL in case of error, rather than
7520 calling error. */
7521
7522 struct value *
7523 ada_value_struct_elt (struct value *arg, const char *name, int no_err)
7524 {
7525 struct type *t, *t1;
7526 struct value *v;
7527 int check_tag;
7528
7529 v = NULL;
7530 t1 = t = ada_check_typedef (value_type (arg));
7531 if (TYPE_CODE (t) == TYPE_CODE_REF)
7532 {
7533 t1 = TYPE_TARGET_TYPE (t);
7534 if (t1 == NULL)
7535 goto BadValue;
7536 t1 = ada_check_typedef (t1);
7537 if (TYPE_CODE (t1) == TYPE_CODE_PTR)
7538 {
7539 arg = coerce_ref (arg);
7540 t = t1;
7541 }
7542 }
7543
7544 while (TYPE_CODE (t) == TYPE_CODE_PTR)
7545 {
7546 t1 = TYPE_TARGET_TYPE (t);
7547 if (t1 == NULL)
7548 goto BadValue;
7549 t1 = ada_check_typedef (t1);
7550 if (TYPE_CODE (t1) == TYPE_CODE_PTR)
7551 {
7552 arg = value_ind (arg);
7553 t = t1;
7554 }
7555 else
7556 break;
7557 }
7558
7559 if (TYPE_CODE (t1) != TYPE_CODE_STRUCT && TYPE_CODE (t1) != TYPE_CODE_UNION)
7560 goto BadValue;
7561
7562 if (t1 == t)
7563 v = ada_search_struct_field (name, arg, 0, t);
7564 else
7565 {
7566 int bit_offset, bit_size, byte_offset;
7567 struct type *field_type;
7568 CORE_ADDR address;
7569
7570 if (TYPE_CODE (t) == TYPE_CODE_PTR)
7571 address = value_address (ada_value_ind (arg));
7572 else
7573 address = value_address (ada_coerce_ref (arg));
7574
7575 /* Check to see if this is a tagged type. We also need to handle
7576 the case where the type is a reference to a tagged type, but
7577 we have to be careful to exclude pointers to tagged types.
7578 The latter should be shown as usual (as a pointer), whereas
7579 a reference should mostly be transparent to the user. */
7580
7581 if (ada_is_tagged_type (t1, 0)
7582 || (TYPE_CODE (t1) == TYPE_CODE_REF
7583 && ada_is_tagged_type (TYPE_TARGET_TYPE (t1), 0)))
7584 {
7585 /* We first try to find the searched field in the current type.
7586 If not found then let's look in the fixed type. */
7587
7588 if (!find_struct_field (name, t1, 0,
7589 &field_type, &byte_offset, &bit_offset,
7590 &bit_size, NULL))
7591 check_tag = 1;
7592 else
7593 check_tag = 0;
7594 }
7595 else
7596 check_tag = 0;
7597
7598 /* Convert to fixed type in all cases, so that we have proper
7599 offsets to each field in unconstrained record types. */
7600 t1 = ada_to_fixed_type (ada_get_base_type (t1), NULL,
7601 address, NULL, check_tag);
7602
7603 if (find_struct_field (name, t1, 0,
7604 &field_type, &byte_offset, &bit_offset,
7605 &bit_size, NULL))
7606 {
7607 if (bit_size != 0)
7608 {
7609 if (TYPE_CODE (t) == TYPE_CODE_REF)
7610 arg = ada_coerce_ref (arg);
7611 else
7612 arg = ada_value_ind (arg);
7613 v = ada_value_primitive_packed_val (arg, NULL, byte_offset,
7614 bit_offset, bit_size,
7615 field_type);
7616 }
7617 else
7618 v = value_at_lazy (field_type, address + byte_offset);
7619 }
7620 }
7621
7622 if (v != NULL || no_err)
7623 return v;
7624 else
7625 error (_("There is no member named %s."), name);
7626
7627 BadValue:
7628 if (no_err)
7629 return NULL;
7630 else
7631 error (_("Attempt to extract a component of "
7632 "a value that is not a record."));
7633 }
7634
7635 /* Return a string representation of type TYPE. */
7636
7637 static std::string
7638 type_as_string (struct type *type)
7639 {
7640 string_file tmp_stream;
7641
7642 type_print (type, "", &tmp_stream, -1);
7643
7644 return std::move (tmp_stream.string ());
7645 }
7646
7647 /* Given a type TYPE, look up the type of the component of type named NAME.
7648 If DISPP is non-null, add its byte displacement from the beginning of a
7649 structure (pointed to by a value) of type TYPE to *DISPP (does not
7650 work for packed fields).
7651
7652 Matches any field whose name has NAME as a prefix, possibly
7653 followed by "___".
7654
7655 TYPE can be either a struct or union. If REFOK, TYPE may also
7656 be a (pointer or reference)+ to a struct or union, and the
7657 ultimate target type will be searched.
7658
7659 Looks recursively into variant clauses and parent types.
7660
7661 In the case of homonyms in the tagged types, please refer to the
7662 long explanation in find_struct_field's function documentation.
7663
7664 If NOERR is nonzero, return NULL if NAME is not suitably defined or
7665 TYPE is not a type of the right kind. */
7666
7667 static struct type *
7668 ada_lookup_struct_elt_type (struct type *type, const char *name, int refok,
7669 int noerr)
7670 {
7671 int i;
7672 int parent_offset = -1;
7673
7674 if (name == NULL)
7675 goto BadName;
7676
7677 if (refok && type != NULL)
7678 while (1)
7679 {
7680 type = ada_check_typedef (type);
7681 if (TYPE_CODE (type) != TYPE_CODE_PTR
7682 && TYPE_CODE (type) != TYPE_CODE_REF)
7683 break;
7684 type = TYPE_TARGET_TYPE (type);
7685 }
7686
7687 if (type == NULL
7688 || (TYPE_CODE (type) != TYPE_CODE_STRUCT
7689 && TYPE_CODE (type) != TYPE_CODE_UNION))
7690 {
7691 if (noerr)
7692 return NULL;
7693
7694 error (_("Type %s is not a structure or union type"),
7695 type != NULL ? type_as_string (type).c_str () : _("(null)"));
7696 }
7697
7698 type = to_static_fixed_type (type);
7699
7700 for (i = 0; i < TYPE_NFIELDS (type); i += 1)
7701 {
7702 const char *t_field_name = TYPE_FIELD_NAME (type, i);
7703 struct type *t;
7704
7705 if (t_field_name == NULL)
7706 continue;
7707
7708 else if (ada_is_parent_field (type, i))
7709 {
7710 /* This is a field pointing us to the parent type of a tagged
7711 type. As hinted in this function's documentation, we give
7712 preference to fields in the current record first, so what
7713 we do here is just record the index of this field before
7714 we skip it. If it turns out we couldn't find our field
7715 in the current record, then we'll get back to it and search
7716 inside it whether the field might exist in the parent. */
7717
7718 parent_offset = i;
7719 continue;
7720 }
7721
7722 else if (field_name_match (t_field_name, name))
7723 return TYPE_FIELD_TYPE (type, i);
7724
7725 else if (ada_is_wrapper_field (type, i))
7726 {
7727 t = ada_lookup_struct_elt_type (TYPE_FIELD_TYPE (type, i), name,
7728 0, 1);
7729 if (t != NULL)
7730 return t;
7731 }
7732
7733 else if (ada_is_variant_part (type, i))
7734 {
7735 int j;
7736 struct type *field_type = ada_check_typedef (TYPE_FIELD_TYPE (type,
7737 i));
7738
7739 for (j = TYPE_NFIELDS (field_type) - 1; j >= 0; j -= 1)
7740 {
7741 /* FIXME pnh 2008/01/26: We check for a field that is
7742 NOT wrapped in a struct, since the compiler sometimes
7743 generates these for unchecked variant types. Revisit
7744 if the compiler changes this practice. */
7745 const char *v_field_name = TYPE_FIELD_NAME (field_type, j);
7746
7747 if (v_field_name != NULL
7748 && field_name_match (v_field_name, name))
7749 t = TYPE_FIELD_TYPE (field_type, j);
7750 else
7751 t = ada_lookup_struct_elt_type (TYPE_FIELD_TYPE (field_type,
7752 j),
7753 name, 0, 1);
7754
7755 if (t != NULL)
7756 return t;
7757 }
7758 }
7759
7760 }
7761
7762 /* Field not found so far. If this is a tagged type which
7763 has a parent, try finding that field in the parent now. */
7764
7765 if (parent_offset != -1)
7766 {
7767 struct type *t;
7768
7769 t = ada_lookup_struct_elt_type (TYPE_FIELD_TYPE (type, parent_offset),
7770 name, 0, 1);
7771 if (t != NULL)
7772 return t;
7773 }
7774
7775 BadName:
7776 if (!noerr)
7777 {
7778 const char *name_str = name != NULL ? name : _("<null>");
7779
7780 error (_("Type %s has no component named %s"),
7781 type_as_string (type).c_str (), name_str);
7782 }
7783
7784 return NULL;
7785 }
7786
7787 /* Assuming that VAR_TYPE is the type of a variant part of a record (a union),
7788 within a value of type OUTER_TYPE, return true iff VAR_TYPE
7789 represents an unchecked union (that is, the variant part of a
7790 record that is named in an Unchecked_Union pragma). */
7791
7792 static int
7793 is_unchecked_variant (struct type *var_type, struct type *outer_type)
7794 {
7795 const char *discrim_name = ada_variant_discrim_name (var_type);
7796
7797 return (ada_lookup_struct_elt_type (outer_type, discrim_name, 0, 1) == NULL);
7798 }
7799
7800
7801 /* Assuming that VAR_TYPE is the type of a variant part of a record (a union),
7802 within a value of type OUTER_TYPE that is stored in GDB at
7803 OUTER_VALADDR, determine which variant clause (field number in VAR_TYPE,
7804 numbering from 0) is applicable. Returns -1 if none are. */
7805
7806 int
7807 ada_which_variant_applies (struct type *var_type, struct type *outer_type,
7808 const gdb_byte *outer_valaddr)
7809 {
7810 int others_clause;
7811 int i;
7812 const char *discrim_name = ada_variant_discrim_name (var_type);
7813 struct value *outer;
7814 struct value *discrim;
7815 LONGEST discrim_val;
7816
7817 /* Using plain value_from_contents_and_address here causes problems
7818 because we will end up trying to resolve a type that is currently
7819 being constructed. */
7820 outer = value_from_contents_and_address_unresolved (outer_type,
7821 outer_valaddr, 0);
7822 discrim = ada_value_struct_elt (outer, discrim_name, 1);
7823 if (discrim == NULL)
7824 return -1;
7825 discrim_val = value_as_long (discrim);
7826
7827 others_clause = -1;
7828 for (i = 0; i < TYPE_NFIELDS (var_type); i += 1)
7829 {
7830 if (ada_is_others_clause (var_type, i))
7831 others_clause = i;
7832 else if (ada_in_variant (discrim_val, var_type, i))
7833 return i;
7834 }
7835
7836 return others_clause;
7837 }
7838 \f
7839
7840
7841 /* Dynamic-Sized Records */
7842
7843 /* Strategy: The type ostensibly attached to a value with dynamic size
7844 (i.e., a size that is not statically recorded in the debugging
7845 data) does not accurately reflect the size or layout of the value.
7846 Our strategy is to convert these values to values with accurate,
7847 conventional types that are constructed on the fly. */
7848
7849 /* There is a subtle and tricky problem here. In general, we cannot
7850 determine the size of dynamic records without its data. However,
7851 the 'struct value' data structure, which GDB uses to represent
7852 quantities in the inferior process (the target), requires the size
7853 of the type at the time of its allocation in order to reserve space
7854 for GDB's internal copy of the data. That's why the
7855 'to_fixed_xxx_type' routines take (target) addresses as parameters,
7856 rather than struct value*s.
7857
7858 However, GDB's internal history variables ($1, $2, etc.) are
7859 struct value*s containing internal copies of the data that are not, in
7860 general, the same as the data at their corresponding addresses in
7861 the target. Fortunately, the types we give to these values are all
7862 conventional, fixed-size types (as per the strategy described
7863 above), so that we don't usually have to perform the
7864 'to_fixed_xxx_type' conversions to look at their values.
7865 Unfortunately, there is one exception: if one of the internal
7866 history variables is an array whose elements are unconstrained
7867 records, then we will need to create distinct fixed types for each
7868 element selected. */
7869
7870 /* The upshot of all of this is that many routines take a (type, host
7871 address, target address) triple as arguments to represent a value.
7872 The host address, if non-null, is supposed to contain an internal
7873 copy of the relevant data; otherwise, the program is to consult the
7874 target at the target address. */
7875
7876 /* Assuming that VAL0 represents a pointer value, the result of
7877 dereferencing it. Differs from value_ind in its treatment of
7878 dynamic-sized types. */
7879
7880 struct value *
7881 ada_value_ind (struct value *val0)
7882 {
7883 struct value *val = value_ind (val0);
7884
7885 if (ada_is_tagged_type (value_type (val), 0))
7886 val = ada_tag_value_at_base_address (val);
7887
7888 return ada_to_fixed_value (val);
7889 }
7890
7891 /* The value resulting from dereferencing any "reference to"
7892 qualifiers on VAL0. */
7893
7894 static struct value *
7895 ada_coerce_ref (struct value *val0)
7896 {
7897 if (TYPE_CODE (value_type (val0)) == TYPE_CODE_REF)
7898 {
7899 struct value *val = val0;
7900
7901 val = coerce_ref (val);
7902
7903 if (ada_is_tagged_type (value_type (val), 0))
7904 val = ada_tag_value_at_base_address (val);
7905
7906 return ada_to_fixed_value (val);
7907 }
7908 else
7909 return val0;
7910 }
7911
7912 /* Return OFF rounded upward if necessary to a multiple of
7913 ALIGNMENT (a power of 2). */
7914
7915 static unsigned int
7916 align_value (unsigned int off, unsigned int alignment)
7917 {
7918 return (off + alignment - 1) & ~(alignment - 1);
7919 }
7920
7921 /* Return the bit alignment required for field #F of template type TYPE. */
7922
7923 static unsigned int
7924 field_alignment (struct type *type, int f)
7925 {
7926 const char *name = TYPE_FIELD_NAME (type, f);
7927 int len;
7928 int align_offset;
7929
7930 /* The field name should never be null, unless the debugging information
7931 is somehow malformed. In this case, we assume the field does not
7932 require any alignment. */
7933 if (name == NULL)
7934 return 1;
7935
7936 len = strlen (name);
7937
7938 if (!isdigit (name[len - 1]))
7939 return 1;
7940
7941 if (isdigit (name[len - 2]))
7942 align_offset = len - 2;
7943 else
7944 align_offset = len - 1;
7945
7946 if (align_offset < 7 || !startswith (name + align_offset - 6, "___XV"))
7947 return TARGET_CHAR_BIT;
7948
7949 return atoi (name + align_offset) * TARGET_CHAR_BIT;
7950 }
7951
7952 /* Find a typedef or tag symbol named NAME. Ignores ambiguity. */
7953
7954 static struct symbol *
7955 ada_find_any_type_symbol (const char *name)
7956 {
7957 struct symbol *sym;
7958
7959 sym = standard_lookup (name, get_selected_block (NULL), VAR_DOMAIN);
7960 if (sym != NULL && SYMBOL_CLASS (sym) == LOC_TYPEDEF)
7961 return sym;
7962
7963 sym = standard_lookup (name, NULL, STRUCT_DOMAIN);
7964 return sym;
7965 }
7966
7967 /* Find a type named NAME. Ignores ambiguity. This routine will look
7968 solely for types defined by debug info, it will not search the GDB
7969 primitive types. */
7970
7971 static struct type *
7972 ada_find_any_type (const char *name)
7973 {
7974 struct symbol *sym = ada_find_any_type_symbol (name);
7975
7976 if (sym != NULL)
7977 return SYMBOL_TYPE (sym);
7978
7979 return NULL;
7980 }
7981
7982 /* Given NAME_SYM and an associated BLOCK, find a "renaming" symbol
7983 associated with NAME_SYM's name. NAME_SYM may itself be a renaming
7984 symbol, in which case it is returned. Otherwise, this looks for
7985 symbols whose name is that of NAME_SYM suffixed with "___XR".
7986 Return symbol if found, and NULL otherwise. */
7987
7988 struct symbol *
7989 ada_find_renaming_symbol (struct symbol *name_sym, const struct block *block)
7990 {
7991 const char *name = SYMBOL_LINKAGE_NAME (name_sym);
7992 struct symbol *sym;
7993
7994 if (strstr (name, "___XR") != NULL)
7995 return name_sym;
7996
7997 sym = find_old_style_renaming_symbol (name, block);
7998
7999 if (sym != NULL)
8000 return sym;
8001
8002 /* Not right yet. FIXME pnh 7/20/2007. */
8003 sym = ada_find_any_type_symbol (name);
8004 if (sym != NULL && strstr (SYMBOL_LINKAGE_NAME (sym), "___XR") != NULL)
8005 return sym;
8006 else
8007 return NULL;
8008 }
8009
8010 static struct symbol *
8011 find_old_style_renaming_symbol (const char *name, const struct block *block)
8012 {
8013 const struct symbol *function_sym = block_linkage_function (block);
8014 char *rename;
8015
8016 if (function_sym != NULL)
8017 {
8018 /* If the symbol is defined inside a function, NAME is not fully
8019 qualified. This means we need to prepend the function name
8020 as well as adding the ``___XR'' suffix to build the name of
8021 the associated renaming symbol. */
8022 const char *function_name = SYMBOL_LINKAGE_NAME (function_sym);
8023 /* Function names sometimes contain suffixes used
8024 for instance to qualify nested subprograms. When building
8025 the XR type name, we need to make sure that this suffix is
8026 not included. So do not include any suffix in the function
8027 name length below. */
8028 int function_name_len = ada_name_prefix_len (function_name);
8029 const int rename_len = function_name_len + 2 /* "__" */
8030 + strlen (name) + 6 /* "___XR\0" */ ;
8031
8032 /* Strip the suffix if necessary. */
8033 ada_remove_trailing_digits (function_name, &function_name_len);
8034 ada_remove_po_subprogram_suffix (function_name, &function_name_len);
8035 ada_remove_Xbn_suffix (function_name, &function_name_len);
8036
8037 /* Library-level functions are a special case, as GNAT adds
8038 a ``_ada_'' prefix to the function name to avoid namespace
8039 pollution. However, the renaming symbols themselves do not
8040 have this prefix, so we need to skip this prefix if present. */
8041 if (function_name_len > 5 /* "_ada_" */
8042 && strstr (function_name, "_ada_") == function_name)
8043 {
8044 function_name += 5;
8045 function_name_len -= 5;
8046 }
8047
8048 rename = (char *) alloca (rename_len * sizeof (char));
8049 strncpy (rename, function_name, function_name_len);
8050 xsnprintf (rename + function_name_len, rename_len - function_name_len,
8051 "__%s___XR", name);
8052 }
8053 else
8054 {
8055 const int rename_len = strlen (name) + 6;
8056
8057 rename = (char *) alloca (rename_len * sizeof (char));
8058 xsnprintf (rename, rename_len * sizeof (char), "%s___XR", name);
8059 }
8060
8061 return ada_find_any_type_symbol (rename);
8062 }
8063
8064 /* Because of GNAT encoding conventions, several GDB symbols may match a
8065 given type name. If the type denoted by TYPE0 is to be preferred to
8066 that of TYPE1 for purposes of type printing, return non-zero;
8067 otherwise return 0. */
8068
8069 int
8070 ada_prefer_type (struct type *type0, struct type *type1)
8071 {
8072 if (type1 == NULL)
8073 return 1;
8074 else if (type0 == NULL)
8075 return 0;
8076 else if (TYPE_CODE (type1) == TYPE_CODE_VOID)
8077 return 1;
8078 else if (TYPE_CODE (type0) == TYPE_CODE_VOID)
8079 return 0;
8080 else if (TYPE_NAME (type1) == NULL && TYPE_NAME (type0) != NULL)
8081 return 1;
8082 else if (ada_is_constrained_packed_array_type (type0))
8083 return 1;
8084 else if (ada_is_array_descriptor_type (type0)
8085 && !ada_is_array_descriptor_type (type1))
8086 return 1;
8087 else
8088 {
8089 const char *type0_name = TYPE_NAME (type0);
8090 const char *type1_name = TYPE_NAME (type1);
8091
8092 if (type0_name != NULL && strstr (type0_name, "___XR") != NULL
8093 && (type1_name == NULL || strstr (type1_name, "___XR") == NULL))
8094 return 1;
8095 }
8096 return 0;
8097 }
8098
8099 /* The name of TYPE, which is its TYPE_NAME. Null if TYPE is
8100 null. */
8101
8102 const char *
8103 ada_type_name (struct type *type)
8104 {
8105 if (type == NULL)
8106 return NULL;
8107 return TYPE_NAME (type);
8108 }
8109
8110 /* Search the list of "descriptive" types associated to TYPE for a type
8111 whose name is NAME. */
8112
8113 static struct type *
8114 find_parallel_type_by_descriptive_type (struct type *type, const char *name)
8115 {
8116 struct type *result, *tmp;
8117
8118 if (ada_ignore_descriptive_types_p)
8119 return NULL;
8120
8121 /* If there no descriptive-type info, then there is no parallel type
8122 to be found. */
8123 if (!HAVE_GNAT_AUX_INFO (type))
8124 return NULL;
8125
8126 result = TYPE_DESCRIPTIVE_TYPE (type);
8127 while (result != NULL)
8128 {
8129 const char *result_name = ada_type_name (result);
8130
8131 if (result_name == NULL)
8132 {
8133 warning (_("unexpected null name on descriptive type"));
8134 return NULL;
8135 }
8136
8137 /* If the names match, stop. */
8138 if (strcmp (result_name, name) == 0)
8139 break;
8140
8141 /* Otherwise, look at the next item on the list, if any. */
8142 if (HAVE_GNAT_AUX_INFO (result))
8143 tmp = TYPE_DESCRIPTIVE_TYPE (result);
8144 else
8145 tmp = NULL;
8146
8147 /* If not found either, try after having resolved the typedef. */
8148 if (tmp != NULL)
8149 result = tmp;
8150 else
8151 {
8152 result = check_typedef (result);
8153 if (HAVE_GNAT_AUX_INFO (result))
8154 result = TYPE_DESCRIPTIVE_TYPE (result);
8155 else
8156 result = NULL;
8157 }
8158 }
8159
8160 /* If we didn't find a match, see whether this is a packed array. With
8161 older compilers, the descriptive type information is either absent or
8162 irrelevant when it comes to packed arrays so the above lookup fails.
8163 Fall back to using a parallel lookup by name in this case. */
8164 if (result == NULL && ada_is_constrained_packed_array_type (type))
8165 return ada_find_any_type (name);
8166
8167 return result;
8168 }
8169
8170 /* Find a parallel type to TYPE with the specified NAME, using the
8171 descriptive type taken from the debugging information, if available,
8172 and otherwise using the (slower) name-based method. */
8173
8174 static struct type *
8175 ada_find_parallel_type_with_name (struct type *type, const char *name)
8176 {
8177 struct type *result = NULL;
8178
8179 if (HAVE_GNAT_AUX_INFO (type))
8180 result = find_parallel_type_by_descriptive_type (type, name);
8181 else
8182 result = ada_find_any_type (name);
8183
8184 return result;
8185 }
8186
8187 /* Same as above, but specify the name of the parallel type by appending
8188 SUFFIX to the name of TYPE. */
8189
8190 struct type *
8191 ada_find_parallel_type (struct type *type, const char *suffix)
8192 {
8193 char *name;
8194 const char *type_name = ada_type_name (type);
8195 int len;
8196
8197 if (type_name == NULL)
8198 return NULL;
8199
8200 len = strlen (type_name);
8201
8202 name = (char *) alloca (len + strlen (suffix) + 1);
8203
8204 strcpy (name, type_name);
8205 strcpy (name + len, suffix);
8206
8207 return ada_find_parallel_type_with_name (type, name);
8208 }
8209
8210 /* If TYPE is a variable-size record type, return the corresponding template
8211 type describing its fields. Otherwise, return NULL. */
8212
8213 static struct type *
8214 dynamic_template_type (struct type *type)
8215 {
8216 type = ada_check_typedef (type);
8217
8218 if (type == NULL || TYPE_CODE (type) != TYPE_CODE_STRUCT
8219 || ada_type_name (type) == NULL)
8220 return NULL;
8221 else
8222 {
8223 int len = strlen (ada_type_name (type));
8224
8225 if (len > 6 && strcmp (ada_type_name (type) + len - 6, "___XVE") == 0)
8226 return type;
8227 else
8228 return ada_find_parallel_type (type, "___XVE");
8229 }
8230 }
8231
8232 /* Assuming that TEMPL_TYPE is a union or struct type, returns
8233 non-zero iff field FIELD_NUM of TEMPL_TYPE has dynamic size. */
8234
8235 static int
8236 is_dynamic_field (struct type *templ_type, int field_num)
8237 {
8238 const char *name = TYPE_FIELD_NAME (templ_type, field_num);
8239
8240 return name != NULL
8241 && TYPE_CODE (TYPE_FIELD_TYPE (templ_type, field_num)) == TYPE_CODE_PTR
8242 && strstr (name, "___XVL") != NULL;
8243 }
8244
8245 /* The index of the variant field of TYPE, or -1 if TYPE does not
8246 represent a variant record type. */
8247
8248 static int
8249 variant_field_index (struct type *type)
8250 {
8251 int f;
8252
8253 if (type == NULL || TYPE_CODE (type) != TYPE_CODE_STRUCT)
8254 return -1;
8255
8256 for (f = 0; f < TYPE_NFIELDS (type); f += 1)
8257 {
8258 if (ada_is_variant_part (type, f))
8259 return f;
8260 }
8261 return -1;
8262 }
8263
8264 /* A record type with no fields. */
8265
8266 static struct type *
8267 empty_record (struct type *templ)
8268 {
8269 struct type *type = alloc_type_copy (templ);
8270
8271 TYPE_CODE (type) = TYPE_CODE_STRUCT;
8272 TYPE_NFIELDS (type) = 0;
8273 TYPE_FIELDS (type) = NULL;
8274 INIT_NONE_SPECIFIC (type);
8275 TYPE_NAME (type) = "<empty>";
8276 TYPE_LENGTH (type) = 0;
8277 return type;
8278 }
8279
8280 /* An ordinary record type (with fixed-length fields) that describes
8281 the value of type TYPE at VALADDR or ADDRESS (see comments at
8282 the beginning of this section) VAL according to GNAT conventions.
8283 DVAL0 should describe the (portion of a) record that contains any
8284 necessary discriminants. It should be NULL if value_type (VAL) is
8285 an outer-level type (i.e., as opposed to a branch of a variant.) A
8286 variant field (unless unchecked) is replaced by a particular branch
8287 of the variant.
8288
8289 If not KEEP_DYNAMIC_FIELDS, then all fields whose position or
8290 length are not statically known are discarded. As a consequence,
8291 VALADDR, ADDRESS and DVAL0 are ignored.
8292
8293 NOTE: Limitations: For now, we assume that dynamic fields and
8294 variants occupy whole numbers of bytes. However, they need not be
8295 byte-aligned. */
8296
8297 struct type *
8298 ada_template_to_fixed_record_type_1 (struct type *type,
8299 const gdb_byte *valaddr,
8300 CORE_ADDR address, struct value *dval0,
8301 int keep_dynamic_fields)
8302 {
8303 struct value *mark = value_mark ();
8304 struct value *dval;
8305 struct type *rtype;
8306 int nfields, bit_len;
8307 int variant_field;
8308 long off;
8309 int fld_bit_len;
8310 int f;
8311
8312 /* Compute the number of fields in this record type that are going
8313 to be processed: unless keep_dynamic_fields, this includes only
8314 fields whose position and length are static will be processed. */
8315 if (keep_dynamic_fields)
8316 nfields = TYPE_NFIELDS (type);
8317 else
8318 {
8319 nfields = 0;
8320 while (nfields < TYPE_NFIELDS (type)
8321 && !ada_is_variant_part (type, nfields)
8322 && !is_dynamic_field (type, nfields))
8323 nfields++;
8324 }
8325
8326 rtype = alloc_type_copy (type);
8327 TYPE_CODE (rtype) = TYPE_CODE_STRUCT;
8328 INIT_NONE_SPECIFIC (rtype);
8329 TYPE_NFIELDS (rtype) = nfields;
8330 TYPE_FIELDS (rtype) = (struct field *)
8331 TYPE_ALLOC (rtype, nfields * sizeof (struct field));
8332 memset (TYPE_FIELDS (rtype), 0, sizeof (struct field) * nfields);
8333 TYPE_NAME (rtype) = ada_type_name (type);
8334 TYPE_FIXED_INSTANCE (rtype) = 1;
8335
8336 off = 0;
8337 bit_len = 0;
8338 variant_field = -1;
8339
8340 for (f = 0; f < nfields; f += 1)
8341 {
8342 off = align_value (off, field_alignment (type, f))
8343 + TYPE_FIELD_BITPOS (type, f);
8344 SET_FIELD_BITPOS (TYPE_FIELD (rtype, f), off);
8345 TYPE_FIELD_BITSIZE (rtype, f) = 0;
8346
8347 if (ada_is_variant_part (type, f))
8348 {
8349 variant_field = f;
8350 fld_bit_len = 0;
8351 }
8352 else if (is_dynamic_field (type, f))
8353 {
8354 const gdb_byte *field_valaddr = valaddr;
8355 CORE_ADDR field_address = address;
8356 struct type *field_type =
8357 TYPE_TARGET_TYPE (TYPE_FIELD_TYPE (type, f));
8358
8359 if (dval0 == NULL)
8360 {
8361 /* rtype's length is computed based on the run-time
8362 value of discriminants. If the discriminants are not
8363 initialized, the type size may be completely bogus and
8364 GDB may fail to allocate a value for it. So check the
8365 size first before creating the value. */
8366 ada_ensure_varsize_limit (rtype);
8367 /* Using plain value_from_contents_and_address here
8368 causes problems because we will end up trying to
8369 resolve a type that is currently being
8370 constructed. */
8371 dval = value_from_contents_and_address_unresolved (rtype,
8372 valaddr,
8373 address);
8374 rtype = value_type (dval);
8375 }
8376 else
8377 dval = dval0;
8378
8379 /* If the type referenced by this field is an aligner type, we need
8380 to unwrap that aligner type, because its size might not be set.
8381 Keeping the aligner type would cause us to compute the wrong
8382 size for this field, impacting the offset of the all the fields
8383 that follow this one. */
8384 if (ada_is_aligner_type (field_type))
8385 {
8386 long field_offset = TYPE_FIELD_BITPOS (field_type, f);
8387
8388 field_valaddr = cond_offset_host (field_valaddr, field_offset);
8389 field_address = cond_offset_target (field_address, field_offset);
8390 field_type = ada_aligned_type (field_type);
8391 }
8392
8393 field_valaddr = cond_offset_host (field_valaddr,
8394 off / TARGET_CHAR_BIT);
8395 field_address = cond_offset_target (field_address,
8396 off / TARGET_CHAR_BIT);
8397
8398 /* Get the fixed type of the field. Note that, in this case,
8399 we do not want to get the real type out of the tag: if
8400 the current field is the parent part of a tagged record,
8401 we will get the tag of the object. Clearly wrong: the real
8402 type of the parent is not the real type of the child. We
8403 would end up in an infinite loop. */
8404 field_type = ada_get_base_type (field_type);
8405 field_type = ada_to_fixed_type (field_type, field_valaddr,
8406 field_address, dval, 0);
8407 /* If the field size is already larger than the maximum
8408 object size, then the record itself will necessarily
8409 be larger than the maximum object size. We need to make
8410 this check now, because the size might be so ridiculously
8411 large (due to an uninitialized variable in the inferior)
8412 that it would cause an overflow when adding it to the
8413 record size. */
8414 ada_ensure_varsize_limit (field_type);
8415
8416 TYPE_FIELD_TYPE (rtype, f) = field_type;
8417 TYPE_FIELD_NAME (rtype, f) = TYPE_FIELD_NAME (type, f);
8418 /* The multiplication can potentially overflow. But because
8419 the field length has been size-checked just above, and
8420 assuming that the maximum size is a reasonable value,
8421 an overflow should not happen in practice. So rather than
8422 adding overflow recovery code to this already complex code,
8423 we just assume that it's not going to happen. */
8424 fld_bit_len =
8425 TYPE_LENGTH (TYPE_FIELD_TYPE (rtype, f)) * TARGET_CHAR_BIT;
8426 }
8427 else
8428 {
8429 /* Note: If this field's type is a typedef, it is important
8430 to preserve the typedef layer.
8431
8432 Otherwise, we might be transforming a typedef to a fat
8433 pointer (encoding a pointer to an unconstrained array),
8434 into a basic fat pointer (encoding an unconstrained
8435 array). As both types are implemented using the same
8436 structure, the typedef is the only clue which allows us
8437 to distinguish between the two options. Stripping it
8438 would prevent us from printing this field appropriately. */
8439 TYPE_FIELD_TYPE (rtype, f) = TYPE_FIELD_TYPE (type, f);
8440 TYPE_FIELD_NAME (rtype, f) = TYPE_FIELD_NAME (type, f);
8441 if (TYPE_FIELD_BITSIZE (type, f) > 0)
8442 fld_bit_len =
8443 TYPE_FIELD_BITSIZE (rtype, f) = TYPE_FIELD_BITSIZE (type, f);
8444 else
8445 {
8446 struct type *field_type = TYPE_FIELD_TYPE (type, f);
8447
8448 /* We need to be careful of typedefs when computing
8449 the length of our field. If this is a typedef,
8450 get the length of the target type, not the length
8451 of the typedef. */
8452 if (TYPE_CODE (field_type) == TYPE_CODE_TYPEDEF)
8453 field_type = ada_typedef_target_type (field_type);
8454
8455 fld_bit_len =
8456 TYPE_LENGTH (ada_check_typedef (field_type)) * TARGET_CHAR_BIT;
8457 }
8458 }
8459 if (off + fld_bit_len > bit_len)
8460 bit_len = off + fld_bit_len;
8461 off += fld_bit_len;
8462 TYPE_LENGTH (rtype) =
8463 align_value (bit_len, TARGET_CHAR_BIT) / TARGET_CHAR_BIT;
8464 }
8465
8466 /* We handle the variant part, if any, at the end because of certain
8467 odd cases in which it is re-ordered so as NOT to be the last field of
8468 the record. This can happen in the presence of representation
8469 clauses. */
8470 if (variant_field >= 0)
8471 {
8472 struct type *branch_type;
8473
8474 off = TYPE_FIELD_BITPOS (rtype, variant_field);
8475
8476 if (dval0 == NULL)
8477 {
8478 /* Using plain value_from_contents_and_address here causes
8479 problems because we will end up trying to resolve a type
8480 that is currently being constructed. */
8481 dval = value_from_contents_and_address_unresolved (rtype, valaddr,
8482 address);
8483 rtype = value_type (dval);
8484 }
8485 else
8486 dval = dval0;
8487
8488 branch_type =
8489 to_fixed_variant_branch_type
8490 (TYPE_FIELD_TYPE (type, variant_field),
8491 cond_offset_host (valaddr, off / TARGET_CHAR_BIT),
8492 cond_offset_target (address, off / TARGET_CHAR_BIT), dval);
8493 if (branch_type == NULL)
8494 {
8495 for (f = variant_field + 1; f < TYPE_NFIELDS (rtype); f += 1)
8496 TYPE_FIELDS (rtype)[f - 1] = TYPE_FIELDS (rtype)[f];
8497 TYPE_NFIELDS (rtype) -= 1;
8498 }
8499 else
8500 {
8501 TYPE_FIELD_TYPE (rtype, variant_field) = branch_type;
8502 TYPE_FIELD_NAME (rtype, variant_field) = "S";
8503 fld_bit_len =
8504 TYPE_LENGTH (TYPE_FIELD_TYPE (rtype, variant_field)) *
8505 TARGET_CHAR_BIT;
8506 if (off + fld_bit_len > bit_len)
8507 bit_len = off + fld_bit_len;
8508 TYPE_LENGTH (rtype) =
8509 align_value (bit_len, TARGET_CHAR_BIT) / TARGET_CHAR_BIT;
8510 }
8511 }
8512
8513 /* According to exp_dbug.ads, the size of TYPE for variable-size records
8514 should contain the alignment of that record, which should be a strictly
8515 positive value. If null or negative, then something is wrong, most
8516 probably in the debug info. In that case, we don't round up the size
8517 of the resulting type. If this record is not part of another structure,
8518 the current RTYPE length might be good enough for our purposes. */
8519 if (TYPE_LENGTH (type) <= 0)
8520 {
8521 if (TYPE_NAME (rtype))
8522 warning (_("Invalid type size for `%s' detected: %s."),
8523 TYPE_NAME (rtype), pulongest (TYPE_LENGTH (type)));
8524 else
8525 warning (_("Invalid type size for <unnamed> detected: %s."),
8526 pulongest (TYPE_LENGTH (type)));
8527 }
8528 else
8529 {
8530 TYPE_LENGTH (rtype) = align_value (TYPE_LENGTH (rtype),
8531 TYPE_LENGTH (type));
8532 }
8533
8534 value_free_to_mark (mark);
8535 if (TYPE_LENGTH (rtype) > varsize_limit)
8536 error (_("record type with dynamic size is larger than varsize-limit"));
8537 return rtype;
8538 }
8539
8540 /* As for ada_template_to_fixed_record_type_1 with KEEP_DYNAMIC_FIELDS
8541 of 1. */
8542
8543 static struct type *
8544 template_to_fixed_record_type (struct type *type, const gdb_byte *valaddr,
8545 CORE_ADDR address, struct value *dval0)
8546 {
8547 return ada_template_to_fixed_record_type_1 (type, valaddr,
8548 address, dval0, 1);
8549 }
8550
8551 /* An ordinary record type in which ___XVL-convention fields and
8552 ___XVU- and ___XVN-convention field types in TYPE0 are replaced with
8553 static approximations, containing all possible fields. Uses
8554 no runtime values. Useless for use in values, but that's OK,
8555 since the results are used only for type determinations. Works on both
8556 structs and unions. Representation note: to save space, we memorize
8557 the result of this function in the TYPE_TARGET_TYPE of the
8558 template type. */
8559
8560 static struct type *
8561 template_to_static_fixed_type (struct type *type0)
8562 {
8563 struct type *type;
8564 int nfields;
8565 int f;
8566
8567 /* No need no do anything if the input type is already fixed. */
8568 if (TYPE_FIXED_INSTANCE (type0))
8569 return type0;
8570
8571 /* Likewise if we already have computed the static approximation. */
8572 if (TYPE_TARGET_TYPE (type0) != NULL)
8573 return TYPE_TARGET_TYPE (type0);
8574
8575 /* Don't clone TYPE0 until we are sure we are going to need a copy. */
8576 type = type0;
8577 nfields = TYPE_NFIELDS (type0);
8578
8579 /* Whether or not we cloned TYPE0, cache the result so that we don't do
8580 recompute all over next time. */
8581 TYPE_TARGET_TYPE (type0) = type;
8582
8583 for (f = 0; f < nfields; f += 1)
8584 {
8585 struct type *field_type = TYPE_FIELD_TYPE (type0, f);
8586 struct type *new_type;
8587
8588 if (is_dynamic_field (type0, f))
8589 {
8590 field_type = ada_check_typedef (field_type);
8591 new_type = to_static_fixed_type (TYPE_TARGET_TYPE (field_type));
8592 }
8593 else
8594 new_type = static_unwrap_type (field_type);
8595
8596 if (new_type != field_type)
8597 {
8598 /* Clone TYPE0 only the first time we get a new field type. */
8599 if (type == type0)
8600 {
8601 TYPE_TARGET_TYPE (type0) = type = alloc_type_copy (type0);
8602 TYPE_CODE (type) = TYPE_CODE (type0);
8603 INIT_NONE_SPECIFIC (type);
8604 TYPE_NFIELDS (type) = nfields;
8605 TYPE_FIELDS (type) = (struct field *)
8606 TYPE_ALLOC (type, nfields * sizeof (struct field));
8607 memcpy (TYPE_FIELDS (type), TYPE_FIELDS (type0),
8608 sizeof (struct field) * nfields);
8609 TYPE_NAME (type) = ada_type_name (type0);
8610 TYPE_FIXED_INSTANCE (type) = 1;
8611 TYPE_LENGTH (type) = 0;
8612 }
8613 TYPE_FIELD_TYPE (type, f) = new_type;
8614 TYPE_FIELD_NAME (type, f) = TYPE_FIELD_NAME (type0, f);
8615 }
8616 }
8617
8618 return type;
8619 }
8620
8621 /* Given an object of type TYPE whose contents are at VALADDR and
8622 whose address in memory is ADDRESS, returns a revision of TYPE,
8623 which should be a non-dynamic-sized record, in which the variant
8624 part, if any, is replaced with the appropriate branch. Looks
8625 for discriminant values in DVAL0, which can be NULL if the record
8626 contains the necessary discriminant values. */
8627
8628 static struct type *
8629 to_record_with_fixed_variant_part (struct type *type, const gdb_byte *valaddr,
8630 CORE_ADDR address, struct value *dval0)
8631 {
8632 struct value *mark = value_mark ();
8633 struct value *dval;
8634 struct type *rtype;
8635 struct type *branch_type;
8636 int nfields = TYPE_NFIELDS (type);
8637 int variant_field = variant_field_index (type);
8638
8639 if (variant_field == -1)
8640 return type;
8641
8642 if (dval0 == NULL)
8643 {
8644 dval = value_from_contents_and_address (type, valaddr, address);
8645 type = value_type (dval);
8646 }
8647 else
8648 dval = dval0;
8649
8650 rtype = alloc_type_copy (type);
8651 TYPE_CODE (rtype) = TYPE_CODE_STRUCT;
8652 INIT_NONE_SPECIFIC (rtype);
8653 TYPE_NFIELDS (rtype) = nfields;
8654 TYPE_FIELDS (rtype) =
8655 (struct field *) TYPE_ALLOC (rtype, nfields * sizeof (struct field));
8656 memcpy (TYPE_FIELDS (rtype), TYPE_FIELDS (type),
8657 sizeof (struct field) * nfields);
8658 TYPE_NAME (rtype) = ada_type_name (type);
8659 TYPE_FIXED_INSTANCE (rtype) = 1;
8660 TYPE_LENGTH (rtype) = TYPE_LENGTH (type);
8661
8662 branch_type = to_fixed_variant_branch_type
8663 (TYPE_FIELD_TYPE (type, variant_field),
8664 cond_offset_host (valaddr,
8665 TYPE_FIELD_BITPOS (type, variant_field)
8666 / TARGET_CHAR_BIT),
8667 cond_offset_target (address,
8668 TYPE_FIELD_BITPOS (type, variant_field)
8669 / TARGET_CHAR_BIT), dval);
8670 if (branch_type == NULL)
8671 {
8672 int f;
8673
8674 for (f = variant_field + 1; f < nfields; f += 1)
8675 TYPE_FIELDS (rtype)[f - 1] = TYPE_FIELDS (rtype)[f];
8676 TYPE_NFIELDS (rtype) -= 1;
8677 }
8678 else
8679 {
8680 TYPE_FIELD_TYPE (rtype, variant_field) = branch_type;
8681 TYPE_FIELD_NAME (rtype, variant_field) = "S";
8682 TYPE_FIELD_BITSIZE (rtype, variant_field) = 0;
8683 TYPE_LENGTH (rtype) += TYPE_LENGTH (branch_type);
8684 }
8685 TYPE_LENGTH (rtype) -= TYPE_LENGTH (TYPE_FIELD_TYPE (type, variant_field));
8686
8687 value_free_to_mark (mark);
8688 return rtype;
8689 }
8690
8691 /* An ordinary record type (with fixed-length fields) that describes
8692 the value at (TYPE0, VALADDR, ADDRESS) [see explanation at
8693 beginning of this section]. Any necessary discriminants' values
8694 should be in DVAL, a record value; it may be NULL if the object
8695 at ADDR itself contains any necessary discriminant values.
8696 Additionally, VALADDR and ADDRESS may also be NULL if no discriminant
8697 values from the record are needed. Except in the case that DVAL,
8698 VALADDR, and ADDRESS are all 0 or NULL, a variant field (unless
8699 unchecked) is replaced by a particular branch of the variant.
8700
8701 NOTE: the case in which DVAL and VALADDR are NULL and ADDRESS is 0
8702 is questionable and may be removed. It can arise during the
8703 processing of an unconstrained-array-of-record type where all the
8704 variant branches have exactly the same size. This is because in
8705 such cases, the compiler does not bother to use the XVS convention
8706 when encoding the record. I am currently dubious of this
8707 shortcut and suspect the compiler should be altered. FIXME. */
8708
8709 static struct type *
8710 to_fixed_record_type (struct type *type0, const gdb_byte *valaddr,
8711 CORE_ADDR address, struct value *dval)
8712 {
8713 struct type *templ_type;
8714
8715 if (TYPE_FIXED_INSTANCE (type0))
8716 return type0;
8717
8718 templ_type = dynamic_template_type (type0);
8719
8720 if (templ_type != NULL)
8721 return template_to_fixed_record_type (templ_type, valaddr, address, dval);
8722 else if (variant_field_index (type0) >= 0)
8723 {
8724 if (dval == NULL && valaddr == NULL && address == 0)
8725 return type0;
8726 return to_record_with_fixed_variant_part (type0, valaddr, address,
8727 dval);
8728 }
8729 else
8730 {
8731 TYPE_FIXED_INSTANCE (type0) = 1;
8732 return type0;
8733 }
8734
8735 }
8736
8737 /* An ordinary record type (with fixed-length fields) that describes
8738 the value at (VAR_TYPE0, VALADDR, ADDRESS), where VAR_TYPE0 is a
8739 union type. Any necessary discriminants' values should be in DVAL,
8740 a record value. That is, this routine selects the appropriate
8741 branch of the union at ADDR according to the discriminant value
8742 indicated in the union's type name. Returns VAR_TYPE0 itself if
8743 it represents a variant subject to a pragma Unchecked_Union. */
8744
8745 static struct type *
8746 to_fixed_variant_branch_type (struct type *var_type0, const gdb_byte *valaddr,
8747 CORE_ADDR address, struct value *dval)
8748 {
8749 int which;
8750 struct type *templ_type;
8751 struct type *var_type;
8752
8753 if (TYPE_CODE (var_type0) == TYPE_CODE_PTR)
8754 var_type = TYPE_TARGET_TYPE (var_type0);
8755 else
8756 var_type = var_type0;
8757
8758 templ_type = ada_find_parallel_type (var_type, "___XVU");
8759
8760 if (templ_type != NULL)
8761 var_type = templ_type;
8762
8763 if (is_unchecked_variant (var_type, value_type (dval)))
8764 return var_type0;
8765 which =
8766 ada_which_variant_applies (var_type,
8767 value_type (dval), value_contents (dval));
8768
8769 if (which < 0)
8770 return empty_record (var_type);
8771 else if (is_dynamic_field (var_type, which))
8772 return to_fixed_record_type
8773 (TYPE_TARGET_TYPE (TYPE_FIELD_TYPE (var_type, which)),
8774 valaddr, address, dval);
8775 else if (variant_field_index (TYPE_FIELD_TYPE (var_type, which)) >= 0)
8776 return
8777 to_fixed_record_type
8778 (TYPE_FIELD_TYPE (var_type, which), valaddr, address, dval);
8779 else
8780 return TYPE_FIELD_TYPE (var_type, which);
8781 }
8782
8783 /* Assuming RANGE_TYPE is a TYPE_CODE_RANGE, return nonzero if
8784 ENCODING_TYPE, a type following the GNAT conventions for discrete
8785 type encodings, only carries redundant information. */
8786
8787 static int
8788 ada_is_redundant_range_encoding (struct type *range_type,
8789 struct type *encoding_type)
8790 {
8791 const char *bounds_str;
8792 int n;
8793 LONGEST lo, hi;
8794
8795 gdb_assert (TYPE_CODE (range_type) == TYPE_CODE_RANGE);
8796
8797 if (TYPE_CODE (get_base_type (range_type))
8798 != TYPE_CODE (get_base_type (encoding_type)))
8799 {
8800 /* The compiler probably used a simple base type to describe
8801 the range type instead of the range's actual base type,
8802 expecting us to get the real base type from the encoding
8803 anyway. In this situation, the encoding cannot be ignored
8804 as redundant. */
8805 return 0;
8806 }
8807
8808 if (is_dynamic_type (range_type))
8809 return 0;
8810
8811 if (TYPE_NAME (encoding_type) == NULL)
8812 return 0;
8813
8814 bounds_str = strstr (TYPE_NAME (encoding_type), "___XDLU_");
8815 if (bounds_str == NULL)
8816 return 0;
8817
8818 n = 8; /* Skip "___XDLU_". */
8819 if (!ada_scan_number (bounds_str, n, &lo, &n))
8820 return 0;
8821 if (TYPE_LOW_BOUND (range_type) != lo)
8822 return 0;
8823
8824 n += 2; /* Skip the "__" separator between the two bounds. */
8825 if (!ada_scan_number (bounds_str, n, &hi, &n))
8826 return 0;
8827 if (TYPE_HIGH_BOUND (range_type) != hi)
8828 return 0;
8829
8830 return 1;
8831 }
8832
8833 /* Given the array type ARRAY_TYPE, return nonzero if DESC_TYPE,
8834 a type following the GNAT encoding for describing array type
8835 indices, only carries redundant information. */
8836
8837 static int
8838 ada_is_redundant_index_type_desc (struct type *array_type,
8839 struct type *desc_type)
8840 {
8841 struct type *this_layer = check_typedef (array_type);
8842 int i;
8843
8844 for (i = 0; i < TYPE_NFIELDS (desc_type); i++)
8845 {
8846 if (!ada_is_redundant_range_encoding (TYPE_INDEX_TYPE (this_layer),
8847 TYPE_FIELD_TYPE (desc_type, i)))
8848 return 0;
8849 this_layer = check_typedef (TYPE_TARGET_TYPE (this_layer));
8850 }
8851
8852 return 1;
8853 }
8854
8855 /* Assuming that TYPE0 is an array type describing the type of a value
8856 at ADDR, and that DVAL describes a record containing any
8857 discriminants used in TYPE0, returns a type for the value that
8858 contains no dynamic components (that is, no components whose sizes
8859 are determined by run-time quantities). Unless IGNORE_TOO_BIG is
8860 true, gives an error message if the resulting type's size is over
8861 varsize_limit. */
8862
8863 static struct type *
8864 to_fixed_array_type (struct type *type0, struct value *dval,
8865 int ignore_too_big)
8866 {
8867 struct type *index_type_desc;
8868 struct type *result;
8869 int constrained_packed_array_p;
8870 static const char *xa_suffix = "___XA";
8871
8872 type0 = ada_check_typedef (type0);
8873 if (TYPE_FIXED_INSTANCE (type0))
8874 return type0;
8875
8876 constrained_packed_array_p = ada_is_constrained_packed_array_type (type0);
8877 if (constrained_packed_array_p)
8878 type0 = decode_constrained_packed_array_type (type0);
8879
8880 index_type_desc = ada_find_parallel_type (type0, xa_suffix);
8881
8882 /* As mentioned in exp_dbug.ads, for non bit-packed arrays an
8883 encoding suffixed with 'P' may still be generated. If so,
8884 it should be used to find the XA type. */
8885
8886 if (index_type_desc == NULL)
8887 {
8888 const char *type_name = ada_type_name (type0);
8889
8890 if (type_name != NULL)
8891 {
8892 const int len = strlen (type_name);
8893 char *name = (char *) alloca (len + strlen (xa_suffix));
8894
8895 if (type_name[len - 1] == 'P')
8896 {
8897 strcpy (name, type_name);
8898 strcpy (name + len - 1, xa_suffix);
8899 index_type_desc = ada_find_parallel_type_with_name (type0, name);
8900 }
8901 }
8902 }
8903
8904 ada_fixup_array_indexes_type (index_type_desc);
8905 if (index_type_desc != NULL
8906 && ada_is_redundant_index_type_desc (type0, index_type_desc))
8907 {
8908 /* Ignore this ___XA parallel type, as it does not bring any
8909 useful information. This allows us to avoid creating fixed
8910 versions of the array's index types, which would be identical
8911 to the original ones. This, in turn, can also help avoid
8912 the creation of fixed versions of the array itself. */
8913 index_type_desc = NULL;
8914 }
8915
8916 if (index_type_desc == NULL)
8917 {
8918 struct type *elt_type0 = ada_check_typedef (TYPE_TARGET_TYPE (type0));
8919
8920 /* NOTE: elt_type---the fixed version of elt_type0---should never
8921 depend on the contents of the array in properly constructed
8922 debugging data. */
8923 /* Create a fixed version of the array element type.
8924 We're not providing the address of an element here,
8925 and thus the actual object value cannot be inspected to do
8926 the conversion. This should not be a problem, since arrays of
8927 unconstrained objects are not allowed. In particular, all
8928 the elements of an array of a tagged type should all be of
8929 the same type specified in the debugging info. No need to
8930 consult the object tag. */
8931 struct type *elt_type = ada_to_fixed_type (elt_type0, 0, 0, dval, 1);
8932
8933 /* Make sure we always create a new array type when dealing with
8934 packed array types, since we're going to fix-up the array
8935 type length and element bitsize a little further down. */
8936 if (elt_type0 == elt_type && !constrained_packed_array_p)
8937 result = type0;
8938 else
8939 result = create_array_type (alloc_type_copy (type0),
8940 elt_type, TYPE_INDEX_TYPE (type0));
8941 }
8942 else
8943 {
8944 int i;
8945 struct type *elt_type0;
8946
8947 elt_type0 = type0;
8948 for (i = TYPE_NFIELDS (index_type_desc); i > 0; i -= 1)
8949 elt_type0 = TYPE_TARGET_TYPE (elt_type0);
8950
8951 /* NOTE: result---the fixed version of elt_type0---should never
8952 depend on the contents of the array in properly constructed
8953 debugging data. */
8954 /* Create a fixed version of the array element type.
8955 We're not providing the address of an element here,
8956 and thus the actual object value cannot be inspected to do
8957 the conversion. This should not be a problem, since arrays of
8958 unconstrained objects are not allowed. In particular, all
8959 the elements of an array of a tagged type should all be of
8960 the same type specified in the debugging info. No need to
8961 consult the object tag. */
8962 result =
8963 ada_to_fixed_type (ada_check_typedef (elt_type0), 0, 0, dval, 1);
8964
8965 elt_type0 = type0;
8966 for (i = TYPE_NFIELDS (index_type_desc) - 1; i >= 0; i -= 1)
8967 {
8968 struct type *range_type =
8969 to_fixed_range_type (TYPE_FIELD_TYPE (index_type_desc, i), dval);
8970
8971 result = create_array_type (alloc_type_copy (elt_type0),
8972 result, range_type);
8973 elt_type0 = TYPE_TARGET_TYPE (elt_type0);
8974 }
8975 if (!ignore_too_big && TYPE_LENGTH (result) > varsize_limit)
8976 error (_("array type with dynamic size is larger than varsize-limit"));
8977 }
8978
8979 /* We want to preserve the type name. This can be useful when
8980 trying to get the type name of a value that has already been
8981 printed (for instance, if the user did "print VAR; whatis $". */
8982 TYPE_NAME (result) = TYPE_NAME (type0);
8983
8984 if (constrained_packed_array_p)
8985 {
8986 /* So far, the resulting type has been created as if the original
8987 type was a regular (non-packed) array type. As a result, the
8988 bitsize of the array elements needs to be set again, and the array
8989 length needs to be recomputed based on that bitsize. */
8990 int len = TYPE_LENGTH (result) / TYPE_LENGTH (TYPE_TARGET_TYPE (result));
8991 int elt_bitsize = TYPE_FIELD_BITSIZE (type0, 0);
8992
8993 TYPE_FIELD_BITSIZE (result, 0) = TYPE_FIELD_BITSIZE (type0, 0);
8994 TYPE_LENGTH (result) = len * elt_bitsize / HOST_CHAR_BIT;
8995 if (TYPE_LENGTH (result) * HOST_CHAR_BIT < len * elt_bitsize)
8996 TYPE_LENGTH (result)++;
8997 }
8998
8999 TYPE_FIXED_INSTANCE (result) = 1;
9000 return result;
9001 }
9002
9003
9004 /* A standard type (containing no dynamically sized components)
9005 corresponding to TYPE for the value (TYPE, VALADDR, ADDRESS)
9006 DVAL describes a record containing any discriminants used in TYPE0,
9007 and may be NULL if there are none, or if the object of type TYPE at
9008 ADDRESS or in VALADDR contains these discriminants.
9009
9010 If CHECK_TAG is not null, in the case of tagged types, this function
9011 attempts to locate the object's tag and use it to compute the actual
9012 type. However, when ADDRESS is null, we cannot use it to determine the
9013 location of the tag, and therefore compute the tagged type's actual type.
9014 So we return the tagged type without consulting the tag. */
9015
9016 static struct type *
9017 ada_to_fixed_type_1 (struct type *type, const gdb_byte *valaddr,
9018 CORE_ADDR address, struct value *dval, int check_tag)
9019 {
9020 type = ada_check_typedef (type);
9021
9022 /* Only un-fixed types need to be handled here. */
9023 if (!HAVE_GNAT_AUX_INFO (type))
9024 return type;
9025
9026 switch (TYPE_CODE (type))
9027 {
9028 default:
9029 return type;
9030 case TYPE_CODE_STRUCT:
9031 {
9032 struct type *static_type = to_static_fixed_type (type);
9033 struct type *fixed_record_type =
9034 to_fixed_record_type (type, valaddr, address, NULL);
9035
9036 /* If STATIC_TYPE is a tagged type and we know the object's address,
9037 then we can determine its tag, and compute the object's actual
9038 type from there. Note that we have to use the fixed record
9039 type (the parent part of the record may have dynamic fields
9040 and the way the location of _tag is expressed may depend on
9041 them). */
9042
9043 if (check_tag && address != 0 && ada_is_tagged_type (static_type, 0))
9044 {
9045 struct value *tag =
9046 value_tag_from_contents_and_address
9047 (fixed_record_type,
9048 valaddr,
9049 address);
9050 struct type *real_type = type_from_tag (tag);
9051 struct value *obj =
9052 value_from_contents_and_address (fixed_record_type,
9053 valaddr,
9054 address);
9055 fixed_record_type = value_type (obj);
9056 if (real_type != NULL)
9057 return to_fixed_record_type
9058 (real_type, NULL,
9059 value_address (ada_tag_value_at_base_address (obj)), NULL);
9060 }
9061
9062 /* Check to see if there is a parallel ___XVZ variable.
9063 If there is, then it provides the actual size of our type. */
9064 else if (ada_type_name (fixed_record_type) != NULL)
9065 {
9066 const char *name = ada_type_name (fixed_record_type);
9067 char *xvz_name
9068 = (char *) alloca (strlen (name) + 7 /* "___XVZ\0" */);
9069 bool xvz_found = false;
9070 LONGEST size;
9071
9072 xsnprintf (xvz_name, strlen (name) + 7, "%s___XVZ", name);
9073 try
9074 {
9075 xvz_found = get_int_var_value (xvz_name, size);
9076 }
9077 catch (const gdb_exception_error &except)
9078 {
9079 /* We found the variable, but somehow failed to read
9080 its value. Rethrow the same error, but with a little
9081 bit more information, to help the user understand
9082 what went wrong (Eg: the variable might have been
9083 optimized out). */
9084 throw_error (except.error,
9085 _("unable to read value of %s (%s)"),
9086 xvz_name, except.what ());
9087 }
9088
9089 if (xvz_found && TYPE_LENGTH (fixed_record_type) != size)
9090 {
9091 fixed_record_type = copy_type (fixed_record_type);
9092 TYPE_LENGTH (fixed_record_type) = size;
9093
9094 /* The FIXED_RECORD_TYPE may have be a stub. We have
9095 observed this when the debugging info is STABS, and
9096 apparently it is something that is hard to fix.
9097
9098 In practice, we don't need the actual type definition
9099 at all, because the presence of the XVZ variable allows us
9100 to assume that there must be a XVS type as well, which we
9101 should be able to use later, when we need the actual type
9102 definition.
9103
9104 In the meantime, pretend that the "fixed" type we are
9105 returning is NOT a stub, because this can cause trouble
9106 when using this type to create new types targeting it.
9107 Indeed, the associated creation routines often check
9108 whether the target type is a stub and will try to replace
9109 it, thus using a type with the wrong size. This, in turn,
9110 might cause the new type to have the wrong size too.
9111 Consider the case of an array, for instance, where the size
9112 of the array is computed from the number of elements in
9113 our array multiplied by the size of its element. */
9114 TYPE_STUB (fixed_record_type) = 0;
9115 }
9116 }
9117 return fixed_record_type;
9118 }
9119 case TYPE_CODE_ARRAY:
9120 return to_fixed_array_type (type, dval, 1);
9121 case TYPE_CODE_UNION:
9122 if (dval == NULL)
9123 return type;
9124 else
9125 return to_fixed_variant_branch_type (type, valaddr, address, dval);
9126 }
9127 }
9128
9129 /* The same as ada_to_fixed_type_1, except that it preserves the type
9130 if it is a TYPE_CODE_TYPEDEF of a type that is already fixed.
9131
9132 The typedef layer needs be preserved in order to differentiate between
9133 arrays and array pointers when both types are implemented using the same
9134 fat pointer. In the array pointer case, the pointer is encoded as
9135 a typedef of the pointer type. For instance, considering:
9136
9137 type String_Access is access String;
9138 S1 : String_Access := null;
9139
9140 To the debugger, S1 is defined as a typedef of type String. But
9141 to the user, it is a pointer. So if the user tries to print S1,
9142 we should not dereference the array, but print the array address
9143 instead.
9144
9145 If we didn't preserve the typedef layer, we would lose the fact that
9146 the type is to be presented as a pointer (needs de-reference before
9147 being printed). And we would also use the source-level type name. */
9148
9149 struct type *
9150 ada_to_fixed_type (struct type *type, const gdb_byte *valaddr,
9151 CORE_ADDR address, struct value *dval, int check_tag)
9152
9153 {
9154 struct type *fixed_type =
9155 ada_to_fixed_type_1 (type, valaddr, address, dval, check_tag);
9156
9157 /* If TYPE is a typedef and its target type is the same as the FIXED_TYPE,
9158 then preserve the typedef layer.
9159
9160 Implementation note: We can only check the main-type portion of
9161 the TYPE and FIXED_TYPE, because eliminating the typedef layer
9162 from TYPE now returns a type that has the same instance flags
9163 as TYPE. For instance, if TYPE is a "typedef const", and its
9164 target type is a "struct", then the typedef elimination will return
9165 a "const" version of the target type. See check_typedef for more
9166 details about how the typedef layer elimination is done.
9167
9168 brobecker/2010-11-19: It seems to me that the only case where it is
9169 useful to preserve the typedef layer is when dealing with fat pointers.
9170 Perhaps, we could add a check for that and preserve the typedef layer
9171 only in that situation. But this seems unecessary so far, probably
9172 because we call check_typedef/ada_check_typedef pretty much everywhere.
9173 */
9174 if (TYPE_CODE (type) == TYPE_CODE_TYPEDEF
9175 && (TYPE_MAIN_TYPE (ada_typedef_target_type (type))
9176 == TYPE_MAIN_TYPE (fixed_type)))
9177 return type;
9178
9179 return fixed_type;
9180 }
9181
9182 /* A standard (static-sized) type corresponding as well as possible to
9183 TYPE0, but based on no runtime data. */
9184
9185 static struct type *
9186 to_static_fixed_type (struct type *type0)
9187 {
9188 struct type *type;
9189
9190 if (type0 == NULL)
9191 return NULL;
9192
9193 if (TYPE_FIXED_INSTANCE (type0))
9194 return type0;
9195
9196 type0 = ada_check_typedef (type0);
9197
9198 switch (TYPE_CODE (type0))
9199 {
9200 default:
9201 return type0;
9202 case TYPE_CODE_STRUCT:
9203 type = dynamic_template_type (type0);
9204 if (type != NULL)
9205 return template_to_static_fixed_type (type);
9206 else
9207 return template_to_static_fixed_type (type0);
9208 case TYPE_CODE_UNION:
9209 type = ada_find_parallel_type (type0, "___XVU");
9210 if (type != NULL)
9211 return template_to_static_fixed_type (type);
9212 else
9213 return template_to_static_fixed_type (type0);
9214 }
9215 }
9216
9217 /* A static approximation of TYPE with all type wrappers removed. */
9218
9219 static struct type *
9220 static_unwrap_type (struct type *type)
9221 {
9222 if (ada_is_aligner_type (type))
9223 {
9224 struct type *type1 = TYPE_FIELD_TYPE (ada_check_typedef (type), 0);
9225 if (ada_type_name (type1) == NULL)
9226 TYPE_NAME (type1) = ada_type_name (type);
9227
9228 return static_unwrap_type (type1);
9229 }
9230 else
9231 {
9232 struct type *raw_real_type = ada_get_base_type (type);
9233
9234 if (raw_real_type == type)
9235 return type;
9236 else
9237 return to_static_fixed_type (raw_real_type);
9238 }
9239 }
9240
9241 /* In some cases, incomplete and private types require
9242 cross-references that are not resolved as records (for example,
9243 type Foo;
9244 type FooP is access Foo;
9245 V: FooP;
9246 type Foo is array ...;
9247 ). In these cases, since there is no mechanism for producing
9248 cross-references to such types, we instead substitute for FooP a
9249 stub enumeration type that is nowhere resolved, and whose tag is
9250 the name of the actual type. Call these types "non-record stubs". */
9251
9252 /* A type equivalent to TYPE that is not a non-record stub, if one
9253 exists, otherwise TYPE. */
9254
9255 struct type *
9256 ada_check_typedef (struct type *type)
9257 {
9258 if (type == NULL)
9259 return NULL;
9260
9261 /* If our type is an access to an unconstrained array, which is encoded
9262 as a TYPE_CODE_TYPEDEF of a fat pointer, then we're done.
9263 We don't want to strip the TYPE_CODE_TYPDEF layer, because this is
9264 what allows us to distinguish between fat pointers that represent
9265 array types, and fat pointers that represent array access types
9266 (in both cases, the compiler implements them as fat pointers). */
9267 if (ada_is_access_to_unconstrained_array (type))
9268 return type;
9269
9270 type = check_typedef (type);
9271 if (type == NULL || TYPE_CODE (type) != TYPE_CODE_ENUM
9272 || !TYPE_STUB (type)
9273 || TYPE_NAME (type) == NULL)
9274 return type;
9275 else
9276 {
9277 const char *name = TYPE_NAME (type);
9278 struct type *type1 = ada_find_any_type (name);
9279
9280 if (type1 == NULL)
9281 return type;
9282
9283 /* TYPE1 might itself be a TYPE_CODE_TYPEDEF (this can happen with
9284 stubs pointing to arrays, as we don't create symbols for array
9285 types, only for the typedef-to-array types). If that's the case,
9286 strip the typedef layer. */
9287 if (TYPE_CODE (type1) == TYPE_CODE_TYPEDEF)
9288 type1 = ada_check_typedef (type1);
9289
9290 return type1;
9291 }
9292 }
9293
9294 /* A value representing the data at VALADDR/ADDRESS as described by
9295 type TYPE0, but with a standard (static-sized) type that correctly
9296 describes it. If VAL0 is not NULL and TYPE0 already is a standard
9297 type, then return VAL0 [this feature is simply to avoid redundant
9298 creation of struct values]. */
9299
9300 static struct value *
9301 ada_to_fixed_value_create (struct type *type0, CORE_ADDR address,
9302 struct value *val0)
9303 {
9304 struct type *type = ada_to_fixed_type (type0, 0, address, NULL, 1);
9305
9306 if (type == type0 && val0 != NULL)
9307 return val0;
9308
9309 if (VALUE_LVAL (val0) != lval_memory)
9310 {
9311 /* Our value does not live in memory; it could be a convenience
9312 variable, for instance. Create a not_lval value using val0's
9313 contents. */
9314 return value_from_contents (type, value_contents (val0));
9315 }
9316
9317 return value_from_contents_and_address (type, 0, address);
9318 }
9319
9320 /* A value representing VAL, but with a standard (static-sized) type
9321 that correctly describes it. Does not necessarily create a new
9322 value. */
9323
9324 struct value *
9325 ada_to_fixed_value (struct value *val)
9326 {
9327 val = unwrap_value (val);
9328 val = ada_to_fixed_value_create (value_type (val), value_address (val), val);
9329 return val;
9330 }
9331 \f
9332
9333 /* Attributes */
9334
9335 /* Table mapping attribute numbers to names.
9336 NOTE: Keep up to date with enum ada_attribute definition in ada-lang.h. */
9337
9338 static const char *attribute_names[] = {
9339 "<?>",
9340
9341 "first",
9342 "last",
9343 "length",
9344 "image",
9345 "max",
9346 "min",
9347 "modulus",
9348 "pos",
9349 "size",
9350 "tag",
9351 "val",
9352 0
9353 };
9354
9355 const char *
9356 ada_attribute_name (enum exp_opcode n)
9357 {
9358 if (n >= OP_ATR_FIRST && n <= (int) OP_ATR_VAL)
9359 return attribute_names[n - OP_ATR_FIRST + 1];
9360 else
9361 return attribute_names[0];
9362 }
9363
9364 /* Evaluate the 'POS attribute applied to ARG. */
9365
9366 static LONGEST
9367 pos_atr (struct value *arg)
9368 {
9369 struct value *val = coerce_ref (arg);
9370 struct type *type = value_type (val);
9371 LONGEST result;
9372
9373 if (!discrete_type_p (type))
9374 error (_("'POS only defined on discrete types"));
9375
9376 if (!discrete_position (type, value_as_long (val), &result))
9377 error (_("enumeration value is invalid: can't find 'POS"));
9378
9379 return result;
9380 }
9381
9382 static struct value *
9383 value_pos_atr (struct type *type, struct value *arg)
9384 {
9385 return value_from_longest (type, pos_atr (arg));
9386 }
9387
9388 /* Evaluate the TYPE'VAL attribute applied to ARG. */
9389
9390 static struct value *
9391 value_val_atr (struct type *type, struct value *arg)
9392 {
9393 if (!discrete_type_p (type))
9394 error (_("'VAL only defined on discrete types"));
9395 if (!integer_type_p (value_type (arg)))
9396 error (_("'VAL requires integral argument"));
9397
9398 if (TYPE_CODE (type) == TYPE_CODE_ENUM)
9399 {
9400 long pos = value_as_long (arg);
9401
9402 if (pos < 0 || pos >= TYPE_NFIELDS (type))
9403 error (_("argument to 'VAL out of range"));
9404 return value_from_longest (type, TYPE_FIELD_ENUMVAL (type, pos));
9405 }
9406 else
9407 return value_from_longest (type, value_as_long (arg));
9408 }
9409 \f
9410
9411 /* Evaluation */
9412
9413 /* True if TYPE appears to be an Ada character type.
9414 [At the moment, this is true only for Character and Wide_Character;
9415 It is a heuristic test that could stand improvement]. */
9416
9417 bool
9418 ada_is_character_type (struct type *type)
9419 {
9420 const char *name;
9421
9422 /* If the type code says it's a character, then assume it really is,
9423 and don't check any further. */
9424 if (TYPE_CODE (type) == TYPE_CODE_CHAR)
9425 return true;
9426
9427 /* Otherwise, assume it's a character type iff it is a discrete type
9428 with a known character type name. */
9429 name = ada_type_name (type);
9430 return (name != NULL
9431 && (TYPE_CODE (type) == TYPE_CODE_INT
9432 || TYPE_CODE (type) == TYPE_CODE_RANGE)
9433 && (strcmp (name, "character") == 0
9434 || strcmp (name, "wide_character") == 0
9435 || strcmp (name, "wide_wide_character") == 0
9436 || strcmp (name, "unsigned char") == 0));
9437 }
9438
9439 /* True if TYPE appears to be an Ada string type. */
9440
9441 bool
9442 ada_is_string_type (struct type *type)
9443 {
9444 type = ada_check_typedef (type);
9445 if (type != NULL
9446 && TYPE_CODE (type) != TYPE_CODE_PTR
9447 && (ada_is_simple_array_type (type)
9448 || ada_is_array_descriptor_type (type))
9449 && ada_array_arity (type) == 1)
9450 {
9451 struct type *elttype = ada_array_element_type (type, 1);
9452
9453 return ada_is_character_type (elttype);
9454 }
9455 else
9456 return false;
9457 }
9458
9459 /* The compiler sometimes provides a parallel XVS type for a given
9460 PAD type. Normally, it is safe to follow the PAD type directly,
9461 but older versions of the compiler have a bug that causes the offset
9462 of its "F" field to be wrong. Following that field in that case
9463 would lead to incorrect results, but this can be worked around
9464 by ignoring the PAD type and using the associated XVS type instead.
9465
9466 Set to True if the debugger should trust the contents of PAD types.
9467 Otherwise, ignore the PAD type if there is a parallel XVS type. */
9468 static int trust_pad_over_xvs = 1;
9469
9470 /* True if TYPE is a struct type introduced by the compiler to force the
9471 alignment of a value. Such types have a single field with a
9472 distinctive name. */
9473
9474 int
9475 ada_is_aligner_type (struct type *type)
9476 {
9477 type = ada_check_typedef (type);
9478
9479 if (!trust_pad_over_xvs && ada_find_parallel_type (type, "___XVS") != NULL)
9480 return 0;
9481
9482 return (TYPE_CODE (type) == TYPE_CODE_STRUCT
9483 && TYPE_NFIELDS (type) == 1
9484 && strcmp (TYPE_FIELD_NAME (type, 0), "F") == 0);
9485 }
9486
9487 /* If there is an ___XVS-convention type parallel to SUBTYPE, return
9488 the parallel type. */
9489
9490 struct type *
9491 ada_get_base_type (struct type *raw_type)
9492 {
9493 struct type *real_type_namer;
9494 struct type *raw_real_type;
9495
9496 if (raw_type == NULL || TYPE_CODE (raw_type) != TYPE_CODE_STRUCT)
9497 return raw_type;
9498
9499 if (ada_is_aligner_type (raw_type))
9500 /* The encoding specifies that we should always use the aligner type.
9501 So, even if this aligner type has an associated XVS type, we should
9502 simply ignore it.
9503
9504 According to the compiler gurus, an XVS type parallel to an aligner
9505 type may exist because of a stabs limitation. In stabs, aligner
9506 types are empty because the field has a variable-sized type, and
9507 thus cannot actually be used as an aligner type. As a result,
9508 we need the associated parallel XVS type to decode the type.
9509 Since the policy in the compiler is to not change the internal
9510 representation based on the debugging info format, we sometimes
9511 end up having a redundant XVS type parallel to the aligner type. */
9512 return raw_type;
9513
9514 real_type_namer = ada_find_parallel_type (raw_type, "___XVS");
9515 if (real_type_namer == NULL
9516 || TYPE_CODE (real_type_namer) != TYPE_CODE_STRUCT
9517 || TYPE_NFIELDS (real_type_namer) != 1)
9518 return raw_type;
9519
9520 if (TYPE_CODE (TYPE_FIELD_TYPE (real_type_namer, 0)) != TYPE_CODE_REF)
9521 {
9522 /* This is an older encoding form where the base type needs to be
9523 looked up by name. We prefer the newer enconding because it is
9524 more efficient. */
9525 raw_real_type = ada_find_any_type (TYPE_FIELD_NAME (real_type_namer, 0));
9526 if (raw_real_type == NULL)
9527 return raw_type;
9528 else
9529 return raw_real_type;
9530 }
9531
9532 /* The field in our XVS type is a reference to the base type. */
9533 return TYPE_TARGET_TYPE (TYPE_FIELD_TYPE (real_type_namer, 0));
9534 }
9535
9536 /* The type of value designated by TYPE, with all aligners removed. */
9537
9538 struct type *
9539 ada_aligned_type (struct type *type)
9540 {
9541 if (ada_is_aligner_type (type))
9542 return ada_aligned_type (TYPE_FIELD_TYPE (type, 0));
9543 else
9544 return ada_get_base_type (type);
9545 }
9546
9547
9548 /* The address of the aligned value in an object at address VALADDR
9549 having type TYPE. Assumes ada_is_aligner_type (TYPE). */
9550
9551 const gdb_byte *
9552 ada_aligned_value_addr (struct type *type, const gdb_byte *valaddr)
9553 {
9554 if (ada_is_aligner_type (type))
9555 return ada_aligned_value_addr (TYPE_FIELD_TYPE (type, 0),
9556 valaddr +
9557 TYPE_FIELD_BITPOS (type,
9558 0) / TARGET_CHAR_BIT);
9559 else
9560 return valaddr;
9561 }
9562
9563
9564
9565 /* The printed representation of an enumeration literal with encoded
9566 name NAME. The value is good to the next call of ada_enum_name. */
9567 const char *
9568 ada_enum_name (const char *name)
9569 {
9570 static char *result;
9571 static size_t result_len = 0;
9572 const char *tmp;
9573
9574 /* First, unqualify the enumeration name:
9575 1. Search for the last '.' character. If we find one, then skip
9576 all the preceding characters, the unqualified name starts
9577 right after that dot.
9578 2. Otherwise, we may be debugging on a target where the compiler
9579 translates dots into "__". Search forward for double underscores,
9580 but stop searching when we hit an overloading suffix, which is
9581 of the form "__" followed by digits. */
9582
9583 tmp = strrchr (name, '.');
9584 if (tmp != NULL)
9585 name = tmp + 1;
9586 else
9587 {
9588 while ((tmp = strstr (name, "__")) != NULL)
9589 {
9590 if (isdigit (tmp[2]))
9591 break;
9592 else
9593 name = tmp + 2;
9594 }
9595 }
9596
9597 if (name[0] == 'Q')
9598 {
9599 int v;
9600
9601 if (name[1] == 'U' || name[1] == 'W')
9602 {
9603 if (sscanf (name + 2, "%x", &v) != 1)
9604 return name;
9605 }
9606 else
9607 return name;
9608
9609 GROW_VECT (result, result_len, 16);
9610 if (isascii (v) && isprint (v))
9611 xsnprintf (result, result_len, "'%c'", v);
9612 else if (name[1] == 'U')
9613 xsnprintf (result, result_len, "[\"%02x\"]", v);
9614 else
9615 xsnprintf (result, result_len, "[\"%04x\"]", v);
9616
9617 return result;
9618 }
9619 else
9620 {
9621 tmp = strstr (name, "__");
9622 if (tmp == NULL)
9623 tmp = strstr (name, "$");
9624 if (tmp != NULL)
9625 {
9626 GROW_VECT (result, result_len, tmp - name + 1);
9627 strncpy (result, name, tmp - name);
9628 result[tmp - name] = '\0';
9629 return result;
9630 }
9631
9632 return name;
9633 }
9634 }
9635
9636 /* Evaluate the subexpression of EXP starting at *POS as for
9637 evaluate_type, updating *POS to point just past the evaluated
9638 expression. */
9639
9640 static struct value *
9641 evaluate_subexp_type (struct expression *exp, int *pos)
9642 {
9643 return evaluate_subexp (NULL_TYPE, exp, pos, EVAL_AVOID_SIDE_EFFECTS);
9644 }
9645
9646 /* If VAL is wrapped in an aligner or subtype wrapper, return the
9647 value it wraps. */
9648
9649 static struct value *
9650 unwrap_value (struct value *val)
9651 {
9652 struct type *type = ada_check_typedef (value_type (val));
9653
9654 if (ada_is_aligner_type (type))
9655 {
9656 struct value *v = ada_value_struct_elt (val, "F", 0);
9657 struct type *val_type = ada_check_typedef (value_type (v));
9658
9659 if (ada_type_name (val_type) == NULL)
9660 TYPE_NAME (val_type) = ada_type_name (type);
9661
9662 return unwrap_value (v);
9663 }
9664 else
9665 {
9666 struct type *raw_real_type =
9667 ada_check_typedef (ada_get_base_type (type));
9668
9669 /* If there is no parallel XVS or XVE type, then the value is
9670 already unwrapped. Return it without further modification. */
9671 if ((type == raw_real_type)
9672 && ada_find_parallel_type (type, "___XVE") == NULL)
9673 return val;
9674
9675 return
9676 coerce_unspec_val_to_type
9677 (val, ada_to_fixed_type (raw_real_type, 0,
9678 value_address (val),
9679 NULL, 1));
9680 }
9681 }
9682
9683 static struct value *
9684 cast_from_fixed (struct type *type, struct value *arg)
9685 {
9686 struct value *scale = ada_scaling_factor (value_type (arg));
9687 arg = value_cast (value_type (scale), arg);
9688
9689 arg = value_binop (arg, scale, BINOP_MUL);
9690 return value_cast (type, arg);
9691 }
9692
9693 static struct value *
9694 cast_to_fixed (struct type *type, struct value *arg)
9695 {
9696 if (type == value_type (arg))
9697 return arg;
9698
9699 struct value *scale = ada_scaling_factor (type);
9700 if (ada_is_fixed_point_type (value_type (arg)))
9701 arg = cast_from_fixed (value_type (scale), arg);
9702 else
9703 arg = value_cast (value_type (scale), arg);
9704
9705 arg = value_binop (arg, scale, BINOP_DIV);
9706 return value_cast (type, arg);
9707 }
9708
9709 /* Given two array types T1 and T2, return nonzero iff both arrays
9710 contain the same number of elements. */
9711
9712 static int
9713 ada_same_array_size_p (struct type *t1, struct type *t2)
9714 {
9715 LONGEST lo1, hi1, lo2, hi2;
9716
9717 /* Get the array bounds in order to verify that the size of
9718 the two arrays match. */
9719 if (!get_array_bounds (t1, &lo1, &hi1)
9720 || !get_array_bounds (t2, &lo2, &hi2))
9721 error (_("unable to determine array bounds"));
9722
9723 /* To make things easier for size comparison, normalize a bit
9724 the case of empty arrays by making sure that the difference
9725 between upper bound and lower bound is always -1. */
9726 if (lo1 > hi1)
9727 hi1 = lo1 - 1;
9728 if (lo2 > hi2)
9729 hi2 = lo2 - 1;
9730
9731 return (hi1 - lo1 == hi2 - lo2);
9732 }
9733
9734 /* Assuming that VAL is an array of integrals, and TYPE represents
9735 an array with the same number of elements, but with wider integral
9736 elements, return an array "casted" to TYPE. In practice, this
9737 means that the returned array is built by casting each element
9738 of the original array into TYPE's (wider) element type. */
9739
9740 static struct value *
9741 ada_promote_array_of_integrals (struct type *type, struct value *val)
9742 {
9743 struct type *elt_type = TYPE_TARGET_TYPE (type);
9744 LONGEST lo, hi;
9745 struct value *res;
9746 LONGEST i;
9747
9748 /* Verify that both val and type are arrays of scalars, and
9749 that the size of val's elements is smaller than the size
9750 of type's element. */
9751 gdb_assert (TYPE_CODE (type) == TYPE_CODE_ARRAY);
9752 gdb_assert (is_integral_type (TYPE_TARGET_TYPE (type)));
9753 gdb_assert (TYPE_CODE (value_type (val)) == TYPE_CODE_ARRAY);
9754 gdb_assert (is_integral_type (TYPE_TARGET_TYPE (value_type (val))));
9755 gdb_assert (TYPE_LENGTH (TYPE_TARGET_TYPE (type))
9756 > TYPE_LENGTH (TYPE_TARGET_TYPE (value_type (val))));
9757
9758 if (!get_array_bounds (type, &lo, &hi))
9759 error (_("unable to determine array bounds"));
9760
9761 res = allocate_value (type);
9762
9763 /* Promote each array element. */
9764 for (i = 0; i < hi - lo + 1; i++)
9765 {
9766 struct value *elt = value_cast (elt_type, value_subscript (val, lo + i));
9767
9768 memcpy (value_contents_writeable (res) + (i * TYPE_LENGTH (elt_type)),
9769 value_contents_all (elt), TYPE_LENGTH (elt_type));
9770 }
9771
9772 return res;
9773 }
9774
9775 /* Coerce VAL as necessary for assignment to an lval of type TYPE, and
9776 return the converted value. */
9777
9778 static struct value *
9779 coerce_for_assign (struct type *type, struct value *val)
9780 {
9781 struct type *type2 = value_type (val);
9782
9783 if (type == type2)
9784 return val;
9785
9786 type2 = ada_check_typedef (type2);
9787 type = ada_check_typedef (type);
9788
9789 if (TYPE_CODE (type2) == TYPE_CODE_PTR
9790 && TYPE_CODE (type) == TYPE_CODE_ARRAY)
9791 {
9792 val = ada_value_ind (val);
9793 type2 = value_type (val);
9794 }
9795
9796 if (TYPE_CODE (type2) == TYPE_CODE_ARRAY
9797 && TYPE_CODE (type) == TYPE_CODE_ARRAY)
9798 {
9799 if (!ada_same_array_size_p (type, type2))
9800 error (_("cannot assign arrays of different length"));
9801
9802 if (is_integral_type (TYPE_TARGET_TYPE (type))
9803 && is_integral_type (TYPE_TARGET_TYPE (type2))
9804 && TYPE_LENGTH (TYPE_TARGET_TYPE (type2))
9805 < TYPE_LENGTH (TYPE_TARGET_TYPE (type)))
9806 {
9807 /* Allow implicit promotion of the array elements to
9808 a wider type. */
9809 return ada_promote_array_of_integrals (type, val);
9810 }
9811
9812 if (TYPE_LENGTH (TYPE_TARGET_TYPE (type2))
9813 != TYPE_LENGTH (TYPE_TARGET_TYPE (type)))
9814 error (_("Incompatible types in assignment"));
9815 deprecated_set_value_type (val, type);
9816 }
9817 return val;
9818 }
9819
9820 static struct value *
9821 ada_value_binop (struct value *arg1, struct value *arg2, enum exp_opcode op)
9822 {
9823 struct value *val;
9824 struct type *type1, *type2;
9825 LONGEST v, v1, v2;
9826
9827 arg1 = coerce_ref (arg1);
9828 arg2 = coerce_ref (arg2);
9829 type1 = get_base_type (ada_check_typedef (value_type (arg1)));
9830 type2 = get_base_type (ada_check_typedef (value_type (arg2)));
9831
9832 if (TYPE_CODE (type1) != TYPE_CODE_INT
9833 || TYPE_CODE (type2) != TYPE_CODE_INT)
9834 return value_binop (arg1, arg2, op);
9835
9836 switch (op)
9837 {
9838 case BINOP_MOD:
9839 case BINOP_DIV:
9840 case BINOP_REM:
9841 break;
9842 default:
9843 return value_binop (arg1, arg2, op);
9844 }
9845
9846 v2 = value_as_long (arg2);
9847 if (v2 == 0)
9848 error (_("second operand of %s must not be zero."), op_string (op));
9849
9850 if (TYPE_UNSIGNED (type1) || op == BINOP_MOD)
9851 return value_binop (arg1, arg2, op);
9852
9853 v1 = value_as_long (arg1);
9854 switch (op)
9855 {
9856 case BINOP_DIV:
9857 v = v1 / v2;
9858 if (!TRUNCATION_TOWARDS_ZERO && v1 * (v1 % v2) < 0)
9859 v += v > 0 ? -1 : 1;
9860 break;
9861 case BINOP_REM:
9862 v = v1 % v2;
9863 if (v * v1 < 0)
9864 v -= v2;
9865 break;
9866 default:
9867 /* Should not reach this point. */
9868 v = 0;
9869 }
9870
9871 val = allocate_value (type1);
9872 store_unsigned_integer (value_contents_raw (val),
9873 TYPE_LENGTH (value_type (val)),
9874 gdbarch_byte_order (get_type_arch (type1)), v);
9875 return val;
9876 }
9877
9878 static int
9879 ada_value_equal (struct value *arg1, struct value *arg2)
9880 {
9881 if (ada_is_direct_array_type (value_type (arg1))
9882 || ada_is_direct_array_type (value_type (arg2)))
9883 {
9884 struct type *arg1_type, *arg2_type;
9885
9886 /* Automatically dereference any array reference before
9887 we attempt to perform the comparison. */
9888 arg1 = ada_coerce_ref (arg1);
9889 arg2 = ada_coerce_ref (arg2);
9890
9891 arg1 = ada_coerce_to_simple_array (arg1);
9892 arg2 = ada_coerce_to_simple_array (arg2);
9893
9894 arg1_type = ada_check_typedef (value_type (arg1));
9895 arg2_type = ada_check_typedef (value_type (arg2));
9896
9897 if (TYPE_CODE (arg1_type) != TYPE_CODE_ARRAY
9898 || TYPE_CODE (arg2_type) != TYPE_CODE_ARRAY)
9899 error (_("Attempt to compare array with non-array"));
9900 /* FIXME: The following works only for types whose
9901 representations use all bits (no padding or undefined bits)
9902 and do not have user-defined equality. */
9903 return (TYPE_LENGTH (arg1_type) == TYPE_LENGTH (arg2_type)
9904 && memcmp (value_contents (arg1), value_contents (arg2),
9905 TYPE_LENGTH (arg1_type)) == 0);
9906 }
9907 return value_equal (arg1, arg2);
9908 }
9909
9910 /* Total number of component associations in the aggregate starting at
9911 index PC in EXP. Assumes that index PC is the start of an
9912 OP_AGGREGATE. */
9913
9914 static int
9915 num_component_specs (struct expression *exp, int pc)
9916 {
9917 int n, m, i;
9918
9919 m = exp->elts[pc + 1].longconst;
9920 pc += 3;
9921 n = 0;
9922 for (i = 0; i < m; i += 1)
9923 {
9924 switch (exp->elts[pc].opcode)
9925 {
9926 default:
9927 n += 1;
9928 break;
9929 case OP_CHOICES:
9930 n += exp->elts[pc + 1].longconst;
9931 break;
9932 }
9933 ada_evaluate_subexp (NULL, exp, &pc, EVAL_SKIP);
9934 }
9935 return n;
9936 }
9937
9938 /* Assign the result of evaluating EXP starting at *POS to the INDEXth
9939 component of LHS (a simple array or a record), updating *POS past
9940 the expression, assuming that LHS is contained in CONTAINER. Does
9941 not modify the inferior's memory, nor does it modify LHS (unless
9942 LHS == CONTAINER). */
9943
9944 static void
9945 assign_component (struct value *container, struct value *lhs, LONGEST index,
9946 struct expression *exp, int *pos)
9947 {
9948 struct value *mark = value_mark ();
9949 struct value *elt;
9950 struct type *lhs_type = check_typedef (value_type (lhs));
9951
9952 if (TYPE_CODE (lhs_type) == TYPE_CODE_ARRAY)
9953 {
9954 struct type *index_type = builtin_type (exp->gdbarch)->builtin_int;
9955 struct value *index_val = value_from_longest (index_type, index);
9956
9957 elt = unwrap_value (ada_value_subscript (lhs, 1, &index_val));
9958 }
9959 else
9960 {
9961 elt = ada_index_struct_field (index, lhs, 0, value_type (lhs));
9962 elt = ada_to_fixed_value (elt);
9963 }
9964
9965 if (exp->elts[*pos].opcode == OP_AGGREGATE)
9966 assign_aggregate (container, elt, exp, pos, EVAL_NORMAL);
9967 else
9968 value_assign_to_component (container, elt,
9969 ada_evaluate_subexp (NULL, exp, pos,
9970 EVAL_NORMAL));
9971
9972 value_free_to_mark (mark);
9973 }
9974
9975 /* Assuming that LHS represents an lvalue having a record or array
9976 type, and EXP->ELTS[*POS] is an OP_AGGREGATE, evaluate an assignment
9977 of that aggregate's value to LHS, advancing *POS past the
9978 aggregate. NOSIDE is as for evaluate_subexp. CONTAINER is an
9979 lvalue containing LHS (possibly LHS itself). Does not modify
9980 the inferior's memory, nor does it modify the contents of
9981 LHS (unless == CONTAINER). Returns the modified CONTAINER. */
9982
9983 static struct value *
9984 assign_aggregate (struct value *container,
9985 struct value *lhs, struct expression *exp,
9986 int *pos, enum noside noside)
9987 {
9988 struct type *lhs_type;
9989 int n = exp->elts[*pos+1].longconst;
9990 LONGEST low_index, high_index;
9991 int num_specs;
9992 LONGEST *indices;
9993 int max_indices, num_indices;
9994 int i;
9995
9996 *pos += 3;
9997 if (noside != EVAL_NORMAL)
9998 {
9999 for (i = 0; i < n; i += 1)
10000 ada_evaluate_subexp (NULL, exp, pos, noside);
10001 return container;
10002 }
10003
10004 container = ada_coerce_ref (container);
10005 if (ada_is_direct_array_type (value_type (container)))
10006 container = ada_coerce_to_simple_array (container);
10007 lhs = ada_coerce_ref (lhs);
10008 if (!deprecated_value_modifiable (lhs))
10009 error (_("Left operand of assignment is not a modifiable lvalue."));
10010
10011 lhs_type = check_typedef (value_type (lhs));
10012 if (ada_is_direct_array_type (lhs_type))
10013 {
10014 lhs = ada_coerce_to_simple_array (lhs);
10015 lhs_type = check_typedef (value_type (lhs));
10016 low_index = TYPE_ARRAY_LOWER_BOUND_VALUE (lhs_type);
10017 high_index = TYPE_ARRAY_UPPER_BOUND_VALUE (lhs_type);
10018 }
10019 else if (TYPE_CODE (lhs_type) == TYPE_CODE_STRUCT)
10020 {
10021 low_index = 0;
10022 high_index = num_visible_fields (lhs_type) - 1;
10023 }
10024 else
10025 error (_("Left-hand side must be array or record."));
10026
10027 num_specs = num_component_specs (exp, *pos - 3);
10028 max_indices = 4 * num_specs + 4;
10029 indices = XALLOCAVEC (LONGEST, max_indices);
10030 indices[0] = indices[1] = low_index - 1;
10031 indices[2] = indices[3] = high_index + 1;
10032 num_indices = 4;
10033
10034 for (i = 0; i < n; i += 1)
10035 {
10036 switch (exp->elts[*pos].opcode)
10037 {
10038 case OP_CHOICES:
10039 aggregate_assign_from_choices (container, lhs, exp, pos, indices,
10040 &num_indices, max_indices,
10041 low_index, high_index);
10042 break;
10043 case OP_POSITIONAL:
10044 aggregate_assign_positional (container, lhs, exp, pos, indices,
10045 &num_indices, max_indices,
10046 low_index, high_index);
10047 break;
10048 case OP_OTHERS:
10049 if (i != n-1)
10050 error (_("Misplaced 'others' clause"));
10051 aggregate_assign_others (container, lhs, exp, pos, indices,
10052 num_indices, low_index, high_index);
10053 break;
10054 default:
10055 error (_("Internal error: bad aggregate clause"));
10056 }
10057 }
10058
10059 return container;
10060 }
10061
10062 /* Assign into the component of LHS indexed by the OP_POSITIONAL
10063 construct at *POS, updating *POS past the construct, given that
10064 the positions are relative to lower bound LOW, where HIGH is the
10065 upper bound. Record the position in INDICES[0 .. MAX_INDICES-1]
10066 updating *NUM_INDICES as needed. CONTAINER is as for
10067 assign_aggregate. */
10068 static void
10069 aggregate_assign_positional (struct value *container,
10070 struct value *lhs, struct expression *exp,
10071 int *pos, LONGEST *indices, int *num_indices,
10072 int max_indices, LONGEST low, LONGEST high)
10073 {
10074 LONGEST ind = longest_to_int (exp->elts[*pos + 1].longconst) + low;
10075
10076 if (ind - 1 == high)
10077 warning (_("Extra components in aggregate ignored."));
10078 if (ind <= high)
10079 {
10080 add_component_interval (ind, ind, indices, num_indices, max_indices);
10081 *pos += 3;
10082 assign_component (container, lhs, ind, exp, pos);
10083 }
10084 else
10085 ada_evaluate_subexp (NULL, exp, pos, EVAL_SKIP);
10086 }
10087
10088 /* Assign into the components of LHS indexed by the OP_CHOICES
10089 construct at *POS, updating *POS past the construct, given that
10090 the allowable indices are LOW..HIGH. Record the indices assigned
10091 to in INDICES[0 .. MAX_INDICES-1], updating *NUM_INDICES as
10092 needed. CONTAINER is as for assign_aggregate. */
10093 static void
10094 aggregate_assign_from_choices (struct value *container,
10095 struct value *lhs, struct expression *exp,
10096 int *pos, LONGEST *indices, int *num_indices,
10097 int max_indices, LONGEST low, LONGEST high)
10098 {
10099 int j;
10100 int n_choices = longest_to_int (exp->elts[*pos+1].longconst);
10101 int choice_pos, expr_pc;
10102 int is_array = ada_is_direct_array_type (value_type (lhs));
10103
10104 choice_pos = *pos += 3;
10105
10106 for (j = 0; j < n_choices; j += 1)
10107 ada_evaluate_subexp (NULL, exp, pos, EVAL_SKIP);
10108 expr_pc = *pos;
10109 ada_evaluate_subexp (NULL, exp, pos, EVAL_SKIP);
10110
10111 for (j = 0; j < n_choices; j += 1)
10112 {
10113 LONGEST lower, upper;
10114 enum exp_opcode op = exp->elts[choice_pos].opcode;
10115
10116 if (op == OP_DISCRETE_RANGE)
10117 {
10118 choice_pos += 1;
10119 lower = value_as_long (ada_evaluate_subexp (NULL, exp, pos,
10120 EVAL_NORMAL));
10121 upper = value_as_long (ada_evaluate_subexp (NULL, exp, pos,
10122 EVAL_NORMAL));
10123 }
10124 else if (is_array)
10125 {
10126 lower = value_as_long (ada_evaluate_subexp (NULL, exp, &choice_pos,
10127 EVAL_NORMAL));
10128 upper = lower;
10129 }
10130 else
10131 {
10132 int ind;
10133 const char *name;
10134
10135 switch (op)
10136 {
10137 case OP_NAME:
10138 name = &exp->elts[choice_pos + 2].string;
10139 break;
10140 case OP_VAR_VALUE:
10141 name = SYMBOL_NATURAL_NAME (exp->elts[choice_pos + 2].symbol);
10142 break;
10143 default:
10144 error (_("Invalid record component association."));
10145 }
10146 ada_evaluate_subexp (NULL, exp, &choice_pos, EVAL_SKIP);
10147 ind = 0;
10148 if (! find_struct_field (name, value_type (lhs), 0,
10149 NULL, NULL, NULL, NULL, &ind))
10150 error (_("Unknown component name: %s."), name);
10151 lower = upper = ind;
10152 }
10153
10154 if (lower <= upper && (lower < low || upper > high))
10155 error (_("Index in component association out of bounds."));
10156
10157 add_component_interval (lower, upper, indices, num_indices,
10158 max_indices);
10159 while (lower <= upper)
10160 {
10161 int pos1;
10162
10163 pos1 = expr_pc;
10164 assign_component (container, lhs, lower, exp, &pos1);
10165 lower += 1;
10166 }
10167 }
10168 }
10169
10170 /* Assign the value of the expression in the OP_OTHERS construct in
10171 EXP at *POS into the components of LHS indexed from LOW .. HIGH that
10172 have not been previously assigned. The index intervals already assigned
10173 are in INDICES[0 .. NUM_INDICES-1]. Updates *POS to after the
10174 OP_OTHERS clause. CONTAINER is as for assign_aggregate. */
10175 static void
10176 aggregate_assign_others (struct value *container,
10177 struct value *lhs, struct expression *exp,
10178 int *pos, LONGEST *indices, int num_indices,
10179 LONGEST low, LONGEST high)
10180 {
10181 int i;
10182 int expr_pc = *pos + 1;
10183
10184 for (i = 0; i < num_indices - 2; i += 2)
10185 {
10186 LONGEST ind;
10187
10188 for (ind = indices[i + 1] + 1; ind < indices[i + 2]; ind += 1)
10189 {
10190 int localpos;
10191
10192 localpos = expr_pc;
10193 assign_component (container, lhs, ind, exp, &localpos);
10194 }
10195 }
10196 ada_evaluate_subexp (NULL, exp, pos, EVAL_SKIP);
10197 }
10198
10199 /* Add the interval [LOW .. HIGH] to the sorted set of intervals
10200 [ INDICES[0] .. INDICES[1] ],..., [ INDICES[*SIZE-2] .. INDICES[*SIZE-1] ],
10201 modifying *SIZE as needed. It is an error if *SIZE exceeds
10202 MAX_SIZE. The resulting intervals do not overlap. */
10203 static void
10204 add_component_interval (LONGEST low, LONGEST high,
10205 LONGEST* indices, int *size, int max_size)
10206 {
10207 int i, j;
10208
10209 for (i = 0; i < *size; i += 2) {
10210 if (high >= indices[i] && low <= indices[i + 1])
10211 {
10212 int kh;
10213
10214 for (kh = i + 2; kh < *size; kh += 2)
10215 if (high < indices[kh])
10216 break;
10217 if (low < indices[i])
10218 indices[i] = low;
10219 indices[i + 1] = indices[kh - 1];
10220 if (high > indices[i + 1])
10221 indices[i + 1] = high;
10222 memcpy (indices + i + 2, indices + kh, *size - kh);
10223 *size -= kh - i - 2;
10224 return;
10225 }
10226 else if (high < indices[i])
10227 break;
10228 }
10229
10230 if (*size == max_size)
10231 error (_("Internal error: miscounted aggregate components."));
10232 *size += 2;
10233 for (j = *size-1; j >= i+2; j -= 1)
10234 indices[j] = indices[j - 2];
10235 indices[i] = low;
10236 indices[i + 1] = high;
10237 }
10238
10239 /* Perform and Ada cast of ARG2 to type TYPE if the type of ARG2
10240 is different. */
10241
10242 static struct value *
10243 ada_value_cast (struct type *type, struct value *arg2)
10244 {
10245 if (type == ada_check_typedef (value_type (arg2)))
10246 return arg2;
10247
10248 if (ada_is_fixed_point_type (type))
10249 return cast_to_fixed (type, arg2);
10250
10251 if (ada_is_fixed_point_type (value_type (arg2)))
10252 return cast_from_fixed (type, arg2);
10253
10254 return value_cast (type, arg2);
10255 }
10256
10257 /* Evaluating Ada expressions, and printing their result.
10258 ------------------------------------------------------
10259
10260 1. Introduction:
10261 ----------------
10262
10263 We usually evaluate an Ada expression in order to print its value.
10264 We also evaluate an expression in order to print its type, which
10265 happens during the EVAL_AVOID_SIDE_EFFECTS phase of the evaluation,
10266 but we'll focus mostly on the EVAL_NORMAL phase. In practice, the
10267 EVAL_AVOID_SIDE_EFFECTS phase allows us to simplify certain aspects of
10268 the evaluation compared to the EVAL_NORMAL, but is otherwise very
10269 similar.
10270
10271 Evaluating expressions is a little more complicated for Ada entities
10272 than it is for entities in languages such as C. The main reason for
10273 this is that Ada provides types whose definition might be dynamic.
10274 One example of such types is variant records. Or another example
10275 would be an array whose bounds can only be known at run time.
10276
10277 The following description is a general guide as to what should be
10278 done (and what should NOT be done) in order to evaluate an expression
10279 involving such types, and when. This does not cover how the semantic
10280 information is encoded by GNAT as this is covered separatly. For the
10281 document used as the reference for the GNAT encoding, see exp_dbug.ads
10282 in the GNAT sources.
10283
10284 Ideally, we should embed each part of this description next to its
10285 associated code. Unfortunately, the amount of code is so vast right
10286 now that it's hard to see whether the code handling a particular
10287 situation might be duplicated or not. One day, when the code is
10288 cleaned up, this guide might become redundant with the comments
10289 inserted in the code, and we might want to remove it.
10290
10291 2. ``Fixing'' an Entity, the Simple Case:
10292 -----------------------------------------
10293
10294 When evaluating Ada expressions, the tricky issue is that they may
10295 reference entities whose type contents and size are not statically
10296 known. Consider for instance a variant record:
10297
10298 type Rec (Empty : Boolean := True) is record
10299 case Empty is
10300 when True => null;
10301 when False => Value : Integer;
10302 end case;
10303 end record;
10304 Yes : Rec := (Empty => False, Value => 1);
10305 No : Rec := (empty => True);
10306
10307 The size and contents of that record depends on the value of the
10308 descriminant (Rec.Empty). At this point, neither the debugging
10309 information nor the associated type structure in GDB are able to
10310 express such dynamic types. So what the debugger does is to create
10311 "fixed" versions of the type that applies to the specific object.
10312 We also informally refer to this opperation as "fixing" an object,
10313 which means creating its associated fixed type.
10314
10315 Example: when printing the value of variable "Yes" above, its fixed
10316 type would look like this:
10317
10318 type Rec is record
10319 Empty : Boolean;
10320 Value : Integer;
10321 end record;
10322
10323 On the other hand, if we printed the value of "No", its fixed type
10324 would become:
10325
10326 type Rec is record
10327 Empty : Boolean;
10328 end record;
10329
10330 Things become a little more complicated when trying to fix an entity
10331 with a dynamic type that directly contains another dynamic type,
10332 such as an array of variant records, for instance. There are
10333 two possible cases: Arrays, and records.
10334
10335 3. ``Fixing'' Arrays:
10336 ---------------------
10337
10338 The type structure in GDB describes an array in terms of its bounds,
10339 and the type of its elements. By design, all elements in the array
10340 have the same type and we cannot represent an array of variant elements
10341 using the current type structure in GDB. When fixing an array,
10342 we cannot fix the array element, as we would potentially need one
10343 fixed type per element of the array. As a result, the best we can do
10344 when fixing an array is to produce an array whose bounds and size
10345 are correct (allowing us to read it from memory), but without having
10346 touched its element type. Fixing each element will be done later,
10347 when (if) necessary.
10348
10349 Arrays are a little simpler to handle than records, because the same
10350 amount of memory is allocated for each element of the array, even if
10351 the amount of space actually used by each element differs from element
10352 to element. Consider for instance the following array of type Rec:
10353
10354 type Rec_Array is array (1 .. 2) of Rec;
10355
10356 The actual amount of memory occupied by each element might be different
10357 from element to element, depending on the value of their discriminant.
10358 But the amount of space reserved for each element in the array remains
10359 fixed regardless. So we simply need to compute that size using
10360 the debugging information available, from which we can then determine
10361 the array size (we multiply the number of elements of the array by
10362 the size of each element).
10363
10364 The simplest case is when we have an array of a constrained element
10365 type. For instance, consider the following type declarations:
10366
10367 type Bounded_String (Max_Size : Integer) is
10368 Length : Integer;
10369 Buffer : String (1 .. Max_Size);
10370 end record;
10371 type Bounded_String_Array is array (1 ..2) of Bounded_String (80);
10372
10373 In this case, the compiler describes the array as an array of
10374 variable-size elements (identified by its XVS suffix) for which
10375 the size can be read in the parallel XVZ variable.
10376
10377 In the case of an array of an unconstrained element type, the compiler
10378 wraps the array element inside a private PAD type. This type should not
10379 be shown to the user, and must be "unwrap"'ed before printing. Note
10380 that we also use the adjective "aligner" in our code to designate
10381 these wrapper types.
10382
10383 In some cases, the size allocated for each element is statically
10384 known. In that case, the PAD type already has the correct size,
10385 and the array element should remain unfixed.
10386
10387 But there are cases when this size is not statically known.
10388 For instance, assuming that "Five" is an integer variable:
10389
10390 type Dynamic is array (1 .. Five) of Integer;
10391 type Wrapper (Has_Length : Boolean := False) is record
10392 Data : Dynamic;
10393 case Has_Length is
10394 when True => Length : Integer;
10395 when False => null;
10396 end case;
10397 end record;
10398 type Wrapper_Array is array (1 .. 2) of Wrapper;
10399
10400 Hello : Wrapper_Array := (others => (Has_Length => True,
10401 Data => (others => 17),
10402 Length => 1));
10403
10404
10405 The debugging info would describe variable Hello as being an
10406 array of a PAD type. The size of that PAD type is not statically
10407 known, but can be determined using a parallel XVZ variable.
10408 In that case, a copy of the PAD type with the correct size should
10409 be used for the fixed array.
10410
10411 3. ``Fixing'' record type objects:
10412 ----------------------------------
10413
10414 Things are slightly different from arrays in the case of dynamic
10415 record types. In this case, in order to compute the associated
10416 fixed type, we need to determine the size and offset of each of
10417 its components. This, in turn, requires us to compute the fixed
10418 type of each of these components.
10419
10420 Consider for instance the example:
10421
10422 type Bounded_String (Max_Size : Natural) is record
10423 Str : String (1 .. Max_Size);
10424 Length : Natural;
10425 end record;
10426 My_String : Bounded_String (Max_Size => 10);
10427
10428 In that case, the position of field "Length" depends on the size
10429 of field Str, which itself depends on the value of the Max_Size
10430 discriminant. In order to fix the type of variable My_String,
10431 we need to fix the type of field Str. Therefore, fixing a variant
10432 record requires us to fix each of its components.
10433
10434 However, if a component does not have a dynamic size, the component
10435 should not be fixed. In particular, fields that use a PAD type
10436 should not fixed. Here is an example where this might happen
10437 (assuming type Rec above):
10438
10439 type Container (Big : Boolean) is record
10440 First : Rec;
10441 After : Integer;
10442 case Big is
10443 when True => Another : Integer;
10444 when False => null;
10445 end case;
10446 end record;
10447 My_Container : Container := (Big => False,
10448 First => (Empty => True),
10449 After => 42);
10450
10451 In that example, the compiler creates a PAD type for component First,
10452 whose size is constant, and then positions the component After just
10453 right after it. The offset of component After is therefore constant
10454 in this case.
10455
10456 The debugger computes the position of each field based on an algorithm
10457 that uses, among other things, the actual position and size of the field
10458 preceding it. Let's now imagine that the user is trying to print
10459 the value of My_Container. If the type fixing was recursive, we would
10460 end up computing the offset of field After based on the size of the
10461 fixed version of field First. And since in our example First has
10462 only one actual field, the size of the fixed type is actually smaller
10463 than the amount of space allocated to that field, and thus we would
10464 compute the wrong offset of field After.
10465
10466 To make things more complicated, we need to watch out for dynamic
10467 components of variant records (identified by the ___XVL suffix in
10468 the component name). Even if the target type is a PAD type, the size
10469 of that type might not be statically known. So the PAD type needs
10470 to be unwrapped and the resulting type needs to be fixed. Otherwise,
10471 we might end up with the wrong size for our component. This can be
10472 observed with the following type declarations:
10473
10474 type Octal is new Integer range 0 .. 7;
10475 type Octal_Array is array (Positive range <>) of Octal;
10476 pragma Pack (Octal_Array);
10477
10478 type Octal_Buffer (Size : Positive) is record
10479 Buffer : Octal_Array (1 .. Size);
10480 Length : Integer;
10481 end record;
10482
10483 In that case, Buffer is a PAD type whose size is unset and needs
10484 to be computed by fixing the unwrapped type.
10485
10486 4. When to ``Fix'' un-``Fixed'' sub-elements of an entity:
10487 ----------------------------------------------------------
10488
10489 Lastly, when should the sub-elements of an entity that remained unfixed
10490 thus far, be actually fixed?
10491
10492 The answer is: Only when referencing that element. For instance
10493 when selecting one component of a record, this specific component
10494 should be fixed at that point in time. Or when printing the value
10495 of a record, each component should be fixed before its value gets
10496 printed. Similarly for arrays, the element of the array should be
10497 fixed when printing each element of the array, or when extracting
10498 one element out of that array. On the other hand, fixing should
10499 not be performed on the elements when taking a slice of an array!
10500
10501 Note that one of the side effects of miscomputing the offset and
10502 size of each field is that we end up also miscomputing the size
10503 of the containing type. This can have adverse results when computing
10504 the value of an entity. GDB fetches the value of an entity based
10505 on the size of its type, and thus a wrong size causes GDB to fetch
10506 the wrong amount of memory. In the case where the computed size is
10507 too small, GDB fetches too little data to print the value of our
10508 entity. Results in this case are unpredictable, as we usually read
10509 past the buffer containing the data =:-o. */
10510
10511 /* Evaluate a subexpression of EXP, at index *POS, and return a value
10512 for that subexpression cast to TO_TYPE. Advance *POS over the
10513 subexpression. */
10514
10515 static value *
10516 ada_evaluate_subexp_for_cast (expression *exp, int *pos,
10517 enum noside noside, struct type *to_type)
10518 {
10519 int pc = *pos;
10520
10521 if (exp->elts[pc].opcode == OP_VAR_MSYM_VALUE
10522 || exp->elts[pc].opcode == OP_VAR_VALUE)
10523 {
10524 (*pos) += 4;
10525
10526 value *val;
10527 if (exp->elts[pc].opcode == OP_VAR_MSYM_VALUE)
10528 {
10529 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10530 return value_zero (to_type, not_lval);
10531
10532 val = evaluate_var_msym_value (noside,
10533 exp->elts[pc + 1].objfile,
10534 exp->elts[pc + 2].msymbol);
10535 }
10536 else
10537 val = evaluate_var_value (noside,
10538 exp->elts[pc + 1].block,
10539 exp->elts[pc + 2].symbol);
10540
10541 if (noside == EVAL_SKIP)
10542 return eval_skip_value (exp);
10543
10544 val = ada_value_cast (to_type, val);
10545
10546 /* Follow the Ada language semantics that do not allow taking
10547 an address of the result of a cast (view conversion in Ada). */
10548 if (VALUE_LVAL (val) == lval_memory)
10549 {
10550 if (value_lazy (val))
10551 value_fetch_lazy (val);
10552 VALUE_LVAL (val) = not_lval;
10553 }
10554 return val;
10555 }
10556
10557 value *val = evaluate_subexp (to_type, exp, pos, noside);
10558 if (noside == EVAL_SKIP)
10559 return eval_skip_value (exp);
10560 return ada_value_cast (to_type, val);
10561 }
10562
10563 /* Implement the evaluate_exp routine in the exp_descriptor structure
10564 for the Ada language. */
10565
10566 static struct value *
10567 ada_evaluate_subexp (struct type *expect_type, struct expression *exp,
10568 int *pos, enum noside noside)
10569 {
10570 enum exp_opcode op;
10571 int tem;
10572 int pc;
10573 int preeval_pos;
10574 struct value *arg1 = NULL, *arg2 = NULL, *arg3;
10575 struct type *type;
10576 int nargs, oplen;
10577 struct value **argvec;
10578
10579 pc = *pos;
10580 *pos += 1;
10581 op = exp->elts[pc].opcode;
10582
10583 switch (op)
10584 {
10585 default:
10586 *pos -= 1;
10587 arg1 = evaluate_subexp_standard (expect_type, exp, pos, noside);
10588
10589 if (noside == EVAL_NORMAL)
10590 arg1 = unwrap_value (arg1);
10591
10592 /* If evaluating an OP_FLOAT and an EXPECT_TYPE was provided,
10593 then we need to perform the conversion manually, because
10594 evaluate_subexp_standard doesn't do it. This conversion is
10595 necessary in Ada because the different kinds of float/fixed
10596 types in Ada have different representations.
10597
10598 Similarly, we need to perform the conversion from OP_LONG
10599 ourselves. */
10600 if ((op == OP_FLOAT || op == OP_LONG) && expect_type != NULL)
10601 arg1 = ada_value_cast (expect_type, arg1);
10602
10603 return arg1;
10604
10605 case OP_STRING:
10606 {
10607 struct value *result;
10608
10609 *pos -= 1;
10610 result = evaluate_subexp_standard (expect_type, exp, pos, noside);
10611 /* The result type will have code OP_STRING, bashed there from
10612 OP_ARRAY. Bash it back. */
10613 if (TYPE_CODE (value_type (result)) == TYPE_CODE_STRING)
10614 TYPE_CODE (value_type (result)) = TYPE_CODE_ARRAY;
10615 return result;
10616 }
10617
10618 case UNOP_CAST:
10619 (*pos) += 2;
10620 type = exp->elts[pc + 1].type;
10621 return ada_evaluate_subexp_for_cast (exp, pos, noside, type);
10622
10623 case UNOP_QUAL:
10624 (*pos) += 2;
10625 type = exp->elts[pc + 1].type;
10626 return ada_evaluate_subexp (type, exp, pos, noside);
10627
10628 case BINOP_ASSIGN:
10629 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10630 if (exp->elts[*pos].opcode == OP_AGGREGATE)
10631 {
10632 arg1 = assign_aggregate (arg1, arg1, exp, pos, noside);
10633 if (noside == EVAL_SKIP || noside == EVAL_AVOID_SIDE_EFFECTS)
10634 return arg1;
10635 return ada_value_assign (arg1, arg1);
10636 }
10637 /* Force the evaluation of the rhs ARG2 to the type of the lhs ARG1,
10638 except if the lhs of our assignment is a convenience variable.
10639 In the case of assigning to a convenience variable, the lhs
10640 should be exactly the result of the evaluation of the rhs. */
10641 type = value_type (arg1);
10642 if (VALUE_LVAL (arg1) == lval_internalvar)
10643 type = NULL;
10644 arg2 = evaluate_subexp (type, exp, pos, noside);
10645 if (noside == EVAL_SKIP || noside == EVAL_AVOID_SIDE_EFFECTS)
10646 return arg1;
10647 if (ada_is_fixed_point_type (value_type (arg1)))
10648 arg2 = cast_to_fixed (value_type (arg1), arg2);
10649 else if (ada_is_fixed_point_type (value_type (arg2)))
10650 error
10651 (_("Fixed-point values must be assigned to fixed-point variables"));
10652 else
10653 arg2 = coerce_for_assign (value_type (arg1), arg2);
10654 return ada_value_assign (arg1, arg2);
10655
10656 case BINOP_ADD:
10657 arg1 = evaluate_subexp_with_coercion (exp, pos, noside);
10658 arg2 = evaluate_subexp_with_coercion (exp, pos, noside);
10659 if (noside == EVAL_SKIP)
10660 goto nosideret;
10661 if (TYPE_CODE (value_type (arg1)) == TYPE_CODE_PTR)
10662 return (value_from_longest
10663 (value_type (arg1),
10664 value_as_long (arg1) + value_as_long (arg2)));
10665 if (TYPE_CODE (value_type (arg2)) == TYPE_CODE_PTR)
10666 return (value_from_longest
10667 (value_type (arg2),
10668 value_as_long (arg1) + value_as_long (arg2)));
10669 if ((ada_is_fixed_point_type (value_type (arg1))
10670 || ada_is_fixed_point_type (value_type (arg2)))
10671 && value_type (arg1) != value_type (arg2))
10672 error (_("Operands of fixed-point addition must have the same type"));
10673 /* Do the addition, and cast the result to the type of the first
10674 argument. We cannot cast the result to a reference type, so if
10675 ARG1 is a reference type, find its underlying type. */
10676 type = value_type (arg1);
10677 while (TYPE_CODE (type) == TYPE_CODE_REF)
10678 type = TYPE_TARGET_TYPE (type);
10679 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10680 return value_cast (type, value_binop (arg1, arg2, BINOP_ADD));
10681
10682 case BINOP_SUB:
10683 arg1 = evaluate_subexp_with_coercion (exp, pos, noside);
10684 arg2 = evaluate_subexp_with_coercion (exp, pos, noside);
10685 if (noside == EVAL_SKIP)
10686 goto nosideret;
10687 if (TYPE_CODE (value_type (arg1)) == TYPE_CODE_PTR)
10688 return (value_from_longest
10689 (value_type (arg1),
10690 value_as_long (arg1) - value_as_long (arg2)));
10691 if (TYPE_CODE (value_type (arg2)) == TYPE_CODE_PTR)
10692 return (value_from_longest
10693 (value_type (arg2),
10694 value_as_long (arg1) - value_as_long (arg2)));
10695 if ((ada_is_fixed_point_type (value_type (arg1))
10696 || ada_is_fixed_point_type (value_type (arg2)))
10697 && value_type (arg1) != value_type (arg2))
10698 error (_("Operands of fixed-point subtraction "
10699 "must have the same type"));
10700 /* Do the substraction, and cast the result to the type of the first
10701 argument. We cannot cast the result to a reference type, so if
10702 ARG1 is a reference type, find its underlying type. */
10703 type = value_type (arg1);
10704 while (TYPE_CODE (type) == TYPE_CODE_REF)
10705 type = TYPE_TARGET_TYPE (type);
10706 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10707 return value_cast (type, value_binop (arg1, arg2, BINOP_SUB));
10708
10709 case BINOP_MUL:
10710 case BINOP_DIV:
10711 case BINOP_REM:
10712 case BINOP_MOD:
10713 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10714 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10715 if (noside == EVAL_SKIP)
10716 goto nosideret;
10717 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
10718 {
10719 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10720 return value_zero (value_type (arg1), not_lval);
10721 }
10722 else
10723 {
10724 type = builtin_type (exp->gdbarch)->builtin_double;
10725 if (ada_is_fixed_point_type (value_type (arg1)))
10726 arg1 = cast_from_fixed (type, arg1);
10727 if (ada_is_fixed_point_type (value_type (arg2)))
10728 arg2 = cast_from_fixed (type, arg2);
10729 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10730 return ada_value_binop (arg1, arg2, op);
10731 }
10732
10733 case BINOP_EQUAL:
10734 case BINOP_NOTEQUAL:
10735 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10736 arg2 = evaluate_subexp (value_type (arg1), exp, pos, noside);
10737 if (noside == EVAL_SKIP)
10738 goto nosideret;
10739 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10740 tem = 0;
10741 else
10742 {
10743 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
10744 tem = ada_value_equal (arg1, arg2);
10745 }
10746 if (op == BINOP_NOTEQUAL)
10747 tem = !tem;
10748 type = language_bool_type (exp->language_defn, exp->gdbarch);
10749 return value_from_longest (type, (LONGEST) tem);
10750
10751 case UNOP_NEG:
10752 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10753 if (noside == EVAL_SKIP)
10754 goto nosideret;
10755 else if (ada_is_fixed_point_type (value_type (arg1)))
10756 return value_cast (value_type (arg1), value_neg (arg1));
10757 else
10758 {
10759 unop_promote (exp->language_defn, exp->gdbarch, &arg1);
10760 return value_neg (arg1);
10761 }
10762
10763 case BINOP_LOGICAL_AND:
10764 case BINOP_LOGICAL_OR:
10765 case UNOP_LOGICAL_NOT:
10766 {
10767 struct value *val;
10768
10769 *pos -= 1;
10770 val = evaluate_subexp_standard (expect_type, exp, pos, noside);
10771 type = language_bool_type (exp->language_defn, exp->gdbarch);
10772 return value_cast (type, val);
10773 }
10774
10775 case BINOP_BITWISE_AND:
10776 case BINOP_BITWISE_IOR:
10777 case BINOP_BITWISE_XOR:
10778 {
10779 struct value *val;
10780
10781 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, EVAL_AVOID_SIDE_EFFECTS);
10782 *pos = pc;
10783 val = evaluate_subexp_standard (expect_type, exp, pos, noside);
10784
10785 return value_cast (value_type (arg1), val);
10786 }
10787
10788 case OP_VAR_VALUE:
10789 *pos -= 1;
10790
10791 if (noside == EVAL_SKIP)
10792 {
10793 *pos += 4;
10794 goto nosideret;
10795 }
10796
10797 if (SYMBOL_DOMAIN (exp->elts[pc + 2].symbol) == UNDEF_DOMAIN)
10798 /* Only encountered when an unresolved symbol occurs in a
10799 context other than a function call, in which case, it is
10800 invalid. */
10801 error (_("Unexpected unresolved symbol, %s, during evaluation"),
10802 SYMBOL_PRINT_NAME (exp->elts[pc + 2].symbol));
10803
10804 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10805 {
10806 type = static_unwrap_type (SYMBOL_TYPE (exp->elts[pc + 2].symbol));
10807 /* Check to see if this is a tagged type. We also need to handle
10808 the case where the type is a reference to a tagged type, but
10809 we have to be careful to exclude pointers to tagged types.
10810 The latter should be shown as usual (as a pointer), whereas
10811 a reference should mostly be transparent to the user. */
10812 if (ada_is_tagged_type (type, 0)
10813 || (TYPE_CODE (type) == TYPE_CODE_REF
10814 && ada_is_tagged_type (TYPE_TARGET_TYPE (type), 0)))
10815 {
10816 /* Tagged types are a little special in the fact that the real
10817 type is dynamic and can only be determined by inspecting the
10818 object's tag. This means that we need to get the object's
10819 value first (EVAL_NORMAL) and then extract the actual object
10820 type from its tag.
10821
10822 Note that we cannot skip the final step where we extract
10823 the object type from its tag, because the EVAL_NORMAL phase
10824 results in dynamic components being resolved into fixed ones.
10825 This can cause problems when trying to print the type
10826 description of tagged types whose parent has a dynamic size:
10827 We use the type name of the "_parent" component in order
10828 to print the name of the ancestor type in the type description.
10829 If that component had a dynamic size, the resolution into
10830 a fixed type would result in the loss of that type name,
10831 thus preventing us from printing the name of the ancestor
10832 type in the type description. */
10833 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, EVAL_NORMAL);
10834
10835 if (TYPE_CODE (type) != TYPE_CODE_REF)
10836 {
10837 struct type *actual_type;
10838
10839 actual_type = type_from_tag (ada_value_tag (arg1));
10840 if (actual_type == NULL)
10841 /* If, for some reason, we were unable to determine
10842 the actual type from the tag, then use the static
10843 approximation that we just computed as a fallback.
10844 This can happen if the debugging information is
10845 incomplete, for instance. */
10846 actual_type = type;
10847 return value_zero (actual_type, not_lval);
10848 }
10849 else
10850 {
10851 /* In the case of a ref, ada_coerce_ref takes care
10852 of determining the actual type. But the evaluation
10853 should return a ref as it should be valid to ask
10854 for its address; so rebuild a ref after coerce. */
10855 arg1 = ada_coerce_ref (arg1);
10856 return value_ref (arg1, TYPE_CODE_REF);
10857 }
10858 }
10859
10860 /* Records and unions for which GNAT encodings have been
10861 generated need to be statically fixed as well.
10862 Otherwise, non-static fixing produces a type where
10863 all dynamic properties are removed, which prevents "ptype"
10864 from being able to completely describe the type.
10865 For instance, a case statement in a variant record would be
10866 replaced by the relevant components based on the actual
10867 value of the discriminants. */
10868 if ((TYPE_CODE (type) == TYPE_CODE_STRUCT
10869 && dynamic_template_type (type) != NULL)
10870 || (TYPE_CODE (type) == TYPE_CODE_UNION
10871 && ada_find_parallel_type (type, "___XVU") != NULL))
10872 {
10873 *pos += 4;
10874 return value_zero (to_static_fixed_type (type), not_lval);
10875 }
10876 }
10877
10878 arg1 = evaluate_subexp_standard (expect_type, exp, pos, noside);
10879 return ada_to_fixed_value (arg1);
10880
10881 case OP_FUNCALL:
10882 (*pos) += 2;
10883
10884 /* Allocate arg vector, including space for the function to be
10885 called in argvec[0] and a terminating NULL. */
10886 nargs = longest_to_int (exp->elts[pc + 1].longconst);
10887 argvec = XALLOCAVEC (struct value *, nargs + 2);
10888
10889 if (exp->elts[*pos].opcode == OP_VAR_VALUE
10890 && SYMBOL_DOMAIN (exp->elts[pc + 5].symbol) == UNDEF_DOMAIN)
10891 error (_("Unexpected unresolved symbol, %s, during evaluation"),
10892 SYMBOL_PRINT_NAME (exp->elts[pc + 5].symbol));
10893 else
10894 {
10895 for (tem = 0; tem <= nargs; tem += 1)
10896 argvec[tem] = evaluate_subexp (NULL_TYPE, exp, pos, noside);
10897 argvec[tem] = 0;
10898
10899 if (noside == EVAL_SKIP)
10900 goto nosideret;
10901 }
10902
10903 if (ada_is_constrained_packed_array_type
10904 (desc_base_type (value_type (argvec[0]))))
10905 argvec[0] = ada_coerce_to_simple_array (argvec[0]);
10906 else if (TYPE_CODE (value_type (argvec[0])) == TYPE_CODE_ARRAY
10907 && TYPE_FIELD_BITSIZE (value_type (argvec[0]), 0) != 0)
10908 /* This is a packed array that has already been fixed, and
10909 therefore already coerced to a simple array. Nothing further
10910 to do. */
10911 ;
10912 else if (TYPE_CODE (value_type (argvec[0])) == TYPE_CODE_REF)
10913 {
10914 /* Make sure we dereference references so that all the code below
10915 feels like it's really handling the referenced value. Wrapping
10916 types (for alignment) may be there, so make sure we strip them as
10917 well. */
10918 argvec[0] = ada_to_fixed_value (coerce_ref (argvec[0]));
10919 }
10920 else if (TYPE_CODE (value_type (argvec[0])) == TYPE_CODE_ARRAY
10921 && VALUE_LVAL (argvec[0]) == lval_memory)
10922 argvec[0] = value_addr (argvec[0]);
10923
10924 type = ada_check_typedef (value_type (argvec[0]));
10925
10926 /* Ada allows us to implicitly dereference arrays when subscripting
10927 them. So, if this is an array typedef (encoding use for array
10928 access types encoded as fat pointers), strip it now. */
10929 if (TYPE_CODE (type) == TYPE_CODE_TYPEDEF)
10930 type = ada_typedef_target_type (type);
10931
10932 if (TYPE_CODE (type) == TYPE_CODE_PTR)
10933 {
10934 switch (TYPE_CODE (ada_check_typedef (TYPE_TARGET_TYPE (type))))
10935 {
10936 case TYPE_CODE_FUNC:
10937 type = ada_check_typedef (TYPE_TARGET_TYPE (type));
10938 break;
10939 case TYPE_CODE_ARRAY:
10940 break;
10941 case TYPE_CODE_STRUCT:
10942 if (noside != EVAL_AVOID_SIDE_EFFECTS)
10943 argvec[0] = ada_value_ind (argvec[0]);
10944 type = ada_check_typedef (TYPE_TARGET_TYPE (type));
10945 break;
10946 default:
10947 error (_("cannot subscript or call something of type `%s'"),
10948 ada_type_name (value_type (argvec[0])));
10949 break;
10950 }
10951 }
10952
10953 switch (TYPE_CODE (type))
10954 {
10955 case TYPE_CODE_FUNC:
10956 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10957 {
10958 if (TYPE_TARGET_TYPE (type) == NULL)
10959 error_call_unknown_return_type (NULL);
10960 return allocate_value (TYPE_TARGET_TYPE (type));
10961 }
10962 return call_function_by_hand (argvec[0], NULL,
10963 gdb::make_array_view (argvec + 1,
10964 nargs));
10965 case TYPE_CODE_INTERNAL_FUNCTION:
10966 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10967 /* We don't know anything about what the internal
10968 function might return, but we have to return
10969 something. */
10970 return value_zero (builtin_type (exp->gdbarch)->builtin_int,
10971 not_lval);
10972 else
10973 return call_internal_function (exp->gdbarch, exp->language_defn,
10974 argvec[0], nargs, argvec + 1);
10975
10976 case TYPE_CODE_STRUCT:
10977 {
10978 int arity;
10979
10980 arity = ada_array_arity (type);
10981 type = ada_array_element_type (type, nargs);
10982 if (type == NULL)
10983 error (_("cannot subscript or call a record"));
10984 if (arity != nargs)
10985 error (_("wrong number of subscripts; expecting %d"), arity);
10986 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10987 return value_zero (ada_aligned_type (type), lval_memory);
10988 return
10989 unwrap_value (ada_value_subscript
10990 (argvec[0], nargs, argvec + 1));
10991 }
10992 case TYPE_CODE_ARRAY:
10993 if (noside == EVAL_AVOID_SIDE_EFFECTS)
10994 {
10995 type = ada_array_element_type (type, nargs);
10996 if (type == NULL)
10997 error (_("element type of array unknown"));
10998 else
10999 return value_zero (ada_aligned_type (type), lval_memory);
11000 }
11001 return
11002 unwrap_value (ada_value_subscript
11003 (ada_coerce_to_simple_array (argvec[0]),
11004 nargs, argvec + 1));
11005 case TYPE_CODE_PTR: /* Pointer to array */
11006 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11007 {
11008 type = to_fixed_array_type (TYPE_TARGET_TYPE (type), NULL, 1);
11009 type = ada_array_element_type (type, nargs);
11010 if (type == NULL)
11011 error (_("element type of array unknown"));
11012 else
11013 return value_zero (ada_aligned_type (type), lval_memory);
11014 }
11015 return
11016 unwrap_value (ada_value_ptr_subscript (argvec[0],
11017 nargs, argvec + 1));
11018
11019 default:
11020 error (_("Attempt to index or call something other than an "
11021 "array or function"));
11022 }
11023
11024 case TERNOP_SLICE:
11025 {
11026 struct value *array = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11027 struct value *low_bound_val =
11028 evaluate_subexp (NULL_TYPE, exp, pos, noside);
11029 struct value *high_bound_val =
11030 evaluate_subexp (NULL_TYPE, exp, pos, noside);
11031 LONGEST low_bound;
11032 LONGEST high_bound;
11033
11034 low_bound_val = coerce_ref (low_bound_val);
11035 high_bound_val = coerce_ref (high_bound_val);
11036 low_bound = value_as_long (low_bound_val);
11037 high_bound = value_as_long (high_bound_val);
11038
11039 if (noside == EVAL_SKIP)
11040 goto nosideret;
11041
11042 /* If this is a reference to an aligner type, then remove all
11043 the aligners. */
11044 if (TYPE_CODE (value_type (array)) == TYPE_CODE_REF
11045 && ada_is_aligner_type (TYPE_TARGET_TYPE (value_type (array))))
11046 TYPE_TARGET_TYPE (value_type (array)) =
11047 ada_aligned_type (TYPE_TARGET_TYPE (value_type (array)));
11048
11049 if (ada_is_constrained_packed_array_type (value_type (array)))
11050 error (_("cannot slice a packed array"));
11051
11052 /* If this is a reference to an array or an array lvalue,
11053 convert to a pointer. */
11054 if (TYPE_CODE (value_type (array)) == TYPE_CODE_REF
11055 || (TYPE_CODE (value_type (array)) == TYPE_CODE_ARRAY
11056 && VALUE_LVAL (array) == lval_memory))
11057 array = value_addr (array);
11058
11059 if (noside == EVAL_AVOID_SIDE_EFFECTS
11060 && ada_is_array_descriptor_type (ada_check_typedef
11061 (value_type (array))))
11062 return empty_array (ada_type_of_array (array, 0), low_bound,
11063 high_bound);
11064
11065 array = ada_coerce_to_simple_array_ptr (array);
11066
11067 /* If we have more than one level of pointer indirection,
11068 dereference the value until we get only one level. */
11069 while (TYPE_CODE (value_type (array)) == TYPE_CODE_PTR
11070 && (TYPE_CODE (TYPE_TARGET_TYPE (value_type (array)))
11071 == TYPE_CODE_PTR))
11072 array = value_ind (array);
11073
11074 /* Make sure we really do have an array type before going further,
11075 to avoid a SEGV when trying to get the index type or the target
11076 type later down the road if the debug info generated by
11077 the compiler is incorrect or incomplete. */
11078 if (!ada_is_simple_array_type (value_type (array)))
11079 error (_("cannot take slice of non-array"));
11080
11081 if (TYPE_CODE (ada_check_typedef (value_type (array)))
11082 == TYPE_CODE_PTR)
11083 {
11084 struct type *type0 = ada_check_typedef (value_type (array));
11085
11086 if (high_bound < low_bound || noside == EVAL_AVOID_SIDE_EFFECTS)
11087 return empty_array (TYPE_TARGET_TYPE (type0), low_bound, high_bound);
11088 else
11089 {
11090 struct type *arr_type0 =
11091 to_fixed_array_type (TYPE_TARGET_TYPE (type0), NULL, 1);
11092
11093 return ada_value_slice_from_ptr (array, arr_type0,
11094 longest_to_int (low_bound),
11095 longest_to_int (high_bound));
11096 }
11097 }
11098 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11099 return array;
11100 else if (high_bound < low_bound)
11101 return empty_array (value_type (array), low_bound, high_bound);
11102 else
11103 return ada_value_slice (array, longest_to_int (low_bound),
11104 longest_to_int (high_bound));
11105 }
11106
11107 case UNOP_IN_RANGE:
11108 (*pos) += 2;
11109 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11110 type = check_typedef (exp->elts[pc + 1].type);
11111
11112 if (noside == EVAL_SKIP)
11113 goto nosideret;
11114
11115 switch (TYPE_CODE (type))
11116 {
11117 default:
11118 lim_warning (_("Membership test incompletely implemented; "
11119 "always returns true"));
11120 type = language_bool_type (exp->language_defn, exp->gdbarch);
11121 return value_from_longest (type, (LONGEST) 1);
11122
11123 case TYPE_CODE_RANGE:
11124 arg2 = value_from_longest (type, TYPE_LOW_BOUND (type));
11125 arg3 = value_from_longest (type, TYPE_HIGH_BOUND (type));
11126 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
11127 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg3);
11128 type = language_bool_type (exp->language_defn, exp->gdbarch);
11129 return
11130 value_from_longest (type,
11131 (value_less (arg1, arg3)
11132 || value_equal (arg1, arg3))
11133 && (value_less (arg2, arg1)
11134 || value_equal (arg2, arg1)));
11135 }
11136
11137 case BINOP_IN_BOUNDS:
11138 (*pos) += 2;
11139 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11140 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11141
11142 if (noside == EVAL_SKIP)
11143 goto nosideret;
11144
11145 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11146 {
11147 type = language_bool_type (exp->language_defn, exp->gdbarch);
11148 return value_zero (type, not_lval);
11149 }
11150
11151 tem = longest_to_int (exp->elts[pc + 1].longconst);
11152
11153 type = ada_index_type (value_type (arg2), tem, "range");
11154 if (!type)
11155 type = value_type (arg1);
11156
11157 arg3 = value_from_longest (type, ada_array_bound (arg2, tem, 1));
11158 arg2 = value_from_longest (type, ada_array_bound (arg2, tem, 0));
11159
11160 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
11161 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg3);
11162 type = language_bool_type (exp->language_defn, exp->gdbarch);
11163 return
11164 value_from_longest (type,
11165 (value_less (arg1, arg3)
11166 || value_equal (arg1, arg3))
11167 && (value_less (arg2, arg1)
11168 || value_equal (arg2, arg1)));
11169
11170 case TERNOP_IN_RANGE:
11171 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11172 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11173 arg3 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11174
11175 if (noside == EVAL_SKIP)
11176 goto nosideret;
11177
11178 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
11179 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg3);
11180 type = language_bool_type (exp->language_defn, exp->gdbarch);
11181 return
11182 value_from_longest (type,
11183 (value_less (arg1, arg3)
11184 || value_equal (arg1, arg3))
11185 && (value_less (arg2, arg1)
11186 || value_equal (arg2, arg1)));
11187
11188 case OP_ATR_FIRST:
11189 case OP_ATR_LAST:
11190 case OP_ATR_LENGTH:
11191 {
11192 struct type *type_arg;
11193
11194 if (exp->elts[*pos].opcode == OP_TYPE)
11195 {
11196 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11197 arg1 = NULL;
11198 type_arg = check_typedef (exp->elts[pc + 2].type);
11199 }
11200 else
11201 {
11202 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11203 type_arg = NULL;
11204 }
11205
11206 if (exp->elts[*pos].opcode != OP_LONG)
11207 error (_("Invalid operand to '%s"), ada_attribute_name (op));
11208 tem = longest_to_int (exp->elts[*pos + 2].longconst);
11209 *pos += 4;
11210
11211 if (noside == EVAL_SKIP)
11212 goto nosideret;
11213
11214 if (type_arg == NULL)
11215 {
11216 arg1 = ada_coerce_ref (arg1);
11217
11218 if (ada_is_constrained_packed_array_type (value_type (arg1)))
11219 arg1 = ada_coerce_to_simple_array (arg1);
11220
11221 if (op == OP_ATR_LENGTH)
11222 type = builtin_type (exp->gdbarch)->builtin_int;
11223 else
11224 {
11225 type = ada_index_type (value_type (arg1), tem,
11226 ada_attribute_name (op));
11227 if (type == NULL)
11228 type = builtin_type (exp->gdbarch)->builtin_int;
11229 }
11230
11231 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11232 return allocate_value (type);
11233
11234 switch (op)
11235 {
11236 default: /* Should never happen. */
11237 error (_("unexpected attribute encountered"));
11238 case OP_ATR_FIRST:
11239 return value_from_longest
11240 (type, ada_array_bound (arg1, tem, 0));
11241 case OP_ATR_LAST:
11242 return value_from_longest
11243 (type, ada_array_bound (arg1, tem, 1));
11244 case OP_ATR_LENGTH:
11245 return value_from_longest
11246 (type, ada_array_length (arg1, tem));
11247 }
11248 }
11249 else if (discrete_type_p (type_arg))
11250 {
11251 struct type *range_type;
11252 const char *name = ada_type_name (type_arg);
11253
11254 range_type = NULL;
11255 if (name != NULL && TYPE_CODE (type_arg) != TYPE_CODE_ENUM)
11256 range_type = to_fixed_range_type (type_arg, NULL);
11257 if (range_type == NULL)
11258 range_type = type_arg;
11259 switch (op)
11260 {
11261 default:
11262 error (_("unexpected attribute encountered"));
11263 case OP_ATR_FIRST:
11264 return value_from_longest
11265 (range_type, ada_discrete_type_low_bound (range_type));
11266 case OP_ATR_LAST:
11267 return value_from_longest
11268 (range_type, ada_discrete_type_high_bound (range_type));
11269 case OP_ATR_LENGTH:
11270 error (_("the 'length attribute applies only to array types"));
11271 }
11272 }
11273 else if (TYPE_CODE (type_arg) == TYPE_CODE_FLT)
11274 error (_("unimplemented type attribute"));
11275 else
11276 {
11277 LONGEST low, high;
11278
11279 if (ada_is_constrained_packed_array_type (type_arg))
11280 type_arg = decode_constrained_packed_array_type (type_arg);
11281
11282 if (op == OP_ATR_LENGTH)
11283 type = builtin_type (exp->gdbarch)->builtin_int;
11284 else
11285 {
11286 type = ada_index_type (type_arg, tem, ada_attribute_name (op));
11287 if (type == NULL)
11288 type = builtin_type (exp->gdbarch)->builtin_int;
11289 }
11290
11291 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11292 return allocate_value (type);
11293
11294 switch (op)
11295 {
11296 default:
11297 error (_("unexpected attribute encountered"));
11298 case OP_ATR_FIRST:
11299 low = ada_array_bound_from_type (type_arg, tem, 0);
11300 return value_from_longest (type, low);
11301 case OP_ATR_LAST:
11302 high = ada_array_bound_from_type (type_arg, tem, 1);
11303 return value_from_longest (type, high);
11304 case OP_ATR_LENGTH:
11305 low = ada_array_bound_from_type (type_arg, tem, 0);
11306 high = ada_array_bound_from_type (type_arg, tem, 1);
11307 return value_from_longest (type, high - low + 1);
11308 }
11309 }
11310 }
11311
11312 case OP_ATR_TAG:
11313 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11314 if (noside == EVAL_SKIP)
11315 goto nosideret;
11316
11317 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11318 return value_zero (ada_tag_type (arg1), not_lval);
11319
11320 return ada_value_tag (arg1);
11321
11322 case OP_ATR_MIN:
11323 case OP_ATR_MAX:
11324 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11325 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11326 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11327 if (noside == EVAL_SKIP)
11328 goto nosideret;
11329 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11330 return value_zero (value_type (arg1), not_lval);
11331 else
11332 {
11333 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
11334 return value_binop (arg1, arg2,
11335 op == OP_ATR_MIN ? BINOP_MIN : BINOP_MAX);
11336 }
11337
11338 case OP_ATR_MODULUS:
11339 {
11340 struct type *type_arg = check_typedef (exp->elts[pc + 2].type);
11341
11342 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11343 if (noside == EVAL_SKIP)
11344 goto nosideret;
11345
11346 if (!ada_is_modular_type (type_arg))
11347 error (_("'modulus must be applied to modular type"));
11348
11349 return value_from_longest (TYPE_TARGET_TYPE (type_arg),
11350 ada_modulus (type_arg));
11351 }
11352
11353
11354 case OP_ATR_POS:
11355 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11356 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11357 if (noside == EVAL_SKIP)
11358 goto nosideret;
11359 type = builtin_type (exp->gdbarch)->builtin_int;
11360 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11361 return value_zero (type, not_lval);
11362 else
11363 return value_pos_atr (type, arg1);
11364
11365 case OP_ATR_SIZE:
11366 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11367 type = value_type (arg1);
11368
11369 /* If the argument is a reference, then dereference its type, since
11370 the user is really asking for the size of the actual object,
11371 not the size of the pointer. */
11372 if (TYPE_CODE (type) == TYPE_CODE_REF)
11373 type = TYPE_TARGET_TYPE (type);
11374
11375 if (noside == EVAL_SKIP)
11376 goto nosideret;
11377 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11378 return value_zero (builtin_type (exp->gdbarch)->builtin_int, not_lval);
11379 else
11380 return value_from_longest (builtin_type (exp->gdbarch)->builtin_int,
11381 TARGET_CHAR_BIT * TYPE_LENGTH (type));
11382
11383 case OP_ATR_VAL:
11384 evaluate_subexp (NULL_TYPE, exp, pos, EVAL_SKIP);
11385 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11386 type = exp->elts[pc + 2].type;
11387 if (noside == EVAL_SKIP)
11388 goto nosideret;
11389 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11390 return value_zero (type, not_lval);
11391 else
11392 return value_val_atr (type, arg1);
11393
11394 case BINOP_EXP:
11395 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11396 arg2 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11397 if (noside == EVAL_SKIP)
11398 goto nosideret;
11399 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11400 return value_zero (value_type (arg1), not_lval);
11401 else
11402 {
11403 /* For integer exponentiation operations,
11404 only promote the first argument. */
11405 if (is_integral_type (value_type (arg2)))
11406 unop_promote (exp->language_defn, exp->gdbarch, &arg1);
11407 else
11408 binop_promote (exp->language_defn, exp->gdbarch, &arg1, &arg2);
11409
11410 return value_binop (arg1, arg2, op);
11411 }
11412
11413 case UNOP_PLUS:
11414 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11415 if (noside == EVAL_SKIP)
11416 goto nosideret;
11417 else
11418 return arg1;
11419
11420 case UNOP_ABS:
11421 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11422 if (noside == EVAL_SKIP)
11423 goto nosideret;
11424 unop_promote (exp->language_defn, exp->gdbarch, &arg1);
11425 if (value_less (arg1, value_zero (value_type (arg1), not_lval)))
11426 return value_neg (arg1);
11427 else
11428 return arg1;
11429
11430 case UNOP_IND:
11431 preeval_pos = *pos;
11432 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11433 if (noside == EVAL_SKIP)
11434 goto nosideret;
11435 type = ada_check_typedef (value_type (arg1));
11436 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11437 {
11438 if (ada_is_array_descriptor_type (type))
11439 /* GDB allows dereferencing GNAT array descriptors. */
11440 {
11441 struct type *arrType = ada_type_of_array (arg1, 0);
11442
11443 if (arrType == NULL)
11444 error (_("Attempt to dereference null array pointer."));
11445 return value_at_lazy (arrType, 0);
11446 }
11447 else if (TYPE_CODE (type) == TYPE_CODE_PTR
11448 || TYPE_CODE (type) == TYPE_CODE_REF
11449 /* In C you can dereference an array to get the 1st elt. */
11450 || TYPE_CODE (type) == TYPE_CODE_ARRAY)
11451 {
11452 /* As mentioned in the OP_VAR_VALUE case, tagged types can
11453 only be determined by inspecting the object's tag.
11454 This means that we need to evaluate completely the
11455 expression in order to get its type. */
11456
11457 if ((TYPE_CODE (type) == TYPE_CODE_REF
11458 || TYPE_CODE (type) == TYPE_CODE_PTR)
11459 && ada_is_tagged_type (TYPE_TARGET_TYPE (type), 0))
11460 {
11461 arg1 = evaluate_subexp (NULL_TYPE, exp, &preeval_pos,
11462 EVAL_NORMAL);
11463 type = value_type (ada_value_ind (arg1));
11464 }
11465 else
11466 {
11467 type = to_static_fixed_type
11468 (ada_aligned_type
11469 (ada_check_typedef (TYPE_TARGET_TYPE (type))));
11470 }
11471 ada_ensure_varsize_limit (type);
11472 return value_zero (type, lval_memory);
11473 }
11474 else if (TYPE_CODE (type) == TYPE_CODE_INT)
11475 {
11476 /* GDB allows dereferencing an int. */
11477 if (expect_type == NULL)
11478 return value_zero (builtin_type (exp->gdbarch)->builtin_int,
11479 lval_memory);
11480 else
11481 {
11482 expect_type =
11483 to_static_fixed_type (ada_aligned_type (expect_type));
11484 return value_zero (expect_type, lval_memory);
11485 }
11486 }
11487 else
11488 error (_("Attempt to take contents of a non-pointer value."));
11489 }
11490 arg1 = ada_coerce_ref (arg1); /* FIXME: What is this for?? */
11491 type = ada_check_typedef (value_type (arg1));
11492
11493 if (TYPE_CODE (type) == TYPE_CODE_INT)
11494 /* GDB allows dereferencing an int. If we were given
11495 the expect_type, then use that as the target type.
11496 Otherwise, assume that the target type is an int. */
11497 {
11498 if (expect_type != NULL)
11499 return ada_value_ind (value_cast (lookup_pointer_type (expect_type),
11500 arg1));
11501 else
11502 return value_at_lazy (builtin_type (exp->gdbarch)->builtin_int,
11503 (CORE_ADDR) value_as_address (arg1));
11504 }
11505
11506 if (ada_is_array_descriptor_type (type))
11507 /* GDB allows dereferencing GNAT array descriptors. */
11508 return ada_coerce_to_simple_array (arg1);
11509 else
11510 return ada_value_ind (arg1);
11511
11512 case STRUCTOP_STRUCT:
11513 tem = longest_to_int (exp->elts[pc + 1].longconst);
11514 (*pos) += 3 + BYTES_TO_EXP_ELEM (tem + 1);
11515 preeval_pos = *pos;
11516 arg1 = evaluate_subexp (NULL_TYPE, exp, pos, noside);
11517 if (noside == EVAL_SKIP)
11518 goto nosideret;
11519 if (noside == EVAL_AVOID_SIDE_EFFECTS)
11520 {
11521 struct type *type1 = value_type (arg1);
11522
11523 if (ada_is_tagged_type (type1, 1))
11524 {
11525 type = ada_lookup_struct_elt_type (type1,
11526 &exp->elts[pc + 2].string,
11527 1, 1);
11528
11529 /* If the field is not found, check if it exists in the
11530 extension of this object's type. This means that we
11531 need to evaluate completely the expression. */
11532
11533 if (type == NULL)
11534 {
11535 arg1 = evaluate_subexp (NULL_TYPE, exp, &preeval_pos,
11536 EVAL_NORMAL);
11537 arg1 = ada_value_struct_elt (arg1,
11538 &exp->elts[pc + 2].string,
11539 0);
11540 arg1 = unwrap_value (arg1);
11541 type = value_type (ada_to_fixed_value (arg1));
11542 }
11543 }
11544 else
11545 type =
11546 ada_lookup_struct_elt_type (type1, &exp->elts[pc + 2].string, 1,
11547 0);
11548
11549 return value_zero (ada_aligned_type (type), lval_memory);
11550 }
11551 else
11552 {
11553 arg1 = ada_value_struct_elt (arg1, &exp->elts[pc + 2].string, 0);
11554 arg1 = unwrap_value (arg1);
11555 return ada_to_fixed_value (arg1);
11556 }
11557
11558 case OP_TYPE:
11559 /* The value is not supposed to be used. This is here to make it
11560 easier to accommodate expressions that contain types. */
11561 (*pos) += 2;
11562 if (noside == EVAL_SKIP)
11563 goto nosideret;
11564 else if (noside == EVAL_AVOID_SIDE_EFFECTS)
11565 return allocate_value (exp->elts[pc + 1].type);
11566 else
11567 error (_("Attempt to use a type name as an expression"));
11568
11569 case OP_AGGREGATE:
11570 case OP_CHOICES:
11571 case OP_OTHERS:
11572 case OP_DISCRETE_RANGE:
11573 case OP_POSITIONAL:
11574 case OP_NAME:
11575 if (noside == EVAL_NORMAL)
11576 switch (op)
11577 {
11578 case OP_NAME:
11579 error (_("Undefined name, ambiguous name, or renaming used in "
11580 "component association: %s."), &exp->elts[pc+2].string);
11581 case OP_AGGREGATE:
11582 error (_("Aggregates only allowed on the right of an assignment"));
11583 default:
11584 internal_error (__FILE__, __LINE__,
11585 _("aggregate apparently mangled"));
11586 }
11587
11588 ada_forward_operator_length (exp, pc, &oplen, &nargs);
11589 *pos += oplen - 1;
11590 for (tem = 0; tem < nargs; tem += 1)
11591 ada_evaluate_subexp (NULL, exp, pos, noside);
11592 goto nosideret;
11593 }
11594
11595 nosideret:
11596 return eval_skip_value (exp);
11597 }
11598 \f
11599
11600 /* Fixed point */
11601
11602 /* If TYPE encodes an Ada fixed-point type, return the suffix of the
11603 type name that encodes the 'small and 'delta information.
11604 Otherwise, return NULL. */
11605
11606 static const char *
11607 fixed_type_info (struct type *type)
11608 {
11609 const char *name = ada_type_name (type);
11610 enum type_code code = (type == NULL) ? TYPE_CODE_UNDEF : TYPE_CODE (type);
11611
11612 if ((code == TYPE_CODE_INT || code == TYPE_CODE_RANGE) && name != NULL)
11613 {
11614 const char *tail = strstr (name, "___XF_");
11615
11616 if (tail == NULL)
11617 return NULL;
11618 else
11619 return tail + 5;
11620 }
11621 else if (code == TYPE_CODE_RANGE && TYPE_TARGET_TYPE (type) != type)
11622 return fixed_type_info (TYPE_TARGET_TYPE (type));
11623 else
11624 return NULL;
11625 }
11626
11627 /* Returns non-zero iff TYPE represents an Ada fixed-point type. */
11628
11629 int
11630 ada_is_fixed_point_type (struct type *type)
11631 {
11632 return fixed_type_info (type) != NULL;
11633 }
11634
11635 /* Return non-zero iff TYPE represents a System.Address type. */
11636
11637 int
11638 ada_is_system_address_type (struct type *type)
11639 {
11640 return (TYPE_NAME (type)
11641 && strcmp (TYPE_NAME (type), "system__address") == 0);
11642 }
11643
11644 /* Assuming that TYPE is the representation of an Ada fixed-point
11645 type, return the target floating-point type to be used to represent
11646 of this type during internal computation. */
11647
11648 static struct type *
11649 ada_scaling_type (struct type *type)
11650 {
11651 return builtin_type (get_type_arch (type))->builtin_long_double;
11652 }
11653
11654 /* Assuming that TYPE is the representation of an Ada fixed-point
11655 type, return its delta, or NULL if the type is malformed and the
11656 delta cannot be determined. */
11657
11658 struct value *
11659 ada_delta (struct type *type)
11660 {
11661 const char *encoding = fixed_type_info (type);
11662 struct type *scale_type = ada_scaling_type (type);
11663
11664 long long num, den;
11665
11666 if (sscanf (encoding, "_%lld_%lld", &num, &den) < 2)
11667 return nullptr;
11668 else
11669 return value_binop (value_from_longest (scale_type, num),
11670 value_from_longest (scale_type, den), BINOP_DIV);
11671 }
11672
11673 /* Assuming that ada_is_fixed_point_type (TYPE), return the scaling
11674 factor ('SMALL value) associated with the type. */
11675
11676 struct value *
11677 ada_scaling_factor (struct type *type)
11678 {
11679 const char *encoding = fixed_type_info (type);
11680 struct type *scale_type = ada_scaling_type (type);
11681
11682 long long num0, den0, num1, den1;
11683 int n;
11684
11685 n = sscanf (encoding, "_%lld_%lld_%lld_%lld",
11686 &num0, &den0, &num1, &den1);
11687
11688 if (n < 2)
11689 return value_from_longest (scale_type, 1);
11690 else if (n == 4)
11691 return value_binop (value_from_longest (scale_type, num1),
11692 value_from_longest (scale_type, den1), BINOP_DIV);
11693 else
11694 return value_binop (value_from_longest (scale_type, num0),
11695 value_from_longest (scale_type, den0), BINOP_DIV);
11696 }
11697
11698 \f
11699
11700 /* Range types */
11701
11702 /* Scan STR beginning at position K for a discriminant name, and
11703 return the value of that discriminant field of DVAL in *PX. If
11704 PNEW_K is not null, put the position of the character beyond the
11705 name scanned in *PNEW_K. Return 1 if successful; return 0 and do
11706 not alter *PX and *PNEW_K if unsuccessful. */
11707
11708 static int
11709 scan_discrim_bound (const char *str, int k, struct value *dval, LONGEST * px,
11710 int *pnew_k)
11711 {
11712 static char *bound_buffer = NULL;
11713 static size_t bound_buffer_len = 0;
11714 const char *pstart, *pend, *bound;
11715 struct value *bound_val;
11716
11717 if (dval == NULL || str == NULL || str[k] == '\0')
11718 return 0;
11719
11720 pstart = str + k;
11721 pend = strstr (pstart, "__");
11722 if (pend == NULL)
11723 {
11724 bound = pstart;
11725 k += strlen (bound);
11726 }
11727 else
11728 {
11729 int len = pend - pstart;
11730
11731 /* Strip __ and beyond. */
11732 GROW_VECT (bound_buffer, bound_buffer_len, len + 1);
11733 strncpy (bound_buffer, pstart, len);
11734 bound_buffer[len] = '\0';
11735
11736 bound = bound_buffer;
11737 k = pend - str;
11738 }
11739
11740 bound_val = ada_search_struct_field (bound, dval, 0, value_type (dval));
11741 if (bound_val == NULL)
11742 return 0;
11743
11744 *px = value_as_long (bound_val);
11745 if (pnew_k != NULL)
11746 *pnew_k = k;
11747 return 1;
11748 }
11749
11750 /* Value of variable named NAME in the current environment. If
11751 no such variable found, then if ERR_MSG is null, returns 0, and
11752 otherwise causes an error with message ERR_MSG. */
11753
11754 static struct value *
11755 get_var_value (const char *name, const char *err_msg)
11756 {
11757 lookup_name_info lookup_name (name, symbol_name_match_type::FULL);
11758
11759 std::vector<struct block_symbol> syms;
11760 int nsyms = ada_lookup_symbol_list_worker (lookup_name,
11761 get_selected_block (0),
11762 VAR_DOMAIN, &syms, 1);
11763
11764 if (nsyms != 1)
11765 {
11766 if (err_msg == NULL)
11767 return 0;
11768 else
11769 error (("%s"), err_msg);
11770 }
11771
11772 return value_of_variable (syms[0].symbol, syms[0].block);
11773 }
11774
11775 /* Value of integer variable named NAME in the current environment.
11776 If no such variable is found, returns false. Otherwise, sets VALUE
11777 to the variable's value and returns true. */
11778
11779 bool
11780 get_int_var_value (const char *name, LONGEST &value)
11781 {
11782 struct value *var_val = get_var_value (name, 0);
11783
11784 if (var_val == 0)
11785 return false;
11786
11787 value = value_as_long (var_val);
11788 return true;
11789 }
11790
11791
11792 /* Return a range type whose base type is that of the range type named
11793 NAME in the current environment, and whose bounds are calculated
11794 from NAME according to the GNAT range encoding conventions.
11795 Extract discriminant values, if needed, from DVAL. ORIG_TYPE is the
11796 corresponding range type from debug information; fall back to using it
11797 if symbol lookup fails. If a new type must be created, allocate it
11798 like ORIG_TYPE was. The bounds information, in general, is encoded
11799 in NAME, the base type given in the named range type. */
11800
11801 static struct type *
11802 to_fixed_range_type (struct type *raw_type, struct value *dval)
11803 {
11804 const char *name;
11805 struct type *base_type;
11806 const char *subtype_info;
11807
11808 gdb_assert (raw_type != NULL);
11809 gdb_assert (TYPE_NAME (raw_type) != NULL);
11810
11811 if (TYPE_CODE (raw_type) == TYPE_CODE_RANGE)
11812 base_type = TYPE_TARGET_TYPE (raw_type);
11813 else
11814 base_type = raw_type;
11815
11816 name = TYPE_NAME (raw_type);
11817 subtype_info = strstr (name, "___XD");
11818 if (subtype_info == NULL)
11819 {
11820 LONGEST L = ada_discrete_type_low_bound (raw_type);
11821 LONGEST U = ada_discrete_type_high_bound (raw_type);
11822
11823 if (L < INT_MIN || U > INT_MAX)
11824 return raw_type;
11825 else
11826 return create_static_range_type (alloc_type_copy (raw_type), raw_type,
11827 L, U);
11828 }
11829 else
11830 {
11831 static char *name_buf = NULL;
11832 static size_t name_len = 0;
11833 int prefix_len = subtype_info - name;
11834 LONGEST L, U;
11835 struct type *type;
11836 const char *bounds_str;
11837 int n;
11838
11839 GROW_VECT (name_buf, name_len, prefix_len + 5);
11840 strncpy (name_buf, name, prefix_len);
11841 name_buf[prefix_len] = '\0';
11842
11843 subtype_info += 5;
11844 bounds_str = strchr (subtype_info, '_');
11845 n = 1;
11846
11847 if (*subtype_info == 'L')
11848 {
11849 if (!ada_scan_number (bounds_str, n, &L, &n)
11850 && !scan_discrim_bound (bounds_str, n, dval, &L, &n))
11851 return raw_type;
11852 if (bounds_str[n] == '_')
11853 n += 2;
11854 else if (bounds_str[n] == '.') /* FIXME? SGI Workshop kludge. */
11855 n += 1;
11856 subtype_info += 1;
11857 }
11858 else
11859 {
11860 strcpy (name_buf + prefix_len, "___L");
11861 if (!get_int_var_value (name_buf, L))
11862 {
11863 lim_warning (_("Unknown lower bound, using 1."));
11864 L = 1;
11865 }
11866 }
11867
11868 if (*subtype_info == 'U')
11869 {
11870 if (!ada_scan_number (bounds_str, n, &U, &n)
11871 && !scan_discrim_bound (bounds_str, n, dval, &U, &n))
11872 return raw_type;
11873 }
11874 else
11875 {
11876 strcpy (name_buf + prefix_len, "___U");
11877 if (!get_int_var_value (name_buf, U))
11878 {
11879 lim_warning (_("Unknown upper bound, using %ld."), (long) L);
11880 U = L;
11881 }
11882 }
11883
11884 type = create_static_range_type (alloc_type_copy (raw_type),
11885 base_type, L, U);
11886 /* create_static_range_type alters the resulting type's length
11887 to match the size of the base_type, which is not what we want.
11888 Set it back to the original range type's length. */
11889 TYPE_LENGTH (type) = TYPE_LENGTH (raw_type);
11890 TYPE_NAME (type) = name;
11891 return type;
11892 }
11893 }
11894
11895 /* True iff NAME is the name of a range type. */
11896
11897 int
11898 ada_is_range_type_name (const char *name)
11899 {
11900 return (name != NULL && strstr (name, "___XD"));
11901 }
11902 \f
11903
11904 /* Modular types */
11905
11906 /* True iff TYPE is an Ada modular type. */
11907
11908 int
11909 ada_is_modular_type (struct type *type)
11910 {
11911 struct type *subranged_type = get_base_type (type);
11912
11913 return (subranged_type != NULL && TYPE_CODE (type) == TYPE_CODE_RANGE
11914 && TYPE_CODE (subranged_type) == TYPE_CODE_INT
11915 && TYPE_UNSIGNED (subranged_type));
11916 }
11917
11918 /* Assuming ada_is_modular_type (TYPE), the modulus of TYPE. */
11919
11920 ULONGEST
11921 ada_modulus (struct type *type)
11922 {
11923 return (ULONGEST) TYPE_HIGH_BOUND (type) + 1;
11924 }
11925 \f
11926
11927 /* Ada exception catchpoint support:
11928 ---------------------------------
11929
11930 We support 3 kinds of exception catchpoints:
11931 . catchpoints on Ada exceptions
11932 . catchpoints on unhandled Ada exceptions
11933 . catchpoints on failed assertions
11934
11935 Exceptions raised during failed assertions, or unhandled exceptions
11936 could perfectly be caught with the general catchpoint on Ada exceptions.
11937 However, we can easily differentiate these two special cases, and having
11938 the option to distinguish these two cases from the rest can be useful
11939 to zero-in on certain situations.
11940
11941 Exception catchpoints are a specialized form of breakpoint,
11942 since they rely on inserting breakpoints inside known routines
11943 of the GNAT runtime. The implementation therefore uses a standard
11944 breakpoint structure of the BP_BREAKPOINT type, but with its own set
11945 of breakpoint_ops.
11946
11947 Support in the runtime for exception catchpoints have been changed
11948 a few times already, and these changes affect the implementation
11949 of these catchpoints. In order to be able to support several
11950 variants of the runtime, we use a sniffer that will determine
11951 the runtime variant used by the program being debugged. */
11952
11953 /* Ada's standard exceptions.
11954
11955 The Ada 83 standard also defined Numeric_Error. But there so many
11956 situations where it was unclear from the Ada 83 Reference Manual
11957 (RM) whether Constraint_Error or Numeric_Error should be raised,
11958 that the ARG (Ada Rapporteur Group) eventually issued a Binding
11959 Interpretation saying that anytime the RM says that Numeric_Error
11960 should be raised, the implementation may raise Constraint_Error.
11961 Ada 95 went one step further and pretty much removed Numeric_Error
11962 from the list of standard exceptions (it made it a renaming of
11963 Constraint_Error, to help preserve compatibility when compiling
11964 an Ada83 compiler). As such, we do not include Numeric_Error from
11965 this list of standard exceptions. */
11966
11967 static const char *standard_exc[] = {
11968 "constraint_error",
11969 "program_error",
11970 "storage_error",
11971 "tasking_error"
11972 };
11973
11974 typedef CORE_ADDR (ada_unhandled_exception_name_addr_ftype) (void);
11975
11976 /* A structure that describes how to support exception catchpoints
11977 for a given executable. */
11978
11979 struct exception_support_info
11980 {
11981 /* The name of the symbol to break on in order to insert
11982 a catchpoint on exceptions. */
11983 const char *catch_exception_sym;
11984
11985 /* The name of the symbol to break on in order to insert
11986 a catchpoint on unhandled exceptions. */
11987 const char *catch_exception_unhandled_sym;
11988
11989 /* The name of the symbol to break on in order to insert
11990 a catchpoint on failed assertions. */
11991 const char *catch_assert_sym;
11992
11993 /* The name of the symbol to break on in order to insert
11994 a catchpoint on exception handling. */
11995 const char *catch_handlers_sym;
11996
11997 /* Assuming that the inferior just triggered an unhandled exception
11998 catchpoint, this function is responsible for returning the address
11999 in inferior memory where the name of that exception is stored.
12000 Return zero if the address could not be computed. */
12001 ada_unhandled_exception_name_addr_ftype *unhandled_exception_name_addr;
12002 };
12003
12004 static CORE_ADDR ada_unhandled_exception_name_addr (void);
12005 static CORE_ADDR ada_unhandled_exception_name_addr_from_raise (void);
12006
12007 /* The following exception support info structure describes how to
12008 implement exception catchpoints with the latest version of the
12009 Ada runtime (as of 2007-03-06). */
12010
12011 static const struct exception_support_info default_exception_support_info =
12012 {
12013 "__gnat_debug_raise_exception", /* catch_exception_sym */
12014 "__gnat_unhandled_exception", /* catch_exception_unhandled_sym */
12015 "__gnat_debug_raise_assert_failure", /* catch_assert_sym */
12016 "__gnat_begin_handler", /* catch_handlers_sym */
12017 ada_unhandled_exception_name_addr
12018 };
12019
12020 /* The following exception support info structure describes how to
12021 implement exception catchpoints with a slightly older version
12022 of the Ada runtime. */
12023
12024 static const struct exception_support_info exception_support_info_fallback =
12025 {
12026 "__gnat_raise_nodefer_with_msg", /* catch_exception_sym */
12027 "__gnat_unhandled_exception", /* catch_exception_unhandled_sym */
12028 "system__assertions__raise_assert_failure", /* catch_assert_sym */
12029 "__gnat_begin_handler", /* catch_handlers_sym */
12030 ada_unhandled_exception_name_addr_from_raise
12031 };
12032
12033 /* Return nonzero if we can detect the exception support routines
12034 described in EINFO.
12035
12036 This function errors out if an abnormal situation is detected
12037 (for instance, if we find the exception support routines, but
12038 that support is found to be incomplete). */
12039
12040 static int
12041 ada_has_this_exception_support (const struct exception_support_info *einfo)
12042 {
12043 struct symbol *sym;
12044
12045 /* The symbol we're looking up is provided by a unit in the GNAT runtime
12046 that should be compiled with debugging information. As a result, we
12047 expect to find that symbol in the symtabs. */
12048
12049 sym = standard_lookup (einfo->catch_exception_sym, NULL, VAR_DOMAIN);
12050 if (sym == NULL)
12051 {
12052 /* Perhaps we did not find our symbol because the Ada runtime was
12053 compiled without debugging info, or simply stripped of it.
12054 It happens on some GNU/Linux distributions for instance, where
12055 users have to install a separate debug package in order to get
12056 the runtime's debugging info. In that situation, let the user
12057 know why we cannot insert an Ada exception catchpoint.
12058
12059 Note: Just for the purpose of inserting our Ada exception
12060 catchpoint, we could rely purely on the associated minimal symbol.
12061 But we would be operating in degraded mode anyway, since we are
12062 still lacking the debugging info needed later on to extract
12063 the name of the exception being raised (this name is printed in
12064 the catchpoint message, and is also used when trying to catch
12065 a specific exception). We do not handle this case for now. */
12066 struct bound_minimal_symbol msym
12067 = lookup_minimal_symbol (einfo->catch_exception_sym, NULL, NULL);
12068
12069 if (msym.minsym && MSYMBOL_TYPE (msym.minsym) != mst_solib_trampoline)
12070 error (_("Your Ada runtime appears to be missing some debugging "
12071 "information.\nCannot insert Ada exception catchpoint "
12072 "in this configuration."));
12073
12074 return 0;
12075 }
12076
12077 /* Make sure that the symbol we found corresponds to a function. */
12078
12079 if (SYMBOL_CLASS (sym) != LOC_BLOCK)
12080 error (_("Symbol \"%s\" is not a function (class = %d)"),
12081 SYMBOL_LINKAGE_NAME (sym), SYMBOL_CLASS (sym));
12082
12083 return 1;
12084 }
12085
12086 /* Inspect the Ada runtime and determine which exception info structure
12087 should be used to provide support for exception catchpoints.
12088
12089 This function will always set the per-inferior exception_info,
12090 or raise an error. */
12091
12092 static void
12093 ada_exception_support_info_sniffer (void)
12094 {
12095 struct ada_inferior_data *data = get_ada_inferior_data (current_inferior ());
12096
12097 /* If the exception info is already known, then no need to recompute it. */
12098 if (data->exception_info != NULL)
12099 return;
12100
12101 /* Check the latest (default) exception support info. */
12102 if (ada_has_this_exception_support (&default_exception_support_info))
12103 {
12104 data->exception_info = &default_exception_support_info;
12105 return;
12106 }
12107
12108 /* Try our fallback exception suport info. */
12109 if (ada_has_this_exception_support (&exception_support_info_fallback))
12110 {
12111 data->exception_info = &exception_support_info_fallback;
12112 return;
12113 }
12114
12115 /* Sometimes, it is normal for us to not be able to find the routine
12116 we are looking for. This happens when the program is linked with
12117 the shared version of the GNAT runtime, and the program has not been
12118 started yet. Inform the user of these two possible causes if
12119 applicable. */
12120
12121 if (ada_update_initial_language (language_unknown) != language_ada)
12122 error (_("Unable to insert catchpoint. Is this an Ada main program?"));
12123
12124 /* If the symbol does not exist, then check that the program is
12125 already started, to make sure that shared libraries have been
12126 loaded. If it is not started, this may mean that the symbol is
12127 in a shared library. */
12128
12129 if (inferior_ptid.pid () == 0)
12130 error (_("Unable to insert catchpoint. Try to start the program first."));
12131
12132 /* At this point, we know that we are debugging an Ada program and
12133 that the inferior has been started, but we still are not able to
12134 find the run-time symbols. That can mean that we are in
12135 configurable run time mode, or that a-except as been optimized
12136 out by the linker... In any case, at this point it is not worth
12137 supporting this feature. */
12138
12139 error (_("Cannot insert Ada exception catchpoints in this configuration."));
12140 }
12141
12142 /* True iff FRAME is very likely to be that of a function that is
12143 part of the runtime system. This is all very heuristic, but is
12144 intended to be used as advice as to what frames are uninteresting
12145 to most users. */
12146
12147 static int
12148 is_known_support_routine (struct frame_info *frame)
12149 {
12150 enum language func_lang;
12151 int i;
12152 const char *fullname;
12153
12154 /* If this code does not have any debugging information (no symtab),
12155 This cannot be any user code. */
12156
12157 symtab_and_line sal = find_frame_sal (frame);
12158 if (sal.symtab == NULL)
12159 return 1;
12160
12161 /* If there is a symtab, but the associated source file cannot be
12162 located, then assume this is not user code: Selecting a frame
12163 for which we cannot display the code would not be very helpful
12164 for the user. This should also take care of case such as VxWorks
12165 where the kernel has some debugging info provided for a few units. */
12166
12167 fullname = symtab_to_fullname (sal.symtab);
12168 if (access (fullname, R_OK) != 0)
12169 return 1;
12170
12171 /* Check the unit filename againt the Ada runtime file naming.
12172 We also check the name of the objfile against the name of some
12173 known system libraries that sometimes come with debugging info
12174 too. */
12175
12176 for (i = 0; known_runtime_file_name_patterns[i] != NULL; i += 1)
12177 {
12178 re_comp (known_runtime_file_name_patterns[i]);
12179 if (re_exec (lbasename (sal.symtab->filename)))
12180 return 1;
12181 if (SYMTAB_OBJFILE (sal.symtab) != NULL
12182 && re_exec (objfile_name (SYMTAB_OBJFILE (sal.symtab))))
12183 return 1;
12184 }
12185
12186 /* Check whether the function is a GNAT-generated entity. */
12187
12188 gdb::unique_xmalloc_ptr<char> func_name
12189 = find_frame_funname (frame, &func_lang, NULL);
12190 if (func_name == NULL)
12191 return 1;
12192
12193 for (i = 0; known_auxiliary_function_name_patterns[i] != NULL; i += 1)
12194 {
12195 re_comp (known_auxiliary_function_name_patterns[i]);
12196 if (re_exec (func_name.get ()))
12197 return 1;
12198 }
12199
12200 return 0;
12201 }
12202
12203 /* Find the first frame that contains debugging information and that is not
12204 part of the Ada run-time, starting from FI and moving upward. */
12205
12206 void
12207 ada_find_printable_frame (struct frame_info *fi)
12208 {
12209 for (; fi != NULL; fi = get_prev_frame (fi))
12210 {
12211 if (!is_known_support_routine (fi))
12212 {
12213 select_frame (fi);
12214 break;
12215 }
12216 }
12217
12218 }
12219
12220 /* Assuming that the inferior just triggered an unhandled exception
12221 catchpoint, return the address in inferior memory where the name
12222 of the exception is stored.
12223
12224 Return zero if the address could not be computed. */
12225
12226 static CORE_ADDR
12227 ada_unhandled_exception_name_addr (void)
12228 {
12229 return parse_and_eval_address ("e.full_name");
12230 }
12231
12232 /* Same as ada_unhandled_exception_name_addr, except that this function
12233 should be used when the inferior uses an older version of the runtime,
12234 where the exception name needs to be extracted from a specific frame
12235 several frames up in the callstack. */
12236
12237 static CORE_ADDR
12238 ada_unhandled_exception_name_addr_from_raise (void)
12239 {
12240 int frame_level;
12241 struct frame_info *fi;
12242 struct ada_inferior_data *data = get_ada_inferior_data (current_inferior ());
12243
12244 /* To determine the name of this exception, we need to select
12245 the frame corresponding to RAISE_SYM_NAME. This frame is
12246 at least 3 levels up, so we simply skip the first 3 frames
12247 without checking the name of their associated function. */
12248 fi = get_current_frame ();
12249 for (frame_level = 0; frame_level < 3; frame_level += 1)
12250 if (fi != NULL)
12251 fi = get_prev_frame (fi);
12252
12253 while (fi != NULL)
12254 {
12255 enum language func_lang;
12256
12257 gdb::unique_xmalloc_ptr<char> func_name
12258 = find_frame_funname (fi, &func_lang, NULL);
12259 if (func_name != NULL)
12260 {
12261 if (strcmp (func_name.get (),
12262 data->exception_info->catch_exception_sym) == 0)
12263 break; /* We found the frame we were looking for... */
12264 }
12265 fi = get_prev_frame (fi);
12266 }
12267
12268 if (fi == NULL)
12269 return 0;
12270
12271 select_frame (fi);
12272 return parse_and_eval_address ("id.full_name");
12273 }
12274
12275 /* Assuming the inferior just triggered an Ada exception catchpoint
12276 (of any type), return the address in inferior memory where the name
12277 of the exception is stored, if applicable.
12278
12279 Assumes the selected frame is the current frame.
12280
12281 Return zero if the address could not be computed, or if not relevant. */
12282
12283 static CORE_ADDR
12284 ada_exception_name_addr_1 (enum ada_exception_catchpoint_kind ex,
12285 struct breakpoint *b)
12286 {
12287 struct ada_inferior_data *data = get_ada_inferior_data (current_inferior ());
12288
12289 switch (ex)
12290 {
12291 case ada_catch_exception:
12292 return (parse_and_eval_address ("e.full_name"));
12293 break;
12294
12295 case ada_catch_exception_unhandled:
12296 return data->exception_info->unhandled_exception_name_addr ();
12297 break;
12298
12299 case ada_catch_handlers:
12300 return 0; /* The runtimes does not provide access to the exception
12301 name. */
12302 break;
12303
12304 case ada_catch_assert:
12305 return 0; /* Exception name is not relevant in this case. */
12306 break;
12307
12308 default:
12309 internal_error (__FILE__, __LINE__, _("unexpected catchpoint type"));
12310 break;
12311 }
12312
12313 return 0; /* Should never be reached. */
12314 }
12315
12316 /* Assuming the inferior is stopped at an exception catchpoint,
12317 return the message which was associated to the exception, if
12318 available. Return NULL if the message could not be retrieved.
12319
12320 Note: The exception message can be associated to an exception
12321 either through the use of the Raise_Exception function, or
12322 more simply (Ada 2005 and later), via:
12323
12324 raise Exception_Name with "exception message";
12325
12326 */
12327
12328 static gdb::unique_xmalloc_ptr<char>
12329 ada_exception_message_1 (void)
12330 {
12331 struct value *e_msg_val;
12332 int e_msg_len;
12333
12334 /* For runtimes that support this feature, the exception message
12335 is passed as an unbounded string argument called "message". */
12336 e_msg_val = parse_and_eval ("message");
12337 if (e_msg_val == NULL)
12338 return NULL; /* Exception message not supported. */
12339
12340 e_msg_val = ada_coerce_to_simple_array (e_msg_val);
12341 gdb_assert (e_msg_val != NULL);
12342 e_msg_len = TYPE_LENGTH (value_type (e_msg_val));
12343
12344 /* If the message string is empty, then treat it as if there was
12345 no exception message. */
12346 if (e_msg_len <= 0)
12347 return NULL;
12348
12349 gdb::unique_xmalloc_ptr<char> e_msg ((char *) xmalloc (e_msg_len + 1));
12350 read_memory_string (value_address (e_msg_val), e_msg.get (), e_msg_len + 1);
12351 e_msg.get ()[e_msg_len] = '\0';
12352
12353 return e_msg;
12354 }
12355
12356 /* Same as ada_exception_message_1, except that all exceptions are
12357 contained here (returning NULL instead). */
12358
12359 static gdb::unique_xmalloc_ptr<char>
12360 ada_exception_message (void)
12361 {
12362 gdb::unique_xmalloc_ptr<char> e_msg;
12363
12364 try
12365 {
12366 e_msg = ada_exception_message_1 ();
12367 }
12368 catch (const gdb_exception_error &e)
12369 {
12370 e_msg.reset (nullptr);
12371 }
12372
12373 return e_msg;
12374 }
12375
12376 /* Same as ada_exception_name_addr_1, except that it intercepts and contains
12377 any error that ada_exception_name_addr_1 might cause to be thrown.
12378 When an error is intercepted, a warning with the error message is printed,
12379 and zero is returned. */
12380
12381 static CORE_ADDR
12382 ada_exception_name_addr (enum ada_exception_catchpoint_kind ex,
12383 struct breakpoint *b)
12384 {
12385 CORE_ADDR result = 0;
12386
12387 try
12388 {
12389 result = ada_exception_name_addr_1 (ex, b);
12390 }
12391
12392 catch (const gdb_exception_error &e)
12393 {
12394 warning (_("failed to get exception name: %s"), e.what ());
12395 return 0;
12396 }
12397
12398 return result;
12399 }
12400
12401 static std::string ada_exception_catchpoint_cond_string
12402 (const char *excep_string,
12403 enum ada_exception_catchpoint_kind ex);
12404
12405 /* Ada catchpoints.
12406
12407 In the case of catchpoints on Ada exceptions, the catchpoint will
12408 stop the target on every exception the program throws. When a user
12409 specifies the name of a specific exception, we translate this
12410 request into a condition expression (in text form), and then parse
12411 it into an expression stored in each of the catchpoint's locations.
12412 We then use this condition to check whether the exception that was
12413 raised is the one the user is interested in. If not, then the
12414 target is resumed again. We store the name of the requested
12415 exception, in order to be able to re-set the condition expression
12416 when symbols change. */
12417
12418 /* An instance of this type is used to represent an Ada catchpoint
12419 breakpoint location. */
12420
12421 class ada_catchpoint_location : public bp_location
12422 {
12423 public:
12424 ada_catchpoint_location (breakpoint *owner)
12425 : bp_location (owner)
12426 {}
12427
12428 /* The condition that checks whether the exception that was raised
12429 is the specific exception the user specified on catchpoint
12430 creation. */
12431 expression_up excep_cond_expr;
12432 };
12433
12434 /* An instance of this type is used to represent an Ada catchpoint. */
12435
12436 struct ada_catchpoint : public breakpoint
12437 {
12438 /* The name of the specific exception the user specified. */
12439 std::string excep_string;
12440 };
12441
12442 /* Parse the exception condition string in the context of each of the
12443 catchpoint's locations, and store them for later evaluation. */
12444
12445 static void
12446 create_excep_cond_exprs (struct ada_catchpoint *c,
12447 enum ada_exception_catchpoint_kind ex)
12448 {
12449 /* Nothing to do if there's no specific exception to catch. */
12450 if (c->excep_string.empty ())
12451 return;
12452
12453 /* Same if there are no locations... */
12454 if (c->loc == NULL)
12455 return;
12456
12457 /* We have to compute the expression once for each program space,
12458 because the expression may hold the addresses of multiple symbols
12459 in some cases. */
12460 std::multimap<program_space *, struct bp_location *> loc_map;
12461 for (bp_location *bl = c->loc; bl != NULL; bl = bl->next)
12462 loc_map.emplace (bl->pspace, bl);
12463
12464 scoped_restore_current_program_space save_pspace;
12465
12466 std::string cond_string;
12467 program_space *last_ps = nullptr;
12468 for (auto iter : loc_map)
12469 {
12470 struct ada_catchpoint_location *ada_loc
12471 = (struct ada_catchpoint_location *) iter.second;
12472
12473 if (ada_loc->pspace != last_ps)
12474 {
12475 last_ps = ada_loc->pspace;
12476 set_current_program_space (last_ps);
12477
12478 /* Compute the condition expression in text form, from the
12479 specific expection we want to catch. */
12480 cond_string
12481 = ada_exception_catchpoint_cond_string (c->excep_string.c_str (),
12482 ex);
12483 }
12484
12485 expression_up exp;
12486
12487 if (!ada_loc->shlib_disabled)
12488 {
12489 const char *s;
12490
12491 s = cond_string.c_str ();
12492 try
12493 {
12494 exp = parse_exp_1 (&s, ada_loc->address,
12495 block_for_pc (ada_loc->address),
12496 0);
12497 }
12498 catch (const gdb_exception_error &e)
12499 {
12500 warning (_("failed to reevaluate internal exception condition "
12501 "for catchpoint %d: %s"),
12502 c->number, e.what ());
12503 }
12504 }
12505
12506 ada_loc->excep_cond_expr = std::move (exp);
12507 }
12508 }
12509
12510 /* Implement the ALLOCATE_LOCATION method in the breakpoint_ops
12511 structure for all exception catchpoint kinds. */
12512
12513 static struct bp_location *
12514 allocate_location_exception (enum ada_exception_catchpoint_kind ex,
12515 struct breakpoint *self)
12516 {
12517 return new ada_catchpoint_location (self);
12518 }
12519
12520 /* Implement the RE_SET method in the breakpoint_ops structure for all
12521 exception catchpoint kinds. */
12522
12523 static void
12524 re_set_exception (enum ada_exception_catchpoint_kind ex, struct breakpoint *b)
12525 {
12526 struct ada_catchpoint *c = (struct ada_catchpoint *) b;
12527
12528 /* Call the base class's method. This updates the catchpoint's
12529 locations. */
12530 bkpt_breakpoint_ops.re_set (b);
12531
12532 /* Reparse the exception conditional expressions. One for each
12533 location. */
12534 create_excep_cond_exprs (c, ex);
12535 }
12536
12537 /* Returns true if we should stop for this breakpoint hit. If the
12538 user specified a specific exception, we only want to cause a stop
12539 if the program thrown that exception. */
12540
12541 static int
12542 should_stop_exception (const struct bp_location *bl)
12543 {
12544 struct ada_catchpoint *c = (struct ada_catchpoint *) bl->owner;
12545 const struct ada_catchpoint_location *ada_loc
12546 = (const struct ada_catchpoint_location *) bl;
12547 int stop;
12548
12549 /* With no specific exception, should always stop. */
12550 if (c->excep_string.empty ())
12551 return 1;
12552
12553 if (ada_loc->excep_cond_expr == NULL)
12554 {
12555 /* We will have a NULL expression if back when we were creating
12556 the expressions, this location's had failed to parse. */
12557 return 1;
12558 }
12559
12560 stop = 1;
12561 try
12562 {
12563 struct value *mark;
12564
12565 mark = value_mark ();
12566 stop = value_true (evaluate_expression (ada_loc->excep_cond_expr.get ()));
12567 value_free_to_mark (mark);
12568 }
12569 catch (const gdb_exception &ex)
12570 {
12571 exception_fprintf (gdb_stderr, ex,
12572 _("Error in testing exception condition:\n"));
12573 }
12574
12575 return stop;
12576 }
12577
12578 /* Implement the CHECK_STATUS method in the breakpoint_ops structure
12579 for all exception catchpoint kinds. */
12580
12581 static void
12582 check_status_exception (enum ada_exception_catchpoint_kind ex, bpstat bs)
12583 {
12584 bs->stop = should_stop_exception (bs->bp_location_at);
12585 }
12586
12587 /* Implement the PRINT_IT method in the breakpoint_ops structure
12588 for all exception catchpoint kinds. */
12589
12590 static enum print_stop_action
12591 print_it_exception (enum ada_exception_catchpoint_kind ex, bpstat bs)
12592 {
12593 struct ui_out *uiout = current_uiout;
12594 struct breakpoint *b = bs->breakpoint_at;
12595
12596 annotate_catchpoint (b->number);
12597
12598 if (uiout->is_mi_like_p ())
12599 {
12600 uiout->field_string ("reason",
12601 async_reason_lookup (EXEC_ASYNC_BREAKPOINT_HIT));
12602 uiout->field_string ("disp", bpdisp_text (b->disposition));
12603 }
12604
12605 uiout->text (b->disposition == disp_del
12606 ? "\nTemporary catchpoint " : "\nCatchpoint ");
12607 uiout->field_int ("bkptno", b->number);
12608 uiout->text (", ");
12609
12610 /* ada_exception_name_addr relies on the selected frame being the
12611 current frame. Need to do this here because this function may be
12612 called more than once when printing a stop, and below, we'll
12613 select the first frame past the Ada run-time (see
12614 ada_find_printable_frame). */
12615 select_frame (get_current_frame ());
12616
12617 switch (ex)
12618 {
12619 case ada_catch_exception:
12620 case ada_catch_exception_unhandled:
12621 case ada_catch_handlers:
12622 {
12623 const CORE_ADDR addr = ada_exception_name_addr (ex, b);
12624 char exception_name[256];
12625
12626 if (addr != 0)
12627 {
12628 read_memory (addr, (gdb_byte *) exception_name,
12629 sizeof (exception_name) - 1);
12630 exception_name [sizeof (exception_name) - 1] = '\0';
12631 }
12632 else
12633 {
12634 /* For some reason, we were unable to read the exception
12635 name. This could happen if the Runtime was compiled
12636 without debugging info, for instance. In that case,
12637 just replace the exception name by the generic string
12638 "exception" - it will read as "an exception" in the
12639 notification we are about to print. */
12640 memcpy (exception_name, "exception", sizeof ("exception"));
12641 }
12642 /* In the case of unhandled exception breakpoints, we print
12643 the exception name as "unhandled EXCEPTION_NAME", to make
12644 it clearer to the user which kind of catchpoint just got
12645 hit. We used ui_out_text to make sure that this extra
12646 info does not pollute the exception name in the MI case. */
12647 if (ex == ada_catch_exception_unhandled)
12648 uiout->text ("unhandled ");
12649 uiout->field_string ("exception-name", exception_name);
12650 }
12651 break;
12652 case ada_catch_assert:
12653 /* In this case, the name of the exception is not really
12654 important. Just print "failed assertion" to make it clearer
12655 that his program just hit an assertion-failure catchpoint.
12656 We used ui_out_text because this info does not belong in
12657 the MI output. */
12658 uiout->text ("failed assertion");
12659 break;
12660 }
12661
12662 gdb::unique_xmalloc_ptr<char> exception_message = ada_exception_message ();
12663 if (exception_message != NULL)
12664 {
12665 uiout->text (" (");
12666 uiout->field_string ("exception-message", exception_message.get ());
12667 uiout->text (")");
12668 }
12669
12670 uiout->text (" at ");
12671 ada_find_printable_frame (get_current_frame ());
12672
12673 return PRINT_SRC_AND_LOC;
12674 }
12675
12676 /* Implement the PRINT_ONE method in the breakpoint_ops structure
12677 for all exception catchpoint kinds. */
12678
12679 static void
12680 print_one_exception (enum ada_exception_catchpoint_kind ex,
12681 struct breakpoint *b, struct bp_location **last_loc)
12682 {
12683 struct ui_out *uiout = current_uiout;
12684 struct ada_catchpoint *c = (struct ada_catchpoint *) b;
12685 struct value_print_options opts;
12686
12687 get_user_print_options (&opts);
12688 if (opts.addressprint)
12689 {
12690 annotate_field (4);
12691 uiout->field_core_addr ("addr", b->loc->gdbarch, b->loc->address);
12692 }
12693
12694 annotate_field (5);
12695 *last_loc = b->loc;
12696 switch (ex)
12697 {
12698 case ada_catch_exception:
12699 if (!c->excep_string.empty ())
12700 {
12701 std::string msg = string_printf (_("`%s' Ada exception"),
12702 c->excep_string.c_str ());
12703
12704 uiout->field_string ("what", msg);
12705 }
12706 else
12707 uiout->field_string ("what", "all Ada exceptions");
12708
12709 break;
12710
12711 case ada_catch_exception_unhandled:
12712 uiout->field_string ("what", "unhandled Ada exceptions");
12713 break;
12714
12715 case ada_catch_handlers:
12716 if (!c->excep_string.empty ())
12717 {
12718 uiout->field_fmt ("what",
12719 _("`%s' Ada exception handlers"),
12720 c->excep_string.c_str ());
12721 }
12722 else
12723 uiout->field_string ("what", "all Ada exceptions handlers");
12724 break;
12725
12726 case ada_catch_assert:
12727 uiout->field_string ("what", "failed Ada assertions");
12728 break;
12729
12730 default:
12731 internal_error (__FILE__, __LINE__, _("unexpected catchpoint type"));
12732 break;
12733 }
12734 }
12735
12736 /* Implement the PRINT_MENTION method in the breakpoint_ops structure
12737 for all exception catchpoint kinds. */
12738
12739 static void
12740 print_mention_exception (enum ada_exception_catchpoint_kind ex,
12741 struct breakpoint *b)
12742 {
12743 struct ada_catchpoint *c = (struct ada_catchpoint *) b;
12744 struct ui_out *uiout = current_uiout;
12745
12746 uiout->text (b->disposition == disp_del ? _("Temporary catchpoint ")
12747 : _("Catchpoint "));
12748 uiout->field_int ("bkptno", b->number);
12749 uiout->text (": ");
12750
12751 switch (ex)
12752 {
12753 case ada_catch_exception:
12754 if (!c->excep_string.empty ())
12755 {
12756 std::string info = string_printf (_("`%s' Ada exception"),
12757 c->excep_string.c_str ());
12758 uiout->text (info.c_str ());
12759 }
12760 else
12761 uiout->text (_("all Ada exceptions"));
12762 break;
12763
12764 case ada_catch_exception_unhandled:
12765 uiout->text (_("unhandled Ada exceptions"));
12766 break;
12767
12768 case ada_catch_handlers:
12769 if (!c->excep_string.empty ())
12770 {
12771 std::string info
12772 = string_printf (_("`%s' Ada exception handlers"),
12773 c->excep_string.c_str ());
12774 uiout->text (info.c_str ());
12775 }
12776 else
12777 uiout->text (_("all Ada exceptions handlers"));
12778 break;
12779
12780 case ada_catch_assert:
12781 uiout->text (_("failed Ada assertions"));
12782 break;
12783
12784 default:
12785 internal_error (__FILE__, __LINE__, _("unexpected catchpoint type"));
12786 break;
12787 }
12788 }
12789
12790 /* Implement the PRINT_RECREATE method in the breakpoint_ops structure
12791 for all exception catchpoint kinds. */
12792
12793 static void
12794 print_recreate_exception (enum ada_exception_catchpoint_kind ex,
12795 struct breakpoint *b, struct ui_file *fp)
12796 {
12797 struct ada_catchpoint *c = (struct ada_catchpoint *) b;
12798
12799 switch (ex)
12800 {
12801 case ada_catch_exception:
12802 fprintf_filtered (fp, "catch exception");
12803 if (!c->excep_string.empty ())
12804 fprintf_filtered (fp, " %s", c->excep_string.c_str ());
12805 break;
12806
12807 case ada_catch_exception_unhandled:
12808 fprintf_filtered (fp, "catch exception unhandled");
12809 break;
12810
12811 case ada_catch_handlers:
12812 fprintf_filtered (fp, "catch handlers");
12813 break;
12814
12815 case ada_catch_assert:
12816 fprintf_filtered (fp, "catch assert");
12817 break;
12818
12819 default:
12820 internal_error (__FILE__, __LINE__, _("unexpected catchpoint type"));
12821 }
12822 print_recreate_thread (b, fp);
12823 }
12824
12825 /* Virtual table for "catch exception" breakpoints. */
12826
12827 static struct bp_location *
12828 allocate_location_catch_exception (struct breakpoint *self)
12829 {
12830 return allocate_location_exception (ada_catch_exception, self);
12831 }
12832
12833 static void
12834 re_set_catch_exception (struct breakpoint *b)
12835 {
12836 re_set_exception (ada_catch_exception, b);
12837 }
12838
12839 static void
12840 check_status_catch_exception (bpstat bs)
12841 {
12842 check_status_exception (ada_catch_exception, bs);
12843 }
12844
12845 static enum print_stop_action
12846 print_it_catch_exception (bpstat bs)
12847 {
12848 return print_it_exception (ada_catch_exception, bs);
12849 }
12850
12851 static void
12852 print_one_catch_exception (struct breakpoint *b, struct bp_location **last_loc)
12853 {
12854 print_one_exception (ada_catch_exception, b, last_loc);
12855 }
12856
12857 static void
12858 print_mention_catch_exception (struct breakpoint *b)
12859 {
12860 print_mention_exception (ada_catch_exception, b);
12861 }
12862
12863 static void
12864 print_recreate_catch_exception (struct breakpoint *b, struct ui_file *fp)
12865 {
12866 print_recreate_exception (ada_catch_exception, b, fp);
12867 }
12868
12869 static struct breakpoint_ops catch_exception_breakpoint_ops;
12870
12871 /* Virtual table for "catch exception unhandled" breakpoints. */
12872
12873 static struct bp_location *
12874 allocate_location_catch_exception_unhandled (struct breakpoint *self)
12875 {
12876 return allocate_location_exception (ada_catch_exception_unhandled, self);
12877 }
12878
12879 static void
12880 re_set_catch_exception_unhandled (struct breakpoint *b)
12881 {
12882 re_set_exception (ada_catch_exception_unhandled, b);
12883 }
12884
12885 static void
12886 check_status_catch_exception_unhandled (bpstat bs)
12887 {
12888 check_status_exception (ada_catch_exception_unhandled, bs);
12889 }
12890
12891 static enum print_stop_action
12892 print_it_catch_exception_unhandled (bpstat bs)
12893 {
12894 return print_it_exception (ada_catch_exception_unhandled, bs);
12895 }
12896
12897 static void
12898 print_one_catch_exception_unhandled (struct breakpoint *b,
12899 struct bp_location **last_loc)
12900 {
12901 print_one_exception (ada_catch_exception_unhandled, b, last_loc);
12902 }
12903
12904 static void
12905 print_mention_catch_exception_unhandled (struct breakpoint *b)
12906 {
12907 print_mention_exception (ada_catch_exception_unhandled, b);
12908 }
12909
12910 static void
12911 print_recreate_catch_exception_unhandled (struct breakpoint *b,
12912 struct ui_file *fp)
12913 {
12914 print_recreate_exception (ada_catch_exception_unhandled, b, fp);
12915 }
12916
12917 static struct breakpoint_ops catch_exception_unhandled_breakpoint_ops;
12918
12919 /* Virtual table for "catch assert" breakpoints. */
12920
12921 static struct bp_location *
12922 allocate_location_catch_assert (struct breakpoint *self)
12923 {
12924 return allocate_location_exception (ada_catch_assert, self);
12925 }
12926
12927 static void
12928 re_set_catch_assert (struct breakpoint *b)
12929 {
12930 re_set_exception (ada_catch_assert, b);
12931 }
12932
12933 static void
12934 check_status_catch_assert (bpstat bs)
12935 {
12936 check_status_exception (ada_catch_assert, bs);
12937 }
12938
12939 static enum print_stop_action
12940 print_it_catch_assert (bpstat bs)
12941 {
12942 return print_it_exception (ada_catch_assert, bs);
12943 }
12944
12945 static void
12946 print_one_catch_assert (struct breakpoint *b, struct bp_location **last_loc)
12947 {
12948 print_one_exception (ada_catch_assert, b, last_loc);
12949 }
12950
12951 static void
12952 print_mention_catch_assert (struct breakpoint *b)
12953 {
12954 print_mention_exception (ada_catch_assert, b);
12955 }
12956
12957 static void
12958 print_recreate_catch_assert (struct breakpoint *b, struct ui_file *fp)
12959 {
12960 print_recreate_exception (ada_catch_assert, b, fp);
12961 }
12962
12963 static struct breakpoint_ops catch_assert_breakpoint_ops;
12964
12965 /* Virtual table for "catch handlers" breakpoints. */
12966
12967 static struct bp_location *
12968 allocate_location_catch_handlers (struct breakpoint *self)
12969 {
12970 return allocate_location_exception (ada_catch_handlers, self);
12971 }
12972
12973 static void
12974 re_set_catch_handlers (struct breakpoint *b)
12975 {
12976 re_set_exception (ada_catch_handlers, b);
12977 }
12978
12979 static void
12980 check_status_catch_handlers (bpstat bs)
12981 {
12982 check_status_exception (ada_catch_handlers, bs);
12983 }
12984
12985 static enum print_stop_action
12986 print_it_catch_handlers (bpstat bs)
12987 {
12988 return print_it_exception (ada_catch_handlers, bs);
12989 }
12990
12991 static void
12992 print_one_catch_handlers (struct breakpoint *b,
12993 struct bp_location **last_loc)
12994 {
12995 print_one_exception (ada_catch_handlers, b, last_loc);
12996 }
12997
12998 static void
12999 print_mention_catch_handlers (struct breakpoint *b)
13000 {
13001 print_mention_exception (ada_catch_handlers, b);
13002 }
13003
13004 static void
13005 print_recreate_catch_handlers (struct breakpoint *b,
13006 struct ui_file *fp)
13007 {
13008 print_recreate_exception (ada_catch_handlers, b, fp);
13009 }
13010
13011 static struct breakpoint_ops catch_handlers_breakpoint_ops;
13012
13013 /* Split the arguments specified in a "catch exception" command.
13014 Set EX to the appropriate catchpoint type.
13015 Set EXCEP_STRING to the name of the specific exception if
13016 specified by the user.
13017 IS_CATCH_HANDLERS_CMD: True if the arguments are for a
13018 "catch handlers" command. False otherwise.
13019 If a condition is found at the end of the arguments, the condition
13020 expression is stored in COND_STRING (memory must be deallocated
13021 after use). Otherwise COND_STRING is set to NULL. */
13022
13023 static void
13024 catch_ada_exception_command_split (const char *args,
13025 bool is_catch_handlers_cmd,
13026 enum ada_exception_catchpoint_kind *ex,
13027 std::string *excep_string,
13028 std::string *cond_string)
13029 {
13030 std::string exception_name;
13031
13032 exception_name = extract_arg (&args);
13033 if (exception_name == "if")
13034 {
13035 /* This is not an exception name; this is the start of a condition
13036 expression for a catchpoint on all exceptions. So, "un-get"
13037 this token, and set exception_name to NULL. */
13038 exception_name.clear ();
13039 args -= 2;
13040 }
13041
13042 /* Check to see if we have a condition. */
13043
13044 args = skip_spaces (args);
13045 if (startswith (args, "if")
13046 && (isspace (args[2]) || args[2] == '\0'))
13047 {
13048 args += 2;
13049 args = skip_spaces (args);
13050
13051 if (args[0] == '\0')
13052 error (_("Condition missing after `if' keyword"));
13053 *cond_string = args;
13054
13055 args += strlen (args);
13056 }
13057
13058 /* Check that we do not have any more arguments. Anything else
13059 is unexpected. */
13060
13061 if (args[0] != '\0')
13062 error (_("Junk at end of expression"));
13063
13064 if (is_catch_handlers_cmd)
13065 {
13066 /* Catch handling of exceptions. */
13067 *ex = ada_catch_handlers;
13068 *excep_string = exception_name;
13069 }
13070 else if (exception_name.empty ())
13071 {
13072 /* Catch all exceptions. */
13073 *ex = ada_catch_exception;
13074 excep_string->clear ();
13075 }
13076 else if (exception_name == "unhandled")
13077 {
13078 /* Catch unhandled exceptions. */
13079 *ex = ada_catch_exception_unhandled;
13080 excep_string->clear ();
13081 }
13082 else
13083 {
13084 /* Catch a specific exception. */
13085 *ex = ada_catch_exception;
13086 *excep_string = exception_name;
13087 }
13088 }
13089
13090 /* Return the name of the symbol on which we should break in order to
13091 implement a catchpoint of the EX kind. */
13092
13093 static const char *
13094 ada_exception_sym_name (enum ada_exception_catchpoint_kind ex)
13095 {
13096 struct ada_inferior_data *data = get_ada_inferior_data (current_inferior ());
13097
13098 gdb_assert (data->exception_info != NULL);
13099
13100 switch (ex)
13101 {
13102 case ada_catch_exception:
13103 return (data->exception_info->catch_exception_sym);
13104 break;
13105 case ada_catch_exception_unhandled:
13106 return (data->exception_info->catch_exception_unhandled_sym);
13107 break;
13108 case ada_catch_assert:
13109 return (data->exception_info->catch_assert_sym);
13110 break;
13111 case ada_catch_handlers:
13112 return (data->exception_info->catch_handlers_sym);
13113 break;
13114 default:
13115 internal_error (__FILE__, __LINE__,
13116 _("unexpected catchpoint kind (%d)"), ex);
13117 }
13118 }
13119
13120 /* Return the breakpoint ops "virtual table" used for catchpoints
13121 of the EX kind. */
13122
13123 static const struct breakpoint_ops *
13124 ada_exception_breakpoint_ops (enum ada_exception_catchpoint_kind ex)
13125 {
13126 switch (ex)
13127 {
13128 case ada_catch_exception:
13129 return (&catch_exception_breakpoint_ops);
13130 break;
13131 case ada_catch_exception_unhandled:
13132 return (&catch_exception_unhandled_breakpoint_ops);
13133 break;
13134 case ada_catch_assert:
13135 return (&catch_assert_breakpoint_ops);
13136 break;
13137 case ada_catch_handlers:
13138 return (&catch_handlers_breakpoint_ops);
13139 break;
13140 default:
13141 internal_error (__FILE__, __LINE__,
13142 _("unexpected catchpoint kind (%d)"), ex);
13143 }
13144 }
13145
13146 /* Return the condition that will be used to match the current exception
13147 being raised with the exception that the user wants to catch. This
13148 assumes that this condition is used when the inferior just triggered
13149 an exception catchpoint.
13150 EX: the type of catchpoints used for catching Ada exceptions. */
13151
13152 static std::string
13153 ada_exception_catchpoint_cond_string (const char *excep_string,
13154 enum ada_exception_catchpoint_kind ex)
13155 {
13156 int i;
13157 std::string result;
13158 const char *name;
13159
13160 if (ex == ada_catch_handlers)
13161 {
13162 /* For exception handlers catchpoints, the condition string does
13163 not use the same parameter as for the other exceptions. */
13164 name = ("long_integer (GNAT_GCC_exception_Access"
13165 "(gcc_exception).all.occurrence.id)");
13166 }
13167 else
13168 name = "long_integer (e)";
13169
13170 /* The standard exceptions are a special case. They are defined in
13171 runtime units that have been compiled without debugging info; if
13172 EXCEP_STRING is the not-fully-qualified name of a standard
13173 exception (e.g. "constraint_error") then, during the evaluation
13174 of the condition expression, the symbol lookup on this name would
13175 *not* return this standard exception. The catchpoint condition
13176 may then be set only on user-defined exceptions which have the
13177 same not-fully-qualified name (e.g. my_package.constraint_error).
13178
13179 To avoid this unexcepted behavior, these standard exceptions are
13180 systematically prefixed by "standard". This means that "catch
13181 exception constraint_error" is rewritten into "catch exception
13182 standard.constraint_error".
13183
13184 If an exception named contraint_error is defined in another package of
13185 the inferior program, then the only way to specify this exception as a
13186 breakpoint condition is to use its fully-qualified named:
13187 e.g. my_package.constraint_error.
13188
13189 Furthermore, in some situations a standard exception's symbol may
13190 be present in more than one objfile, because the compiler may
13191 choose to emit copy relocations for them. So, we have to compare
13192 against all the possible addresses. */
13193
13194 /* Storage for a rewritten symbol name. */
13195 std::string std_name;
13196 for (i = 0; i < sizeof (standard_exc) / sizeof (char *); i++)
13197 {
13198 if (strcmp (standard_exc [i], excep_string) == 0)
13199 {
13200 std_name = std::string ("standard.") + excep_string;
13201 excep_string = std_name.c_str ();
13202 break;
13203 }
13204 }
13205
13206 excep_string = ada_encode (excep_string);
13207 std::vector<struct bound_minimal_symbol> symbols
13208 = ada_lookup_simple_minsyms (excep_string);
13209 for (const bound_minimal_symbol &msym : symbols)
13210 {
13211 if (!result.empty ())
13212 result += " or ";
13213 string_appendf (result, "%s = %s", name,
13214 pulongest (BMSYMBOL_VALUE_ADDRESS (msym)));
13215 }
13216
13217 return result;
13218 }
13219
13220 /* Return the symtab_and_line that should be used to insert an exception
13221 catchpoint of the TYPE kind.
13222
13223 ADDR_STRING returns the name of the function where the real
13224 breakpoint that implements the catchpoints is set, depending on the
13225 type of catchpoint we need to create. */
13226
13227 static struct symtab_and_line
13228 ada_exception_sal (enum ada_exception_catchpoint_kind ex,
13229 std::string *addr_string, const struct breakpoint_ops **ops)
13230 {
13231 const char *sym_name;
13232 struct symbol *sym;
13233
13234 /* First, find out which exception support info to use. */
13235 ada_exception_support_info_sniffer ();
13236
13237 /* Then lookup the function on which we will break in order to catch
13238 the Ada exceptions requested by the user. */
13239 sym_name = ada_exception_sym_name (ex);
13240 sym = standard_lookup (sym_name, NULL, VAR_DOMAIN);
13241
13242 if (sym == NULL)
13243 error (_("Catchpoint symbol not found: %s"), sym_name);
13244
13245 if (SYMBOL_CLASS (sym) != LOC_BLOCK)
13246 error (_("Unable to insert catchpoint. %s is not a function."), sym_name);
13247
13248 /* Set ADDR_STRING. */
13249 *addr_string = sym_name;
13250
13251 /* Set OPS. */
13252 *ops = ada_exception_breakpoint_ops (ex);
13253
13254 return find_function_start_sal (sym, 1);
13255 }
13256
13257 /* Create an Ada exception catchpoint.
13258
13259 EX_KIND is the kind of exception catchpoint to be created.
13260
13261 If EXCEPT_STRING is empty, this catchpoint is expected to trigger
13262 for all exceptions. Otherwise, EXCEPT_STRING indicates the name
13263 of the exception to which this catchpoint applies.
13264
13265 COND_STRING, if not empty, is the catchpoint condition.
13266
13267 TEMPFLAG, if nonzero, means that the underlying breakpoint
13268 should be temporary.
13269
13270 FROM_TTY is the usual argument passed to all commands implementations. */
13271
13272 void
13273 create_ada_exception_catchpoint (struct gdbarch *gdbarch,
13274 enum ada_exception_catchpoint_kind ex_kind,
13275 const std::string &excep_string,
13276 const std::string &cond_string,
13277 int tempflag,
13278 int disabled,
13279 int from_tty)
13280 {
13281 std::string addr_string;
13282 const struct breakpoint_ops *ops = NULL;
13283 struct symtab_and_line sal = ada_exception_sal (ex_kind, &addr_string, &ops);
13284
13285 std::unique_ptr<ada_catchpoint> c (new ada_catchpoint ());
13286 init_ada_exception_breakpoint (c.get (), gdbarch, sal, addr_string.c_str (),
13287 ops, tempflag, disabled, from_tty);
13288 c->excep_string = excep_string;
13289 create_excep_cond_exprs (c.get (), ex_kind);
13290 if (!cond_string.empty ())
13291 set_breakpoint_condition (c.get (), cond_string.c_str (), from_tty);
13292 install_breakpoint (0, std::move (c), 1);
13293 }
13294
13295 /* Implement the "catch exception" command. */
13296
13297 static void
13298 catch_ada_exception_command (const char *arg_entry, int from_tty,
13299 struct cmd_list_element *command)
13300 {
13301 const char *arg = arg_entry;
13302 struct gdbarch *gdbarch = get_current_arch ();
13303 int tempflag;
13304 enum ada_exception_catchpoint_kind ex_kind;
13305 std::string excep_string;
13306 std::string cond_string;
13307
13308 tempflag = get_cmd_context (command) == CATCH_TEMPORARY;
13309
13310 if (!arg)
13311 arg = "";
13312 catch_ada_exception_command_split (arg, false, &ex_kind, &excep_string,
13313 &cond_string);
13314 create_ada_exception_catchpoint (gdbarch, ex_kind,
13315 excep_string, cond_string,
13316 tempflag, 1 /* enabled */,
13317 from_tty);
13318 }
13319
13320 /* Implement the "catch handlers" command. */
13321
13322 static void
13323 catch_ada_handlers_command (const char *arg_entry, int from_tty,
13324 struct cmd_list_element *command)
13325 {
13326 const char *arg = arg_entry;
13327 struct gdbarch *gdbarch = get_current_arch ();
13328 int tempflag;
13329 enum ada_exception_catchpoint_kind ex_kind;
13330 std::string excep_string;
13331 std::string cond_string;
13332
13333 tempflag = get_cmd_context (command) == CATCH_TEMPORARY;
13334
13335 if (!arg)
13336 arg = "";
13337 catch_ada_exception_command_split (arg, true, &ex_kind, &excep_string,
13338 &cond_string);
13339 create_ada_exception_catchpoint (gdbarch, ex_kind,
13340 excep_string, cond_string,
13341 tempflag, 1 /* enabled */,
13342 from_tty);
13343 }
13344
13345 /* Completion function for the Ada "catch" commands. */
13346
13347 static void
13348 catch_ada_completer (struct cmd_list_element *cmd, completion_tracker &tracker,
13349 const char *text, const char *word)
13350 {
13351 std::vector<ada_exc_info> exceptions = ada_exceptions_list (NULL);
13352
13353 for (const ada_exc_info &info : exceptions)
13354 {
13355 if (startswith (info.name, word))
13356 tracker.add_completion
13357 (gdb::unique_xmalloc_ptr<char> (xstrdup (info.name)));
13358 }
13359 }
13360
13361 /* Split the arguments specified in a "catch assert" command.
13362
13363 ARGS contains the command's arguments (or the empty string if
13364 no arguments were passed).
13365
13366 If ARGS contains a condition, set COND_STRING to that condition
13367 (the memory needs to be deallocated after use). */
13368
13369 static void
13370 catch_ada_assert_command_split (const char *args, std::string &cond_string)
13371 {
13372 args = skip_spaces (args);
13373
13374 /* Check whether a condition was provided. */
13375 if (startswith (args, "if")
13376 && (isspace (args[2]) || args[2] == '\0'))
13377 {
13378 args += 2;
13379 args = skip_spaces (args);
13380 if (args[0] == '\0')
13381 error (_("condition missing after `if' keyword"));
13382 cond_string.assign (args);
13383 }
13384
13385 /* Otherwise, there should be no other argument at the end of
13386 the command. */
13387 else if (args[0] != '\0')
13388 error (_("Junk at end of arguments."));
13389 }
13390
13391 /* Implement the "catch assert" command. */
13392
13393 static void
13394 catch_assert_command (const char *arg_entry, int from_tty,
13395 struct cmd_list_element *command)
13396 {
13397 const char *arg = arg_entry;
13398 struct gdbarch *gdbarch = get_current_arch ();
13399 int tempflag;
13400 std::string cond_string;
13401
13402 tempflag = get_cmd_context (command) == CATCH_TEMPORARY;
13403
13404 if (!arg)
13405 arg = "";
13406 catch_ada_assert_command_split (arg, cond_string);
13407 create_ada_exception_catchpoint (gdbarch, ada_catch_assert,
13408 "", cond_string,
13409 tempflag, 1 /* enabled */,
13410 from_tty);
13411 }
13412
13413 /* Return non-zero if the symbol SYM is an Ada exception object. */
13414
13415 static int
13416 ada_is_exception_sym (struct symbol *sym)
13417 {
13418 const char *type_name = TYPE_NAME (SYMBOL_TYPE (sym));
13419
13420 return (SYMBOL_CLASS (sym) != LOC_TYPEDEF
13421 && SYMBOL_CLASS (sym) != LOC_BLOCK
13422 && SYMBOL_CLASS (sym) != LOC_CONST
13423 && SYMBOL_CLASS (sym) != LOC_UNRESOLVED
13424 && type_name != NULL && strcmp (type_name, "exception") == 0);
13425 }
13426
13427 /* Given a global symbol SYM, return non-zero iff SYM is a non-standard
13428 Ada exception object. This matches all exceptions except the ones
13429 defined by the Ada language. */
13430
13431 static int
13432 ada_is_non_standard_exception_sym (struct symbol *sym)
13433 {
13434 int i;
13435
13436 if (!ada_is_exception_sym (sym))
13437 return 0;
13438
13439 for (i = 0; i < ARRAY_SIZE (standard_exc); i++)
13440 if (strcmp (SYMBOL_LINKAGE_NAME (sym), standard_exc[i]) == 0)
13441 return 0; /* A standard exception. */
13442
13443 /* Numeric_Error is also a standard exception, so exclude it.
13444 See the STANDARD_EXC description for more details as to why
13445 this exception is not listed in that array. */
13446 if (strcmp (SYMBOL_LINKAGE_NAME (sym), "numeric_error") == 0)
13447 return 0;
13448
13449 return 1;
13450 }
13451
13452 /* A helper function for std::sort, comparing two struct ada_exc_info
13453 objects.
13454
13455 The comparison is determined first by exception name, and then
13456 by exception address. */
13457
13458 bool
13459 ada_exc_info::operator< (const ada_exc_info &other) const
13460 {
13461 int result;
13462
13463 result = strcmp (name, other.name);
13464 if (result < 0)
13465 return true;
13466 if (result == 0 && addr < other.addr)
13467 return true;
13468 return false;
13469 }
13470
13471 bool
13472 ada_exc_info::operator== (const ada_exc_info &other) const
13473 {
13474 return addr == other.addr && strcmp (name, other.name) == 0;
13475 }
13476
13477 /* Sort EXCEPTIONS using compare_ada_exception_info as the comparison
13478 routine, but keeping the first SKIP elements untouched.
13479
13480 All duplicates are also removed. */
13481
13482 static void
13483 sort_remove_dups_ada_exceptions_list (std::vector<ada_exc_info> *exceptions,
13484 int skip)
13485 {
13486 std::sort (exceptions->begin () + skip, exceptions->end ());
13487 exceptions->erase (std::unique (exceptions->begin () + skip, exceptions->end ()),
13488 exceptions->end ());
13489 }
13490
13491 /* Add all exceptions defined by the Ada standard whose name match
13492 a regular expression.
13493
13494 If PREG is not NULL, then this regexp_t object is used to
13495 perform the symbol name matching. Otherwise, no name-based
13496 filtering is performed.
13497
13498 EXCEPTIONS is a vector of exceptions to which matching exceptions
13499 gets pushed. */
13500
13501 static void
13502 ada_add_standard_exceptions (compiled_regex *preg,
13503 std::vector<ada_exc_info> *exceptions)
13504 {
13505 int i;
13506
13507 for (i = 0; i < ARRAY_SIZE (standard_exc); i++)
13508 {
13509 if (preg == NULL
13510 || preg->exec (standard_exc[i], 0, NULL, 0) == 0)
13511 {
13512 struct bound_minimal_symbol msymbol
13513 = ada_lookup_simple_minsym (standard_exc[i]);
13514
13515 if (msymbol.minsym != NULL)
13516 {
13517 struct ada_exc_info info
13518 = {standard_exc[i], BMSYMBOL_VALUE_ADDRESS (msymbol)};
13519
13520 exceptions->push_back (info);
13521 }
13522 }
13523 }
13524 }
13525
13526 /* Add all Ada exceptions defined locally and accessible from the given
13527 FRAME.
13528
13529 If PREG is not NULL, then this regexp_t object is used to
13530 perform the symbol name matching. Otherwise, no name-based
13531 filtering is performed.
13532
13533 EXCEPTIONS is a vector of exceptions to which matching exceptions
13534 gets pushed. */
13535
13536 static void
13537 ada_add_exceptions_from_frame (compiled_regex *preg,
13538 struct frame_info *frame,
13539 std::vector<ada_exc_info> *exceptions)
13540 {
13541 const struct block *block = get_frame_block (frame, 0);
13542
13543 while (block != 0)
13544 {
13545 struct block_iterator iter;
13546 struct symbol *sym;
13547
13548 ALL_BLOCK_SYMBOLS (block, iter, sym)
13549 {
13550 switch (SYMBOL_CLASS (sym))
13551 {
13552 case LOC_TYPEDEF:
13553 case LOC_BLOCK:
13554 case LOC_CONST:
13555 break;
13556 default:
13557 if (ada_is_exception_sym (sym))
13558 {
13559 struct ada_exc_info info = {SYMBOL_PRINT_NAME (sym),
13560 SYMBOL_VALUE_ADDRESS (sym)};
13561
13562 exceptions->push_back (info);
13563 }
13564 }
13565 }
13566 if (BLOCK_FUNCTION (block) != NULL)
13567 break;
13568 block = BLOCK_SUPERBLOCK (block);
13569 }
13570 }
13571
13572 /* Return true if NAME matches PREG or if PREG is NULL. */
13573
13574 static bool
13575 name_matches_regex (const char *name, compiled_regex *preg)
13576 {
13577 return (preg == NULL
13578 || preg->exec (ada_decode (name), 0, NULL, 0) == 0);
13579 }
13580
13581 /* Add all exceptions defined globally whose name name match
13582 a regular expression, excluding standard exceptions.
13583
13584 The reason we exclude standard exceptions is that they need
13585 to be handled separately: Standard exceptions are defined inside
13586 a runtime unit which is normally not compiled with debugging info,
13587 and thus usually do not show up in our symbol search. However,
13588 if the unit was in fact built with debugging info, we need to
13589 exclude them because they would duplicate the entry we found
13590 during the special loop that specifically searches for those
13591 standard exceptions.
13592
13593 If PREG is not NULL, then this regexp_t object is used to
13594 perform the symbol name matching. Otherwise, no name-based
13595 filtering is performed.
13596
13597 EXCEPTIONS is a vector of exceptions to which matching exceptions
13598 gets pushed. */
13599
13600 static void
13601 ada_add_global_exceptions (compiled_regex *preg,
13602 std::vector<ada_exc_info> *exceptions)
13603 {
13604 /* In Ada, the symbol "search name" is a linkage name, whereas the
13605 regular expression used to do the matching refers to the natural
13606 name. So match against the decoded name. */
13607 expand_symtabs_matching (NULL,
13608 lookup_name_info::match_any (),
13609 [&] (const char *search_name)
13610 {
13611 const char *decoded = ada_decode (search_name);
13612 return name_matches_regex (decoded, preg);
13613 },
13614 NULL,
13615 VARIABLES_DOMAIN);
13616
13617 for (objfile *objfile : current_program_space->objfiles ())
13618 {
13619 for (compunit_symtab *s : objfile->compunits ())
13620 {
13621 const struct blockvector *bv = COMPUNIT_BLOCKVECTOR (s);
13622 int i;
13623
13624 for (i = GLOBAL_BLOCK; i <= STATIC_BLOCK; i++)
13625 {
13626 const struct block *b = BLOCKVECTOR_BLOCK (bv, i);
13627 struct block_iterator iter;
13628 struct symbol *sym;
13629
13630 ALL_BLOCK_SYMBOLS (b, iter, sym)
13631 if (ada_is_non_standard_exception_sym (sym)
13632 && name_matches_regex (SYMBOL_NATURAL_NAME (sym), preg))
13633 {
13634 struct ada_exc_info info
13635 = {SYMBOL_PRINT_NAME (sym), SYMBOL_VALUE_ADDRESS (sym)};
13636
13637 exceptions->push_back (info);
13638 }
13639 }
13640 }
13641 }
13642 }
13643
13644 /* Implements ada_exceptions_list with the regular expression passed
13645 as a regex_t, rather than a string.
13646
13647 If not NULL, PREG is used to filter out exceptions whose names
13648 do not match. Otherwise, all exceptions are listed. */
13649
13650 static std::vector<ada_exc_info>
13651 ada_exceptions_list_1 (compiled_regex *preg)
13652 {
13653 std::vector<ada_exc_info> result;
13654 int prev_len;
13655
13656 /* First, list the known standard exceptions. These exceptions
13657 need to be handled separately, as they are usually defined in
13658 runtime units that have been compiled without debugging info. */
13659
13660 ada_add_standard_exceptions (preg, &result);
13661
13662 /* Next, find all exceptions whose scope is local and accessible
13663 from the currently selected frame. */
13664
13665 if (has_stack_frames ())
13666 {
13667 prev_len = result.size ();
13668 ada_add_exceptions_from_frame (preg, get_selected_frame (NULL),
13669 &result);
13670 if (result.size () > prev_len)
13671 sort_remove_dups_ada_exceptions_list (&result, prev_len);
13672 }
13673
13674 /* Add all exceptions whose scope is global. */
13675
13676 prev_len = result.size ();
13677 ada_add_global_exceptions (preg, &result);
13678 if (result.size () > prev_len)
13679 sort_remove_dups_ada_exceptions_list (&result, prev_len);
13680
13681 return result;
13682 }
13683
13684 /* Return a vector of ada_exc_info.
13685
13686 If REGEXP is NULL, all exceptions are included in the result.
13687 Otherwise, it should contain a valid regular expression,
13688 and only the exceptions whose names match that regular expression
13689 are included in the result.
13690
13691 The exceptions are sorted in the following order:
13692 - Standard exceptions (defined by the Ada language), in
13693 alphabetical order;
13694 - Exceptions only visible from the current frame, in
13695 alphabetical order;
13696 - Exceptions whose scope is global, in alphabetical order. */
13697
13698 std::vector<ada_exc_info>
13699 ada_exceptions_list (const char *regexp)
13700 {
13701 if (regexp == NULL)
13702 return ada_exceptions_list_1 (NULL);
13703
13704 compiled_regex reg (regexp, REG_NOSUB, _("invalid regular expression"));
13705 return ada_exceptions_list_1 (&reg);
13706 }
13707
13708 /* Implement the "info exceptions" command. */
13709
13710 static void
13711 info_exceptions_command (const char *regexp, int from_tty)
13712 {
13713 struct gdbarch *gdbarch = get_current_arch ();
13714
13715 std::vector<ada_exc_info> exceptions = ada_exceptions_list (regexp);
13716
13717 if (regexp != NULL)
13718 printf_filtered
13719 (_("All Ada exceptions matching regular expression \"%s\":\n"), regexp);
13720 else
13721 printf_filtered (_("All defined Ada exceptions:\n"));
13722
13723 for (const ada_exc_info &info : exceptions)
13724 printf_filtered ("%s: %s\n", info.name, paddress (gdbarch, info.addr));
13725 }
13726
13727 /* Operators */
13728 /* Information about operators given special treatment in functions
13729 below. */
13730 /* Format: OP_DEFN (<operator>, <operator length>, <# args>, <binop>). */
13731
13732 #define ADA_OPERATORS \
13733 OP_DEFN (OP_VAR_VALUE, 4, 0, 0) \
13734 OP_DEFN (BINOP_IN_BOUNDS, 3, 2, 0) \
13735 OP_DEFN (TERNOP_IN_RANGE, 1, 3, 0) \
13736 OP_DEFN (OP_ATR_FIRST, 1, 2, 0) \
13737 OP_DEFN (OP_ATR_LAST, 1, 2, 0) \
13738 OP_DEFN (OP_ATR_LENGTH, 1, 2, 0) \
13739 OP_DEFN (OP_ATR_IMAGE, 1, 2, 0) \
13740 OP_DEFN (OP_ATR_MAX, 1, 3, 0) \
13741 OP_DEFN (OP_ATR_MIN, 1, 3, 0) \
13742 OP_DEFN (OP_ATR_MODULUS, 1, 1, 0) \
13743 OP_DEFN (OP_ATR_POS, 1, 2, 0) \
13744 OP_DEFN (OP_ATR_SIZE, 1, 1, 0) \
13745 OP_DEFN (OP_ATR_TAG, 1, 1, 0) \
13746 OP_DEFN (OP_ATR_VAL, 1, 2, 0) \
13747 OP_DEFN (UNOP_QUAL, 3, 1, 0) \
13748 OP_DEFN (UNOP_IN_RANGE, 3, 1, 0) \
13749 OP_DEFN (OP_OTHERS, 1, 1, 0) \
13750 OP_DEFN (OP_POSITIONAL, 3, 1, 0) \
13751 OP_DEFN (OP_DISCRETE_RANGE, 1, 2, 0)
13752
13753 static void
13754 ada_operator_length (const struct expression *exp, int pc, int *oplenp,
13755 int *argsp)
13756 {
13757 switch (exp->elts[pc - 1].opcode)
13758 {
13759 default:
13760 operator_length_standard (exp, pc, oplenp, argsp);
13761 break;
13762
13763 #define OP_DEFN(op, len, args, binop) \
13764 case op: *oplenp = len; *argsp = args; break;
13765 ADA_OPERATORS;
13766 #undef OP_DEFN
13767
13768 case OP_AGGREGATE:
13769 *oplenp = 3;
13770 *argsp = longest_to_int (exp->elts[pc - 2].longconst);
13771 break;
13772
13773 case OP_CHOICES:
13774 *oplenp = 3;
13775 *argsp = longest_to_int (exp->elts[pc - 2].longconst) + 1;
13776 break;
13777 }
13778 }
13779
13780 /* Implementation of the exp_descriptor method operator_check. */
13781
13782 static int
13783 ada_operator_check (struct expression *exp, int pos,
13784 int (*objfile_func) (struct objfile *objfile, void *data),
13785 void *data)
13786 {
13787 const union exp_element *const elts = exp->elts;
13788 struct type *type = NULL;
13789
13790 switch (elts[pos].opcode)
13791 {
13792 case UNOP_IN_RANGE:
13793 case UNOP_QUAL:
13794 type = elts[pos + 1].type;
13795 break;
13796
13797 default:
13798 return operator_check_standard (exp, pos, objfile_func, data);
13799 }
13800
13801 /* Invoke callbacks for TYPE and OBJFILE if they were set as non-NULL. */
13802
13803 if (type && TYPE_OBJFILE (type)
13804 && (*objfile_func) (TYPE_OBJFILE (type), data))
13805 return 1;
13806
13807 return 0;
13808 }
13809
13810 static const char *
13811 ada_op_name (enum exp_opcode opcode)
13812 {
13813 switch (opcode)
13814 {
13815 default:
13816 return op_name_standard (opcode);
13817
13818 #define OP_DEFN(op, len, args, binop) case op: return #op;
13819 ADA_OPERATORS;
13820 #undef OP_DEFN
13821
13822 case OP_AGGREGATE:
13823 return "OP_AGGREGATE";
13824 case OP_CHOICES:
13825 return "OP_CHOICES";
13826 case OP_NAME:
13827 return "OP_NAME";
13828 }
13829 }
13830
13831 /* As for operator_length, but assumes PC is pointing at the first
13832 element of the operator, and gives meaningful results only for the
13833 Ada-specific operators, returning 0 for *OPLENP and *ARGSP otherwise. */
13834
13835 static void
13836 ada_forward_operator_length (struct expression *exp, int pc,
13837 int *oplenp, int *argsp)
13838 {
13839 switch (exp->elts[pc].opcode)
13840 {
13841 default:
13842 *oplenp = *argsp = 0;
13843 break;
13844
13845 #define OP_DEFN(op, len, args, binop) \
13846 case op: *oplenp = len; *argsp = args; break;
13847 ADA_OPERATORS;
13848 #undef OP_DEFN
13849
13850 case OP_AGGREGATE:
13851 *oplenp = 3;
13852 *argsp = longest_to_int (exp->elts[pc + 1].longconst);
13853 break;
13854
13855 case OP_CHOICES:
13856 *oplenp = 3;
13857 *argsp = longest_to_int (exp->elts[pc + 1].longconst) + 1;
13858 break;
13859
13860 case OP_STRING:
13861 case OP_NAME:
13862 {
13863 int len = longest_to_int (exp->elts[pc + 1].longconst);
13864
13865 *oplenp = 4 + BYTES_TO_EXP_ELEM (len + 1);
13866 *argsp = 0;
13867 break;
13868 }
13869 }
13870 }
13871
13872 static int
13873 ada_dump_subexp_body (struct expression *exp, struct ui_file *stream, int elt)
13874 {
13875 enum exp_opcode op = exp->elts[elt].opcode;
13876 int oplen, nargs;
13877 int pc = elt;
13878 int i;
13879
13880 ada_forward_operator_length (exp, elt, &oplen, &nargs);
13881
13882 switch (op)
13883 {
13884 /* Ada attributes ('Foo). */
13885 case OP_ATR_FIRST:
13886 case OP_ATR_LAST:
13887 case OP_ATR_LENGTH:
13888 case OP_ATR_IMAGE:
13889 case OP_ATR_MAX:
13890 case OP_ATR_MIN:
13891 case OP_ATR_MODULUS:
13892 case OP_ATR_POS:
13893 case OP_ATR_SIZE:
13894 case OP_ATR_TAG:
13895 case OP_ATR_VAL:
13896 break;
13897
13898 case UNOP_IN_RANGE:
13899 case UNOP_QUAL:
13900 /* XXX: gdb_sprint_host_address, type_sprint */
13901 fprintf_filtered (stream, _("Type @"));
13902 gdb_print_host_address (exp->elts[pc + 1].type, stream);
13903 fprintf_filtered (stream, " (");
13904 type_print (exp->elts[pc + 1].type, NULL, stream, 0);
13905 fprintf_filtered (stream, ")");
13906 break;
13907 case BINOP_IN_BOUNDS:
13908 fprintf_filtered (stream, " (%d)",
13909 longest_to_int (exp->elts[pc + 2].longconst));
13910 break;
13911 case TERNOP_IN_RANGE:
13912 break;
13913
13914 case OP_AGGREGATE:
13915 case OP_OTHERS:
13916 case OP_DISCRETE_RANGE:
13917 case OP_POSITIONAL:
13918 case OP_CHOICES:
13919 break;
13920
13921 case OP_NAME:
13922 case OP_STRING:
13923 {
13924 char *name = &exp->elts[elt + 2].string;
13925 int len = longest_to_int (exp->elts[elt + 1].longconst);
13926
13927 fprintf_filtered (stream, "Text: `%.*s'", len, name);
13928 break;
13929 }
13930
13931 default:
13932 return dump_subexp_body_standard (exp, stream, elt);
13933 }
13934
13935 elt += oplen;
13936 for (i = 0; i < nargs; i += 1)
13937 elt = dump_subexp (exp, stream, elt);
13938
13939 return elt;
13940 }
13941
13942 /* The Ada extension of print_subexp (q.v.). */
13943
13944 static void
13945 ada_print_subexp (struct expression *exp, int *pos,
13946 struct ui_file *stream, enum precedence prec)
13947 {
13948 int oplen, nargs, i;
13949 int pc = *pos;
13950 enum exp_opcode op = exp->elts[pc].opcode;
13951
13952 ada_forward_operator_length (exp, pc, &oplen, &nargs);
13953
13954 *pos += oplen;
13955 switch (op)
13956 {
13957 default:
13958 *pos -= oplen;
13959 print_subexp_standard (exp, pos, stream, prec);
13960 return;
13961
13962 case OP_VAR_VALUE:
13963 fputs_filtered (SYMBOL_NATURAL_NAME (exp->elts[pc + 2].symbol), stream);
13964 return;
13965
13966 case BINOP_IN_BOUNDS:
13967 /* XXX: sprint_subexp */
13968 print_subexp (exp, pos, stream, PREC_SUFFIX);
13969 fputs_filtered (" in ", stream);
13970 print_subexp (exp, pos, stream, PREC_SUFFIX);
13971 fputs_filtered ("'range", stream);
13972 if (exp->elts[pc + 1].longconst > 1)
13973 fprintf_filtered (stream, "(%ld)",
13974 (long) exp->elts[pc + 1].longconst);
13975 return;
13976
13977 case TERNOP_IN_RANGE:
13978 if (prec >= PREC_EQUAL)
13979 fputs_filtered ("(", stream);
13980 /* XXX: sprint_subexp */
13981 print_subexp (exp, pos, stream, PREC_SUFFIX);
13982 fputs_filtered (" in ", stream);
13983 print_subexp (exp, pos, stream, PREC_EQUAL);
13984 fputs_filtered (" .. ", stream);
13985 print_subexp (exp, pos, stream, PREC_EQUAL);
13986 if (prec >= PREC_EQUAL)
13987 fputs_filtered (")", stream);
13988 return;
13989
13990 case OP_ATR_FIRST:
13991 case OP_ATR_LAST:
13992 case OP_ATR_LENGTH:
13993 case OP_ATR_IMAGE:
13994 case OP_ATR_MAX:
13995 case OP_ATR_MIN:
13996 case OP_ATR_MODULUS:
13997 case OP_ATR_POS:
13998 case OP_ATR_SIZE:
13999 case OP_ATR_TAG:
14000 case OP_ATR_VAL:
14001 if (exp->elts[*pos].opcode == OP_TYPE)
14002 {
14003 if (TYPE_CODE (exp->elts[*pos + 1].type) != TYPE_CODE_VOID)
14004 LA_PRINT_TYPE (exp->elts[*pos + 1].type, "", stream, 0, 0,
14005 &type_print_raw_options);
14006 *pos += 3;
14007 }
14008 else
14009 print_subexp (exp, pos, stream, PREC_SUFFIX);
14010 fprintf_filtered (stream, "'%s", ada_attribute_name (op));
14011 if (nargs > 1)
14012 {
14013 int tem;
14014
14015 for (tem = 1; tem < nargs; tem += 1)
14016 {
14017 fputs_filtered ((tem == 1) ? " (" : ", ", stream);
14018 print_subexp (exp, pos, stream, PREC_ABOVE_COMMA);
14019 }
14020 fputs_filtered (")", stream);
14021 }
14022 return;
14023
14024 case UNOP_QUAL:
14025 type_print (exp->elts[pc + 1].type, "", stream, 0);
14026 fputs_filtered ("'(", stream);
14027 print_subexp (exp, pos, stream, PREC_PREFIX);
14028 fputs_filtered (")", stream);
14029 return;
14030
14031 case UNOP_IN_RANGE:
14032 /* XXX: sprint_subexp */
14033 print_subexp (exp, pos, stream, PREC_SUFFIX);
14034 fputs_filtered (" in ", stream);
14035 LA_PRINT_TYPE (exp->elts[pc + 1].type, "", stream, 1, 0,
14036 &type_print_raw_options);
14037 return;
14038
14039 case OP_DISCRETE_RANGE:
14040 print_subexp (exp, pos, stream, PREC_SUFFIX);
14041 fputs_filtered ("..", stream);
14042 print_subexp (exp, pos, stream, PREC_SUFFIX);
14043 return;
14044
14045 case OP_OTHERS:
14046 fputs_filtered ("others => ", stream);
14047 print_subexp (exp, pos, stream, PREC_SUFFIX);
14048 return;
14049
14050 case OP_CHOICES:
14051 for (i = 0; i < nargs-1; i += 1)
14052 {
14053 if (i > 0)
14054 fputs_filtered ("|", stream);
14055 print_subexp (exp, pos, stream, PREC_SUFFIX);
14056 }
14057 fputs_filtered (" => ", stream);
14058 print_subexp (exp, pos, stream, PREC_SUFFIX);
14059 return;
14060
14061 case OP_POSITIONAL:
14062 print_subexp (exp, pos, stream, PREC_SUFFIX);
14063 return;
14064
14065 case OP_AGGREGATE:
14066 fputs_filtered ("(", stream);
14067 for (i = 0; i < nargs; i += 1)
14068 {
14069 if (i > 0)
14070 fputs_filtered (", ", stream);
14071 print_subexp (exp, pos, stream, PREC_SUFFIX);
14072 }
14073 fputs_filtered (")", stream);
14074 return;
14075 }
14076 }
14077
14078 /* Table mapping opcodes into strings for printing operators
14079 and precedences of the operators. */
14080
14081 static const struct op_print ada_op_print_tab[] = {
14082 {":=", BINOP_ASSIGN, PREC_ASSIGN, 1},
14083 {"or else", BINOP_LOGICAL_OR, PREC_LOGICAL_OR, 0},
14084 {"and then", BINOP_LOGICAL_AND, PREC_LOGICAL_AND, 0},
14085 {"or", BINOP_BITWISE_IOR, PREC_BITWISE_IOR, 0},
14086 {"xor", BINOP_BITWISE_XOR, PREC_BITWISE_XOR, 0},
14087 {"and", BINOP_BITWISE_AND, PREC_BITWISE_AND, 0},
14088 {"=", BINOP_EQUAL, PREC_EQUAL, 0},
14089 {"/=", BINOP_NOTEQUAL, PREC_EQUAL, 0},
14090 {"<=", BINOP_LEQ, PREC_ORDER, 0},
14091 {">=", BINOP_GEQ, PREC_ORDER, 0},
14092 {">", BINOP_GTR, PREC_ORDER, 0},
14093 {"<", BINOP_LESS, PREC_ORDER, 0},
14094 {">>", BINOP_RSH, PREC_SHIFT, 0},
14095 {"<<", BINOP_LSH, PREC_SHIFT, 0},
14096 {"+", BINOP_ADD, PREC_ADD, 0},
14097 {"-", BINOP_SUB, PREC_ADD, 0},
14098 {"&", BINOP_CONCAT, PREC_ADD, 0},
14099 {"*", BINOP_MUL, PREC_MUL, 0},
14100 {"/", BINOP_DIV, PREC_MUL, 0},
14101 {"rem", BINOP_REM, PREC_MUL, 0},
14102 {"mod", BINOP_MOD, PREC_MUL, 0},
14103 {"**", BINOP_EXP, PREC_REPEAT, 0},
14104 {"@", BINOP_REPEAT, PREC_REPEAT, 0},
14105 {"-", UNOP_NEG, PREC_PREFIX, 0},
14106 {"+", UNOP_PLUS, PREC_PREFIX, 0},
14107 {"not ", UNOP_LOGICAL_NOT, PREC_PREFIX, 0},
14108 {"not ", UNOP_COMPLEMENT, PREC_PREFIX, 0},
14109 {"abs ", UNOP_ABS, PREC_PREFIX, 0},
14110 {".all", UNOP_IND, PREC_SUFFIX, 1},
14111 {"'access", UNOP_ADDR, PREC_SUFFIX, 1},
14112 {"'size", OP_ATR_SIZE, PREC_SUFFIX, 1},
14113 {NULL, OP_NULL, PREC_SUFFIX, 0}
14114 };
14115 \f
14116 enum ada_primitive_types {
14117 ada_primitive_type_int,
14118 ada_primitive_type_long,
14119 ada_primitive_type_short,
14120 ada_primitive_type_char,
14121 ada_primitive_type_float,
14122 ada_primitive_type_double,
14123 ada_primitive_type_void,
14124 ada_primitive_type_long_long,
14125 ada_primitive_type_long_double,
14126 ada_primitive_type_natural,
14127 ada_primitive_type_positive,
14128 ada_primitive_type_system_address,
14129 ada_primitive_type_storage_offset,
14130 nr_ada_primitive_types
14131 };
14132
14133 static void
14134 ada_language_arch_info (struct gdbarch *gdbarch,
14135 struct language_arch_info *lai)
14136 {
14137 const struct builtin_type *builtin = builtin_type (gdbarch);
14138
14139 lai->primitive_type_vector
14140 = GDBARCH_OBSTACK_CALLOC (gdbarch, nr_ada_primitive_types + 1,
14141 struct type *);
14142
14143 lai->primitive_type_vector [ada_primitive_type_int]
14144 = arch_integer_type (gdbarch, gdbarch_int_bit (gdbarch),
14145 0, "integer");
14146 lai->primitive_type_vector [ada_primitive_type_long]
14147 = arch_integer_type (gdbarch, gdbarch_long_bit (gdbarch),
14148 0, "long_integer");
14149 lai->primitive_type_vector [ada_primitive_type_short]
14150 = arch_integer_type (gdbarch, gdbarch_short_bit (gdbarch),
14151 0, "short_integer");
14152 lai->string_char_type
14153 = lai->primitive_type_vector [ada_primitive_type_char]
14154 = arch_character_type (gdbarch, TARGET_CHAR_BIT, 0, "character");
14155 lai->primitive_type_vector [ada_primitive_type_float]
14156 = arch_float_type (gdbarch, gdbarch_float_bit (gdbarch),
14157 "float", gdbarch_float_format (gdbarch));
14158 lai->primitive_type_vector [ada_primitive_type_double]
14159 = arch_float_type (gdbarch, gdbarch_double_bit (gdbarch),
14160 "long_float", gdbarch_double_format (gdbarch));
14161 lai->primitive_type_vector [ada_primitive_type_long_long]
14162 = arch_integer_type (gdbarch, gdbarch_long_long_bit (gdbarch),
14163 0, "long_long_integer");
14164 lai->primitive_type_vector [ada_primitive_type_long_double]
14165 = arch_float_type (gdbarch, gdbarch_long_double_bit (gdbarch),
14166 "long_long_float", gdbarch_long_double_format (gdbarch));
14167 lai->primitive_type_vector [ada_primitive_type_natural]
14168 = arch_integer_type (gdbarch, gdbarch_int_bit (gdbarch),
14169 0, "natural");
14170 lai->primitive_type_vector [ada_primitive_type_positive]
14171 = arch_integer_type (gdbarch, gdbarch_int_bit (gdbarch),
14172 0, "positive");
14173 lai->primitive_type_vector [ada_primitive_type_void]
14174 = builtin->builtin_void;
14175
14176 lai->primitive_type_vector [ada_primitive_type_system_address]
14177 = lookup_pointer_type (arch_type (gdbarch, TYPE_CODE_VOID, TARGET_CHAR_BIT,
14178 "void"));
14179 TYPE_NAME (lai->primitive_type_vector [ada_primitive_type_system_address])
14180 = "system__address";
14181
14182 /* Create the equivalent of the System.Storage_Elements.Storage_Offset
14183 type. This is a signed integral type whose size is the same as
14184 the size of addresses. */
14185 {
14186 unsigned int addr_length = TYPE_LENGTH
14187 (lai->primitive_type_vector [ada_primitive_type_system_address]);
14188
14189 lai->primitive_type_vector [ada_primitive_type_storage_offset]
14190 = arch_integer_type (gdbarch, addr_length * HOST_CHAR_BIT, 0,
14191 "storage_offset");
14192 }
14193
14194 lai->bool_type_symbol = NULL;
14195 lai->bool_type_default = builtin->builtin_bool;
14196 }
14197 \f
14198 /* Language vector */
14199
14200 /* Not really used, but needed in the ada_language_defn. */
14201
14202 static void
14203 emit_char (int c, struct type *type, struct ui_file *stream, int quoter)
14204 {
14205 ada_emit_char (c, type, stream, quoter, 1);
14206 }
14207
14208 static int
14209 parse (struct parser_state *ps)
14210 {
14211 warnings_issued = 0;
14212 return ada_parse (ps);
14213 }
14214
14215 static const struct exp_descriptor ada_exp_descriptor = {
14216 ada_print_subexp,
14217 ada_operator_length,
14218 ada_operator_check,
14219 ada_op_name,
14220 ada_dump_subexp_body,
14221 ada_evaluate_subexp
14222 };
14223
14224 /* symbol_name_matcher_ftype adapter for wild_match. */
14225
14226 static bool
14227 do_wild_match (const char *symbol_search_name,
14228 const lookup_name_info &lookup_name,
14229 completion_match_result *comp_match_res)
14230 {
14231 return wild_match (symbol_search_name, ada_lookup_name (lookup_name));
14232 }
14233
14234 /* symbol_name_matcher_ftype adapter for full_match. */
14235
14236 static bool
14237 do_full_match (const char *symbol_search_name,
14238 const lookup_name_info &lookup_name,
14239 completion_match_result *comp_match_res)
14240 {
14241 return full_match (symbol_search_name, ada_lookup_name (lookup_name));
14242 }
14243
14244 /* symbol_name_matcher_ftype for exact (verbatim) matches. */
14245
14246 static bool
14247 do_exact_match (const char *symbol_search_name,
14248 const lookup_name_info &lookup_name,
14249 completion_match_result *comp_match_res)
14250 {
14251 return strcmp (symbol_search_name, ada_lookup_name (lookup_name)) == 0;
14252 }
14253
14254 /* Build the Ada lookup name for LOOKUP_NAME. */
14255
14256 ada_lookup_name_info::ada_lookup_name_info (const lookup_name_info &lookup_name)
14257 {
14258 const std::string &user_name = lookup_name.name ();
14259
14260 if (user_name[0] == '<')
14261 {
14262 if (user_name.back () == '>')
14263 m_encoded_name = user_name.substr (1, user_name.size () - 2);
14264 else
14265 m_encoded_name = user_name.substr (1, user_name.size () - 1);
14266 m_encoded_p = true;
14267 m_verbatim_p = true;
14268 m_wild_match_p = false;
14269 m_standard_p = false;
14270 }
14271 else
14272 {
14273 m_verbatim_p = false;
14274
14275 m_encoded_p = user_name.find ("__") != std::string::npos;
14276
14277 if (!m_encoded_p)
14278 {
14279 const char *folded = ada_fold_name (user_name.c_str ());
14280 const char *encoded = ada_encode_1 (folded, false);
14281 if (encoded != NULL)
14282 m_encoded_name = encoded;
14283 else
14284 m_encoded_name = user_name;
14285 }
14286 else
14287 m_encoded_name = user_name;
14288
14289 /* Handle the 'package Standard' special case. See description
14290 of m_standard_p. */
14291 if (startswith (m_encoded_name.c_str (), "standard__"))
14292 {
14293 m_encoded_name = m_encoded_name.substr (sizeof ("standard__") - 1);
14294 m_standard_p = true;
14295 }
14296 else
14297 m_standard_p = false;
14298
14299 /* If the name contains a ".", then the user is entering a fully
14300 qualified entity name, and the match must not be done in wild
14301 mode. Similarly, if the user wants to complete what looks
14302 like an encoded name, the match must not be done in wild
14303 mode. Also, in the standard__ special case always do
14304 non-wild matching. */
14305 m_wild_match_p
14306 = (lookup_name.match_type () != symbol_name_match_type::FULL
14307 && !m_encoded_p
14308 && !m_standard_p
14309 && user_name.find ('.') == std::string::npos);
14310 }
14311 }
14312
14313 /* symbol_name_matcher_ftype method for Ada. This only handles
14314 completion mode. */
14315
14316 static bool
14317 ada_symbol_name_matches (const char *symbol_search_name,
14318 const lookup_name_info &lookup_name,
14319 completion_match_result *comp_match_res)
14320 {
14321 return lookup_name.ada ().matches (symbol_search_name,
14322 lookup_name.match_type (),
14323 comp_match_res);
14324 }
14325
14326 /* A name matcher that matches the symbol name exactly, with
14327 strcmp. */
14328
14329 static bool
14330 literal_symbol_name_matcher (const char *symbol_search_name,
14331 const lookup_name_info &lookup_name,
14332 completion_match_result *comp_match_res)
14333 {
14334 const std::string &name = lookup_name.name ();
14335
14336 int cmp = (lookup_name.completion_mode ()
14337 ? strncmp (symbol_search_name, name.c_str (), name.size ())
14338 : strcmp (symbol_search_name, name.c_str ()));
14339 if (cmp == 0)
14340 {
14341 if (comp_match_res != NULL)
14342 comp_match_res->set_match (symbol_search_name);
14343 return true;
14344 }
14345 else
14346 return false;
14347 }
14348
14349 /* Implement the "la_get_symbol_name_matcher" language_defn method for
14350 Ada. */
14351
14352 static symbol_name_matcher_ftype *
14353 ada_get_symbol_name_matcher (const lookup_name_info &lookup_name)
14354 {
14355 if (lookup_name.match_type () == symbol_name_match_type::SEARCH_NAME)
14356 return literal_symbol_name_matcher;
14357
14358 if (lookup_name.completion_mode ())
14359 return ada_symbol_name_matches;
14360 else
14361 {
14362 if (lookup_name.ada ().wild_match_p ())
14363 return do_wild_match;
14364 else if (lookup_name.ada ().verbatim_p ())
14365 return do_exact_match;
14366 else
14367 return do_full_match;
14368 }
14369 }
14370
14371 /* Implement the "la_read_var_value" language_defn method for Ada. */
14372
14373 static struct value *
14374 ada_read_var_value (struct symbol *var, const struct block *var_block,
14375 struct frame_info *frame)
14376 {
14377 const struct block *frame_block = NULL;
14378 struct symbol *renaming_sym = NULL;
14379
14380 /* The only case where default_read_var_value is not sufficient
14381 is when VAR is a renaming... */
14382 if (frame)
14383 frame_block = get_frame_block (frame, NULL);
14384 if (frame_block)
14385 renaming_sym = ada_find_renaming_symbol (var, frame_block);
14386 if (renaming_sym != NULL)
14387 return ada_read_renaming_var_value (renaming_sym, frame_block);
14388
14389 /* This is a typical case where we expect the default_read_var_value
14390 function to work. */
14391 return default_read_var_value (var, var_block, frame);
14392 }
14393
14394 static const char *ada_extensions[] =
14395 {
14396 ".adb", ".ads", ".a", ".ada", ".dg", NULL
14397 };
14398
14399 extern const struct language_defn ada_language_defn = {
14400 "ada", /* Language name */
14401 "Ada",
14402 language_ada,
14403 range_check_off,
14404 case_sensitive_on, /* Yes, Ada is case-insensitive, but
14405 that's not quite what this means. */
14406 array_row_major,
14407 macro_expansion_no,
14408 ada_extensions,
14409 &ada_exp_descriptor,
14410 parse,
14411 resolve,
14412 ada_printchar, /* Print a character constant */
14413 ada_printstr, /* Function to print string constant */
14414 emit_char, /* Function to print single char (not used) */
14415 ada_print_type, /* Print a type using appropriate syntax */
14416 ada_print_typedef, /* Print a typedef using appropriate syntax */
14417 ada_val_print, /* Print a value using appropriate syntax */
14418 ada_value_print, /* Print a top-level value */
14419 ada_read_var_value, /* la_read_var_value */
14420 NULL, /* Language specific skip_trampoline */
14421 NULL, /* name_of_this */
14422 true, /* la_store_sym_names_in_linkage_form_p */
14423 ada_lookup_symbol_nonlocal, /* Looking up non-local symbols. */
14424 basic_lookup_transparent_type, /* lookup_transparent_type */
14425 ada_la_decode, /* Language specific symbol demangler */
14426 ada_sniff_from_mangled_name,
14427 NULL, /* Language specific
14428 class_name_from_physname */
14429 ada_op_print_tab, /* expression operators for printing */
14430 0, /* c-style arrays */
14431 1, /* String lower bound */
14432 ada_get_gdb_completer_word_break_characters,
14433 ada_collect_symbol_completion_matches,
14434 ada_language_arch_info,
14435 ada_print_array_index,
14436 default_pass_by_reference,
14437 c_get_string,
14438 ada_watch_location_expression,
14439 ada_get_symbol_name_matcher, /* la_get_symbol_name_matcher */
14440 ada_iterate_over_symbols,
14441 default_search_name_hash,
14442 &ada_varobj_ops,
14443 NULL,
14444 NULL,
14445 ada_is_string_type,
14446 "(...)" /* la_struct_too_deep_ellipsis */
14447 };
14448
14449 /* Command-list for the "set/show ada" prefix command. */
14450 static struct cmd_list_element *set_ada_list;
14451 static struct cmd_list_element *show_ada_list;
14452
14453 /* Implement the "set ada" prefix command. */
14454
14455 static void
14456 set_ada_command (const char *arg, int from_tty)
14457 {
14458 printf_unfiltered (_(\
14459 "\"set ada\" must be followed by the name of a setting.\n"));
14460 help_list (set_ada_list, "set ada ", all_commands, gdb_stdout);
14461 }
14462
14463 /* Implement the "show ada" prefix command. */
14464
14465 static void
14466 show_ada_command (const char *args, int from_tty)
14467 {
14468 cmd_show_list (show_ada_list, from_tty, "");
14469 }
14470
14471 static void
14472 initialize_ada_catchpoint_ops (void)
14473 {
14474 struct breakpoint_ops *ops;
14475
14476 initialize_breakpoint_ops ();
14477
14478 ops = &catch_exception_breakpoint_ops;
14479 *ops = bkpt_breakpoint_ops;
14480 ops->allocate_location = allocate_location_catch_exception;
14481 ops->re_set = re_set_catch_exception;
14482 ops->check_status = check_status_catch_exception;
14483 ops->print_it = print_it_catch_exception;
14484 ops->print_one = print_one_catch_exception;
14485 ops->print_mention = print_mention_catch_exception;
14486 ops->print_recreate = print_recreate_catch_exception;
14487
14488 ops = &catch_exception_unhandled_breakpoint_ops;
14489 *ops = bkpt_breakpoint_ops;
14490 ops->allocate_location = allocate_location_catch_exception_unhandled;
14491 ops->re_set = re_set_catch_exception_unhandled;
14492 ops->check_status = check_status_catch_exception_unhandled;
14493 ops->print_it = print_it_catch_exception_unhandled;
14494 ops->print_one = print_one_catch_exception_unhandled;
14495 ops->print_mention = print_mention_catch_exception_unhandled;
14496 ops->print_recreate = print_recreate_catch_exception_unhandled;
14497
14498 ops = &catch_assert_breakpoint_ops;
14499 *ops = bkpt_breakpoint_ops;
14500 ops->allocate_location = allocate_location_catch_assert;
14501 ops->re_set = re_set_catch_assert;
14502 ops->check_status = check_status_catch_assert;
14503 ops->print_it = print_it_catch_assert;
14504 ops->print_one = print_one_catch_assert;
14505 ops->print_mention = print_mention_catch_assert;
14506 ops->print_recreate = print_recreate_catch_assert;
14507
14508 ops = &catch_handlers_breakpoint_ops;
14509 *ops = bkpt_breakpoint_ops;
14510 ops->allocate_location = allocate_location_catch_handlers;
14511 ops->re_set = re_set_catch_handlers;
14512 ops->check_status = check_status_catch_handlers;
14513 ops->print_it = print_it_catch_handlers;
14514 ops->print_one = print_one_catch_handlers;
14515 ops->print_mention = print_mention_catch_handlers;
14516 ops->print_recreate = print_recreate_catch_handlers;
14517 }
14518
14519 /* This module's 'new_objfile' observer. */
14520
14521 static void
14522 ada_new_objfile_observer (struct objfile *objfile)
14523 {
14524 ada_clear_symbol_cache ();
14525 }
14526
14527 /* This module's 'free_objfile' observer. */
14528
14529 static void
14530 ada_free_objfile_observer (struct objfile *objfile)
14531 {
14532 ada_clear_symbol_cache ();
14533 }
14534
14535 void
14536 _initialize_ada_language (void)
14537 {
14538 initialize_ada_catchpoint_ops ();
14539
14540 add_prefix_cmd ("ada", no_class, set_ada_command,
14541 _("Prefix command for changing Ada-specific settings"),
14542 &set_ada_list, "set ada ", 0, &setlist);
14543
14544 add_prefix_cmd ("ada", no_class, show_ada_command,
14545 _("Generic command for showing Ada-specific settings."),
14546 &show_ada_list, "show ada ", 0, &showlist);
14547
14548 add_setshow_boolean_cmd ("trust-PAD-over-XVS", class_obscure,
14549 &trust_pad_over_xvs, _("\
14550 Enable or disable an optimization trusting PAD types over XVS types"), _("\
14551 Show whether an optimization trusting PAD types over XVS types is activated"),
14552 _("\
14553 This is related to the encoding used by the GNAT compiler. The debugger\n\
14554 should normally trust the contents of PAD types, but certain older versions\n\
14555 of GNAT have a bug that sometimes causes the information in the PAD type\n\
14556 to be incorrect. Turning this setting \"off\" allows the debugger to\n\
14557 work around this bug. It is always safe to turn this option \"off\", but\n\
14558 this incurs a slight performance penalty, so it is recommended to NOT change\n\
14559 this option to \"off\" unless necessary."),
14560 NULL, NULL, &set_ada_list, &show_ada_list);
14561
14562 add_setshow_boolean_cmd ("print-signatures", class_vars,
14563 &print_signatures, _("\
14564 Enable or disable the output of formal and return types for functions in the \
14565 overloads selection menu"), _("\
14566 Show whether the output of formal and return types for functions in the \
14567 overloads selection menu is activated"),
14568 NULL, NULL, NULL, &set_ada_list, &show_ada_list);
14569
14570 add_catch_command ("exception", _("\
14571 Catch Ada exceptions, when raised.\n\
14572 Usage: catch exception [ ARG ]\n\
14573 \n\
14574 Without any argument, stop when any Ada exception is raised.\n\
14575 If ARG is \"unhandled\" (without the quotes), only stop when the exception\n\
14576 being raised does not have a handler (and will therefore lead to the task's\n\
14577 termination).\n\
14578 Otherwise, the catchpoint only stops when the name of the exception being\n\
14579 raised is the same as ARG."),
14580 catch_ada_exception_command,
14581 catch_ada_completer,
14582 CATCH_PERMANENT,
14583 CATCH_TEMPORARY);
14584
14585 add_catch_command ("handlers", _("\
14586 Catch Ada exceptions, when handled.\n\
14587 With an argument, catch only exceptions with the given name."),
14588 catch_ada_handlers_command,
14589 catch_ada_completer,
14590 CATCH_PERMANENT,
14591 CATCH_TEMPORARY);
14592 add_catch_command ("assert", _("\
14593 Catch failed Ada assertions, when raised.\n\
14594 With an argument, catch only exceptions with the given name."),
14595 catch_assert_command,
14596 NULL,
14597 CATCH_PERMANENT,
14598 CATCH_TEMPORARY);
14599
14600 varsize_limit = 65536;
14601 add_setshow_uinteger_cmd ("varsize-limit", class_support,
14602 &varsize_limit, _("\
14603 Set the maximum number of bytes allowed in a variable-size object."), _("\
14604 Show the maximum number of bytes allowed in a variable-size object."), _("\
14605 Attempts to access an object whose size is not a compile-time constant\n\
14606 and exceeds this limit will cause an error."),
14607 NULL, NULL, &setlist, &showlist);
14608
14609 add_info ("exceptions", info_exceptions_command,
14610 _("\
14611 List all Ada exception names.\n\
14612 If a regular expression is passed as an argument, only those matching\n\
14613 the regular expression are listed."));
14614
14615 add_prefix_cmd ("ada", class_maintenance, maint_set_ada_cmd,
14616 _("Set Ada maintenance-related variables."),
14617 &maint_set_ada_cmdlist, "maintenance set ada ",
14618 0/*allow-unknown*/, &maintenance_set_cmdlist);
14619
14620 add_prefix_cmd ("ada", class_maintenance, maint_show_ada_cmd,
14621 _("Show Ada maintenance-related variables"),
14622 &maint_show_ada_cmdlist, "maintenance show ada ",
14623 0/*allow-unknown*/, &maintenance_show_cmdlist);
14624
14625 add_setshow_boolean_cmd
14626 ("ignore-descriptive-types", class_maintenance,
14627 &ada_ignore_descriptive_types_p,
14628 _("Set whether descriptive types generated by GNAT should be ignored."),
14629 _("Show whether descriptive types generated by GNAT should be ignored."),
14630 _("\
14631 When enabled, the debugger will stop using the DW_AT_GNAT_descriptive_type\n\
14632 DWARF attribute."),
14633 NULL, NULL, &maint_set_ada_cmdlist, &maint_show_ada_cmdlist);
14634
14635 decoded_names_store = htab_create_alloc (256, htab_hash_string, streq_hash,
14636 NULL, xcalloc, xfree);
14637
14638 /* The ada-lang observers. */
14639 gdb::observers::new_objfile.attach (ada_new_objfile_observer);
14640 gdb::observers::free_objfile.attach (ada_free_objfile_observer);
14641 gdb::observers::inferior_exit.attach (ada_inferior_exit);
14642 }
This page took 0.413779 seconds and 4 git commands to generate.