gdb/valarith.c

   1 /* Perform arithmetic and other operations on values, for GDB.
   2
   3    Copyright 1986, 1988, 1989, 1990, 1991, 1992, 1993, 1994, 1995,
   4    1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004 Free Software
   5    Foundation, Inc.
   6
   7    This file is part of GDB.
   8
   9    This program is free software; you can redistribute it and/or modify
  10    it under the terms of the GNU General Public License as published by
  11    the Free Software Foundation; either version 2 of the License, or
  12    (at your option) any later version.
  13
  14    This program is distributed in the hope that it will be useful,
  15    but WITHOUT ANY WARRANTY; without even the implied warranty of
  16    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17    GNU General Public License for more details.
  18
  19    You should have received a copy of the GNU General Public License
  20    along with this program; if not, write to the Free Software
  21    Foundation, Inc., 59 Temple Place - Suite 330,
  22    Boston, MA 02111-1307, USA.  */
  23
  24 #include "defs.h"
  25 #include "value.h"
  26 #include "symtab.h"
  27 #include "gdbtypes.h"
  28 #include "expression.h"
  29 #include "target.h"
  30 #include "language.h"
  31 #include "gdb_string.h"
  32 #include "doublest.h"
  33 #include <math.h>
  34 #include "infcall.h"
  35
  36 /* Define whether or not the C operator '/' truncates towards zero for
  37    differently signed operands (truncation direction is undefined in C). */
  38
  39 #ifndef TRUNCATION_TOWARDS_ZERO
  40 #define TRUNCATION_TOWARDS_ZERO ((-5 / 2) == -2)
  41 #endif
  42
  43 static struct value *value_subscripted_rvalue (struct value *, struct value *, int);
  44
  45 void _initialize_valarith (void);
  46 \f
  47
  48 /* Given a pointer, return the size of its target.
  49    If the pointer type is void *, then return 1.
  50    If the target type is incomplete, then error out.
  51    This isn't a general purpose function, but just a
  52    helper for value_sub & value_add.
  53 */
  54
  55 static LONGEST
  56 find_size_for_pointer_math (struct type *ptr_type)
  57 {
  58   LONGEST sz = -1;
  59   struct type *ptr_target;
  60
  61   ptr_target = check_typedef (TYPE_TARGET_TYPE (ptr_type));
  62
  63   sz = TYPE_LENGTH (ptr_target);
  64   if (sz == 0)
  65     {
  66       if (TYPE_CODE (ptr_type) == TYPE_CODE_VOID)
  67         sz = 1;
  68       else
  69         {
  70           char *name;
  71
  72           name = TYPE_NAME (ptr_target);
  73           if (name == NULL)
  74             name = TYPE_TAG_NAME (ptr_target);
  75           if (name == NULL)
  76             error ("Cannot perform pointer math on incomplete types, "
  77                    "try casting to a known type, or void *.");
  78           else
  79             error ("Cannot perform pointer math on incomplete type \"%s\", "
  80                    "try casting to a known type, or void *.", name);
  81         }
  82     }
  83   return sz;
  84 }
  85
  86 struct value *
  87 value_add (struct value *arg1, struct value *arg2)
  88 {
  89   struct value *valint;
  90   struct value *valptr;
  91   LONGEST sz;
  92   struct type *type1, *type2, *valptrtype;
  93
  94   COERCE_ARRAY (arg1);
  95   COERCE_ARRAY (arg2);
  96   type1 = check_typedef (VALUE_TYPE (arg1));
  97   type2 = check_typedef (VALUE_TYPE (arg2));
  98
  99   if ((TYPE_CODE (type1) == TYPE_CODE_PTR
 100        || TYPE_CODE (type2) == TYPE_CODE_PTR)
 101       &&
 102       (is_integral_type (type1) || is_integral_type (type2)))
 103     /* Exactly one argument is a pointer, and one is an integer.  */
 104     {
 105       struct value *retval;
 106
 107       if (TYPE_CODE (type1) == TYPE_CODE_PTR)
 108         {
 109           valptr = arg1;
 110           valint = arg2;
 111           valptrtype = type1;
 112         }
 113       else
 114         {
 115           valptr = arg2;
 116           valint = arg1;
 117           valptrtype = type2;
 118         }
 119
 120       sz = find_size_for_pointer_math (valptrtype);
 121
 122       retval = value_from_pointer (valptrtype,
 123                                    value_as_address (valptr)
 124                                    + (sz * value_as_long (valint)));
 125       VALUE_BFD_SECTION (retval) = VALUE_BFD_SECTION (valptr);
 126       return retval;
 127     }
 128
 129   return value_binop (arg1, arg2, BINOP_ADD);
 130 }
 131
 132 struct value *
 133 value_sub (struct value *arg1, struct value *arg2)
 134 {
 135   struct type *type1, *type2;
 136   COERCE_ARRAY (arg1);
 137   COERCE_ARRAY (arg2);
 138   type1 = check_typedef (VALUE_TYPE (arg1));
 139   type2 = check_typedef (VALUE_TYPE (arg2));
 140
 141   if (TYPE_CODE (type1) == TYPE_CODE_PTR)
 142     {
 143       if (is_integral_type (type2))
 144         {
 145           /* pointer - integer.  */
 146           LONGEST sz = find_size_for_pointer_math (type1);
 147
 148           return value_from_pointer (type1,
 149                                      (value_as_address (arg1)
 150                                       - (sz * value_as_long (arg2))));
 151         }
 152       else if (TYPE_CODE (type2) == TYPE_CODE_PTR
 153                && TYPE_LENGTH (check_typedef (TYPE_TARGET_TYPE (type1)))
 154                == TYPE_LENGTH (check_typedef (TYPE_TARGET_TYPE (type2))))
 155         {
 156           /* pointer to <type x> - pointer to <type x>.  */
 157           LONGEST sz = TYPE_LENGTH (check_typedef (TYPE_TARGET_TYPE (type1)));
 158           return value_from_longest
 159             (builtin_type_long, /* FIXME -- should be ptrdiff_t */
 160              (value_as_long (arg1) - value_as_long (arg2)) / sz);
 161         }
 162       else
 163         {
 164           error ("\
 165 First argument of `-' is a pointer and second argument is neither\n\
 166 an integer nor a pointer of the same type.");
 167         }
 168     }
 169
 170   return value_binop (arg1, arg2, BINOP_SUB);
 171 }
 172
 173 /* Return the value of ARRAY[IDX].
 174    See comments in value_coerce_array() for rationale for reason for
 175    doing lower bounds adjustment here rather than there.
 176    FIXME:  Perhaps we should validate that the index is valid and if
 177    verbosity is set, warn about invalid indices (but still use them). */
 178
 179 struct value *
 180 value_subscript (struct value *array, struct value *idx)
 181 {
 182   struct value *bound;
 183   int c_style = current_language->c_style_arrays;
 184   struct type *tarray;
 185
 186   COERCE_REF (array);
 187   tarray = check_typedef (VALUE_TYPE (array));
 188   COERCE_VARYING_ARRAY (array, tarray);
 189
 190   if (TYPE_CODE (tarray) == TYPE_CODE_ARRAY
 191       || TYPE_CODE (tarray) == TYPE_CODE_STRING)
 192     {
 193       struct type *range_type = TYPE_INDEX_TYPE (tarray);
 194       LONGEST lowerbound, upperbound;
 195       get_discrete_bounds (range_type, &lowerbound, &upperbound);
 196
 197       if (VALUE_LVAL (array) != lval_memory)
 198         return value_subscripted_rvalue (array, idx, lowerbound);
 199
 200       if (c_style == 0)
 201         {
 202           LONGEST index = value_as_long (idx);
 203           if (index >= lowerbound && index <= upperbound)
 204             return value_subscripted_rvalue (array, idx, lowerbound);
 205           warning ("array or string index out of range");
 206           /* fall doing C stuff */
 207           c_style = 1;
 208         }
 209
 210       if (lowerbound != 0)
 211         {
 212           bound = value_from_longest (builtin_type_int, (LONGEST) lowerbound);
 213           idx = value_sub (idx, bound);
 214         }
 215
 216       array = value_coerce_array (array);
 217     }
 218
 219   if (TYPE_CODE (tarray) == TYPE_CODE_BITSTRING)
 220     {
 221       struct type *range_type = TYPE_INDEX_TYPE (tarray);
 222       LONGEST index = value_as_long (idx);
 223       struct value *v;
 224       int offset, byte, bit_index;
 225       LONGEST lowerbound, upperbound;
 226       get_discrete_bounds (range_type, &lowerbound, &upperbound);
 227       if (index < lowerbound || index > upperbound)
 228         error ("bitstring index out of range");
 229       index -= lowerbound;
 230       offset = index / TARGET_CHAR_BIT;
 231       byte = *((char *) VALUE_CONTENTS (array) + offset);
 232       bit_index = index % TARGET_CHAR_BIT;
 233       byte >>= (BITS_BIG_ENDIAN ? TARGET_CHAR_BIT - 1 - bit_index : bit_index);
 234       v = value_from_longest (LA_BOOL_TYPE, byte & 1);
 235       VALUE_BITPOS (v) = bit_index;
 236       VALUE_BITSIZE (v) = 1;
 237       VALUE_LVAL (v) = VALUE_LVAL (array);
 238       if (VALUE_LVAL (array) == lval_internalvar)
 239         VALUE_LVAL (v) = lval_internalvar_component;
 240       VALUE_ADDRESS (v) = VALUE_ADDRESS (array);
 241       VALUE_OFFSET (v) = offset + VALUE_OFFSET (array);
 242       return v;
 243     }
 244
 245   if (c_style)
 246     return value_ind (value_add (array, idx));
 247   else
 248     error ("not an array or string");
 249 }
 250
 251 /* Return the value of EXPR[IDX], expr an aggregate rvalue
 252    (eg, a vector register).  This routine used to promote floats
 253    to doubles, but no longer does.  */
 254
 255 static struct value *
 256 value_subscripted_rvalue (struct value *array, struct value *idx, int lowerbound)
 257 {
 258   struct type *array_type = check_typedef (VALUE_TYPE (array));
 259   struct type *elt_type = check_typedef (TYPE_TARGET_TYPE (array_type));
 260   unsigned int elt_size = TYPE_LENGTH (elt_type);
 261   LONGEST index = value_as_long (idx);
 262   unsigned int elt_offs = elt_size * longest_to_int (index - lowerbound);
 263   struct value *v;
 264
 265   if (index < lowerbound || elt_offs >= TYPE_LENGTH (array_type))
 266     error ("no such vector element");
 267
 268   v = allocate_value (elt_type);
 269   if (VALUE_LAZY (array))
 270     VALUE_LAZY (v) = 1;
 271   else
 272     memcpy (VALUE_CONTENTS (v), VALUE_CONTENTS (array) + elt_offs, elt_size);
 273
 274   if (VALUE_LVAL (array) == lval_internalvar)
 275     VALUE_LVAL (v) = lval_internalvar_component;
 276   else
 277     VALUE_LVAL (v) = VALUE_LVAL (array);
 278   VALUE_ADDRESS (v) = VALUE_ADDRESS (array);
 279   VALUE_REGNO (v) = VALUE_REGNO (array);
 280   VALUE_OFFSET (v) = VALUE_OFFSET (array) + elt_offs;
 281   return v;
 282 }
 283 \f
 284 /* Check to see if either argument is a structure.  This is called so
 285    we know whether to go ahead with the normal binop or look for a
 286    user defined function instead.
 287
 288    For now, we do not overload the `=' operator.  */
 289
 290 int
 291 binop_user_defined_p (enum exp_opcode op, struct value *arg1, struct value *arg2)
 292 {
 293   struct type *type1, *type2;
 294   if (op == BINOP_ASSIGN || op == BINOP_CONCAT)
 295     return 0;
 296   type1 = check_typedef (VALUE_TYPE (arg1));
 297   type2 = check_typedef (VALUE_TYPE (arg2));
 298   return (TYPE_CODE (type1) == TYPE_CODE_STRUCT
 299           || TYPE_CODE (type2) == TYPE_CODE_STRUCT
 300           || (TYPE_CODE (type1) == TYPE_CODE_REF
 301               && TYPE_CODE (TYPE_TARGET_TYPE (type1)) == TYPE_CODE_STRUCT)
 302           || (TYPE_CODE (type2) == TYPE_CODE_REF
 303               && TYPE_CODE (TYPE_TARGET_TYPE (type2)) == TYPE_CODE_STRUCT));
 304 }
 305
 306 /* Check to see if argument is a structure.  This is called so
 307    we know whether to go ahead with the normal unop or look for a
 308    user defined function instead.
 309
 310    For now, we do not overload the `&' operator.  */
 311
 312 int
 313 unop_user_defined_p (enum exp_opcode op, struct value *arg1)
 314 {
 315   struct type *type1;
 316   if (op == UNOP_ADDR)
 317     return 0;
 318   type1 = check_typedef (VALUE_TYPE (arg1));
 319   for (;;)
 320     {
 321       if (TYPE_CODE (type1) == TYPE_CODE_STRUCT)
 322         return 1;
 323       else if (TYPE_CODE (type1) == TYPE_CODE_REF)
 324         type1 = TYPE_TARGET_TYPE (type1);
 325       else
 326         return 0;
 327     }
 328 }
 329
 330 /* We know either arg1 or arg2 is a structure, so try to find the right
 331    user defined function.  Create an argument vector that calls
 332    arg1.operator @ (arg1,arg2) and return that value (where '@' is any
 333    binary operator which is legal for GNU C++).
 334
 335    OP is the operatore, and if it is BINOP_ASSIGN_MODIFY, then OTHEROP
 336    is the opcode saying how to modify it.  Otherwise, OTHEROP is
 337    unused.  */
 338
 339 struct value *
 340 value_x_binop (struct value *arg1, struct value *arg2, enum exp_opcode op,
 341                enum exp_opcode otherop, enum noside noside)
 342 {
 343   struct value **argvec;
 344   char *ptr;
 345   char tstr[13];
 346   int static_memfuncp;
 347
 348   COERCE_REF (arg1);
 349   COERCE_REF (arg2);
 350   COERCE_ENUM (arg1);
 351   COERCE_ENUM (arg2);
 352
 353   /* now we know that what we have to do is construct our
 354      arg vector and find the right function to call it with.  */
 355
 356   if (TYPE_CODE (check_typedef (VALUE_TYPE (arg1))) != TYPE_CODE_STRUCT)
 357     error ("Can't do that binary op on that type");     /* FIXME be explicit */
 358
 359   argvec = (struct value **) alloca (sizeof (struct value *) * 4);
 360   argvec[1] = value_addr (arg1);
 361   argvec[2] = arg2;
 362   argvec[3] = 0;
 363
 364   /* make the right function name up */
 365   strcpy (tstr, "operator__");
 366   ptr = tstr + 8;
 367   switch (op)
 368     {
 369     case BINOP_ADD:
 370       strcpy (ptr, "+");
 371       break;
 372     case BINOP_SUB:
 373       strcpy (ptr, "-");
 374       break;
 375     case BINOP_MUL:
 376       strcpy (ptr, "*");
 377       break;
 378     case BINOP_DIV:
 379       strcpy (ptr, "/");
 380       break;
 381     case BINOP_REM:
 382       strcpy (ptr, "%");
 383       break;
 384     case BINOP_LSH:
 385       strcpy (ptr, "<<");
 386       break;
 387     case BINOP_RSH:
 388       strcpy (ptr, ">>");
 389       break;
 390     case BINOP_BITWISE_AND:
 391       strcpy (ptr, "&");
 392       break;
 393     case BINOP_BITWISE_IOR:
 394       strcpy (ptr, "|");
 395       break;
 396     case BINOP_BITWISE_XOR:
 397       strcpy (ptr, "^");
 398       break;
 399     case BINOP_LOGICAL_AND:
 400       strcpy (ptr, "&&");
 401       break;
 402     case BINOP_LOGICAL_OR:
 403       strcpy (ptr, "||");
 404       break;
 405     case BINOP_MIN:
 406       strcpy (ptr, "<?");
 407       break;
 408     case BINOP_MAX:
 409       strcpy (ptr, ">?");
 410       break;
 411     case BINOP_ASSIGN:
 412       strcpy (ptr, "=");
 413       break;
 414     case BINOP_ASSIGN_MODIFY:
 415       switch (otherop)
 416         {
 417         case BINOP_ADD:
 418           strcpy (ptr, "+=");
 419           break;
 420         case BINOP_SUB:
 421           strcpy (ptr, "-=");
 422           break;
 423         case BINOP_MUL:
 424           strcpy (ptr, "*=");
 425           break;
 426         case BINOP_DIV:
 427           strcpy (ptr, "/=");
 428           break;
 429         case BINOP_REM:
 430           strcpy (ptr, "%=");
 431           break;
 432         case BINOP_BITWISE_AND:
 433           strcpy (ptr, "&=");
 434           break;
 435         case BINOP_BITWISE_IOR:
 436           strcpy (ptr, "|=");
 437           break;
 438         case BINOP_BITWISE_XOR:
 439           strcpy (ptr, "^=");
 440           break;
 441         case BINOP_MOD: /* invalid */
 442         default:
 443           error ("Invalid binary operation specified.");
 444         }
 445       break;
 446     case BINOP_SUBSCRIPT:
 447       strcpy (ptr, "[]");
 448       break;
 449     case BINOP_EQUAL:
 450       strcpy (ptr, "==");
 451       break;
 452     case BINOP_NOTEQUAL:
 453       strcpy (ptr, "!=");
 454       break;
 455     case BINOP_LESS:
 456       strcpy (ptr, "<");
 457       break;
 458     case BINOP_GTR:
 459       strcpy (ptr, ">");
 460       break;
 461     case BINOP_GEQ:
 462       strcpy (ptr, ">=");
 463       break;
 464     case BINOP_LEQ:
 465       strcpy (ptr, "<=");
 466       break;
 467     case BINOP_MOD:             /* invalid */
 468     default:
 469       error ("Invalid binary operation specified.");
 470     }
 471
 472   argvec[0] = value_struct_elt (&arg1, argvec + 1, tstr, &static_memfuncp, "structure");
 473
 474   if (argvec[0])
 475     {
 476       if (static_memfuncp)
 477         {
 478           argvec[1] = argvec[0];
 479           argvec++;
 480         }
 481       if (noside == EVAL_AVOID_SIDE_EFFECTS)
 482         {
 483           struct type *return_type;
 484           return_type
 485             = TYPE_TARGET_TYPE (check_typedef (VALUE_TYPE (argvec[0])));
 486           return value_zero (return_type, VALUE_LVAL (arg1));
 487         }
 488       return call_function_by_hand (argvec[0], 2 - static_memfuncp, argvec + 1);
 489     }
 490   error ("member function %s not found", tstr);
 491 #ifdef lint
 492   return call_function_by_hand (argvec[0], 2 - static_memfuncp, argvec + 1);
 493 #endif
 494 }
 495
 496 /* We know that arg1 is a structure, so try to find a unary user
 497    defined operator that matches the operator in question.
 498    Create an argument vector that calls arg1.operator @ (arg1)
 499    and return that value (where '@' is (almost) any unary operator which
 500    is legal for GNU C++).  */
 501
 502 struct value *
 503 value_x_unop (struct value *arg1, enum exp_opcode op, enum noside noside)
 504 {
 505   struct value **argvec;
 506   char *ptr, *mangle_ptr;
 507   char tstr[13], mangle_tstr[13];
 508   int static_memfuncp, nargs;
 509
 510   COERCE_REF (arg1);
 511   COERCE_ENUM (arg1);
 512
 513   /* now we know that what we have to do is construct our
 514      arg vector and find the right function to call it with.  */
 515
 516   if (TYPE_CODE (check_typedef (VALUE_TYPE (arg1))) != TYPE_CODE_STRUCT)
 517     error ("Can't do that unary op on that type");      /* FIXME be explicit */
 518
 519   argvec = (struct value **) alloca (sizeof (struct value *) * 4);
 520   argvec[1] = value_addr (arg1);
 521   argvec[2] = 0;
 522
 523   nargs = 1;
 524
 525   /* make the right function name up */
 526   strcpy (tstr, "operator__");
 527   ptr = tstr + 8;
 528   strcpy (mangle_tstr, "__");
 529   mangle_ptr = mangle_tstr + 2;
 530   switch (op)
 531     {
 532     case UNOP_PREINCREMENT:
 533       strcpy (ptr, "++");
 534       break;
 535     case UNOP_PREDECREMENT:
 536       strcpy (ptr, "--");
 537       break;
 538     case UNOP_POSTINCREMENT:
 539       strcpy (ptr, "++");
 540       argvec[2] = value_from_longest (builtin_type_int, 0);
 541       argvec[3] = 0;
 542       nargs ++;
 543       break;
 544     case UNOP_POSTDECREMENT:
 545       strcpy (ptr, "--");
 546       argvec[2] = value_from_longest (builtin_type_int, 0);
 547       argvec[3] = 0;
 548       nargs ++;
 549       break;
 550     case UNOP_LOGICAL_NOT:
 551       strcpy (ptr, "!");
 552       break;
 553     case UNOP_COMPLEMENT:
 554       strcpy (ptr, "~");
 555       break;
 556     case UNOP_NEG:
 557       strcpy (ptr, "-");
 558       break;
 559     case UNOP_IND:
 560       strcpy (ptr, "*");
 561       break;
 562     default:
 563       error ("Invalid unary operation specified.");
 564     }
 565
 566   argvec[0] = value_struct_elt (&arg1, argvec + 1, tstr, &static_memfuncp, "structure");
 567
 568   if (argvec[0])
 569     {
 570       if (static_memfuncp)
 571         {
 572           argvec[1] = argvec[0];
 573           nargs --;
 574           argvec++;
 575         }
 576       if (noside == EVAL_AVOID_SIDE_EFFECTS)
 577         {
 578           struct type *return_type;
 579           return_type
 580             = TYPE_TARGET_TYPE (check_typedef (VALUE_TYPE (argvec[0])));
 581           return value_zero (return_type, VALUE_LVAL (arg1));
 582         }
 583       return call_function_by_hand (argvec[0], nargs, argvec + 1);
 584     }
 585   error ("member function %s not found", tstr);
 586   return 0;                     /* For lint -- never reached */
 587 }
 588 \f
 589
 590 /* Concatenate two values with the following conditions:
 591
 592    (1)  Both values must be either bitstring values or character string
 593    values and the resulting value consists of the concatenation of
 594    ARG1 followed by ARG2.
 595
 596    or
 597
 598    One value must be an integer value and the other value must be
 599    either a bitstring value or character string value, which is
 600    to be repeated by the number of times specified by the integer
 601    value.
 602
 603
 604    (2)  Boolean values are also allowed and are treated as bit string
 605    values of length 1.
 606
 607    (3)  Character values are also allowed and are treated as character
 608    string values of length 1.
 609  */
 610
 611 struct value *
 612 value_concat (struct value *arg1, struct value *arg2)
 613 {
 614   struct value *inval1;
 615   struct value *inval2;
 616   struct value *outval = NULL;
 617   int inval1len, inval2len;
 618   int count, idx;
 619   char *ptr;
 620   char inchar;
 621   struct type *type1 = check_typedef (VALUE_TYPE (arg1));
 622   struct type *type2 = check_typedef (VALUE_TYPE (arg2));
 623
 624   COERCE_VARYING_ARRAY (arg1, type1);
 625   COERCE_VARYING_ARRAY (arg2, type2);
 626
 627   /* First figure out if we are dealing with two values to be concatenated
 628      or a repeat count and a value to be repeated.  INVAL1 is set to the
 629      first of two concatenated values, or the repeat count.  INVAL2 is set
 630      to the second of the two concatenated values or the value to be
 631      repeated. */
 632
 633   if (TYPE_CODE (type2) == TYPE_CODE_INT)
 634     {
 635       struct type *tmp = type1;
 636       type1 = tmp;
 637       tmp = type2;
 638       inval1 = arg2;
 639       inval2 = arg1;
 640     }
 641   else
 642     {
 643       inval1 = arg1;
 644       inval2 = arg2;
 645     }
 646
 647   /* Now process the input values. */
 648
 649   if (TYPE_CODE (type1) == TYPE_CODE_INT)
 650     {
 651       /* We have a repeat count.  Validate the second value and then
 652          construct a value repeated that many times. */
 653       if (TYPE_CODE (type2) == TYPE_CODE_STRING
 654           || TYPE_CODE (type2) == TYPE_CODE_CHAR)
 655         {
 656           count = longest_to_int (value_as_long (inval1));
 657           inval2len = TYPE_LENGTH (type2);
 658           ptr = (char *) alloca (count * inval2len);
 659           if (TYPE_CODE (type2) == TYPE_CODE_CHAR)
 660             {
 661               inchar = (char) unpack_long (type2,
 662                                            VALUE_CONTENTS (inval2));
 663               for (idx = 0; idx < count; idx++)
 664                 {
 665                   *(ptr + idx) = inchar;
 666                 }
 667             }
 668           else
 669             {
 670               for (idx = 0; idx < count; idx++)
 671                 {
 672                   memcpy (ptr + (idx * inval2len), VALUE_CONTENTS (inval2),
 673                           inval2len);
 674                 }
 675             }
 676           outval = value_string (ptr, count * inval2len);
 677         }
 678       else if (TYPE_CODE (type2) == TYPE_CODE_BITSTRING
 679                || TYPE_CODE (type2) == TYPE_CODE_BOOL)
 680         {
 681           error ("unimplemented support for bitstring/boolean repeats");
 682         }
 683       else
 684         {
 685           error ("can't repeat values of that type");
 686         }
 687     }
 688   else if (TYPE_CODE (type1) == TYPE_CODE_STRING
 689            || TYPE_CODE (type1) == TYPE_CODE_CHAR)
 690     {
 691       /* We have two character strings to concatenate. */
 692       if (TYPE_CODE (type2) != TYPE_CODE_STRING
 693           && TYPE_CODE (type2) != TYPE_CODE_CHAR)
 694         {
 695           error ("Strings can only be concatenated with other strings.");
 696         }
 697       inval1len = TYPE_LENGTH (type1);
 698       inval2len = TYPE_LENGTH (type2);
 699       ptr = (char *) alloca (inval1len + inval2len);
 700       if (TYPE_CODE (type1) == TYPE_CODE_CHAR)
 701         {
 702           *ptr = (char) unpack_long (type1, VALUE_CONTENTS (inval1));
 703         }
 704       else
 705         {
 706           memcpy (ptr, VALUE_CONTENTS (inval1), inval1len);
 707         }
 708       if (TYPE_CODE (type2) == TYPE_CODE_CHAR)
 709         {
 710           *(ptr + inval1len) =
 711             (char) unpack_long (type2, VALUE_CONTENTS (inval2));
 712         }
 713       else
 714         {
 715           memcpy (ptr + inval1len, VALUE_CONTENTS (inval2), inval2len);
 716         }
 717       outval = value_string (ptr, inval1len + inval2len);
 718     }
 719   else if (TYPE_CODE (type1) == TYPE_CODE_BITSTRING
 720            || TYPE_CODE (type1) == TYPE_CODE_BOOL)
 721     {
 722       /* We have two bitstrings to concatenate. */
 723       if (TYPE_CODE (type2) != TYPE_CODE_BITSTRING
 724           && TYPE_CODE (type2) != TYPE_CODE_BOOL)
 725         {
 726           error ("Bitstrings or booleans can only be concatenated with other bitstrings or booleans.");
 727         }
 728       error ("unimplemented support for bitstring/boolean concatenation.");
 729     }
 730   else
 731     {
 732       /* We don't know how to concatenate these operands. */
 733       error ("illegal operands for concatenation.");
 734     }
 735   return (outval);
 736 }
 737 \f
 738
 739
 740 /* Perform a binary operation on two operands which have reasonable
 741    representations as integers or floats.  This includes booleans,
 742    characters, integers, or floats.
 743    Does not support addition and subtraction on pointers;
 744    use value_add or value_sub if you want to handle those possibilities.  */
 745
 746 struct value *
 747 value_binop (struct value *arg1, struct value *arg2, enum exp_opcode op)
 748 {
 749   struct value *val;
 750   struct type *type1, *type2;
 751
 752   COERCE_REF (arg1);
 753   COERCE_REF (arg2);
 754   type1 = check_typedef (VALUE_TYPE (arg1));
 755   type2 = check_typedef (VALUE_TYPE (arg2));
 756
 757   if ((TYPE_CODE (type1) != TYPE_CODE_FLT && !is_integral_type (type1))
 758       ||
 759       (TYPE_CODE (type2) != TYPE_CODE_FLT && !is_integral_type (type2)))
 760     error ("Argument to arithmetic operation not a number or boolean.");
 761
 762   if (TYPE_CODE (type1) == TYPE_CODE_FLT
 763       ||
 764       TYPE_CODE (type2) == TYPE_CODE_FLT)
 765     {
 766       /* FIXME-if-picky-about-floating-accuracy: Should be doing this
 767          in target format.  real.c in GCC probably has the necessary
 768          code.  */
 769       DOUBLEST v1, v2, v = 0;
 770       v1 = value_as_double (arg1);
 771       v2 = value_as_double (arg2);
 772       switch (op)
 773         {
 774         case BINOP_ADD:
 775           v = v1 + v2;
 776           break;
 777
 778         case BINOP_SUB:
 779           v = v1 - v2;
 780           break;
 781
 782         case BINOP_MUL:
 783           v = v1 * v2;
 784           break;
 785
 786         case BINOP_DIV:
 787           v = v1 / v2;
 788           break;
 789
 790         case BINOP_EXP:
 791           v = pow (v1, v2);
 792           if (errno)
 793             error ("Cannot perform exponentiation: %s", safe_strerror (errno));
 794           break;
 795
 796         default:
 797           error ("Integer-only operation on floating point number.");
 798         }
 799
 800       /* If either arg was long double, make sure that value is also long
 801          double.  */
 802
 803       if (TYPE_LENGTH (type1) * 8 > TARGET_DOUBLE_BIT
 804           || TYPE_LENGTH (type2) * 8 > TARGET_DOUBLE_BIT)
 805         val = allocate_value (builtin_type_long_double);
 806       else
 807         val = allocate_value (builtin_type_double);
 808
 809       store_typed_floating (VALUE_CONTENTS_RAW (val), VALUE_TYPE (val), v);
 810     }
 811   else if (TYPE_CODE (type1) == TYPE_CODE_BOOL
 812            &&
 813            TYPE_CODE (type2) == TYPE_CODE_BOOL)
 814     {
 815       LONGEST v1, v2, v = 0;
 816       v1 = value_as_long (arg1);
 817       v2 = value_as_long (arg2);
 818
 819       switch (op)
 820         {
 821         case BINOP_BITWISE_AND:
 822           v = v1 & v2;
 823           break;
 824
 825         case BINOP_BITWISE_IOR:
 826           v = v1 | v2;
 827           break;
 828
 829         case BINOP_BITWISE_XOR:
 830           v = v1 ^ v2;
 831           break;
 832
 833         case BINOP_EQUAL:
 834           v = v1 == v2;
 835           break;
 836
 837         case BINOP_NOTEQUAL:
 838           v = v1 != v2;
 839           break;
 840
 841         default:
 842           error ("Invalid operation on booleans.");
 843         }
 844
 845       val = allocate_value (type1);
 846       store_signed_integer (VALUE_CONTENTS_RAW (val),
 847                             TYPE_LENGTH (type1),
 848                             v);
 849     }
 850   else
 851     /* Integral operations here.  */
 852     /* FIXME:  Also mixed integral/booleans, with result an integer. */
 853     /* FIXME: This implements ANSI C rules (also correct for C++).
 854        What about FORTRAN and (the deleted) chill ?  */
 855     {
 856       unsigned int promoted_len1 = TYPE_LENGTH (type1);
 857       unsigned int promoted_len2 = TYPE_LENGTH (type2);
 858       int is_unsigned1 = TYPE_UNSIGNED (type1);
 859       int is_unsigned2 = TYPE_UNSIGNED (type2);
 860       unsigned int result_len;
 861       int unsigned_operation;
 862
 863       /* Determine type length and signedness after promotion for
 864          both operands.  */
 865       if (promoted_len1 < TYPE_LENGTH (builtin_type_int))
 866         {
 867           is_unsigned1 = 0;
 868           promoted_len1 = TYPE_LENGTH (builtin_type_int);
 869         }
 870       if (promoted_len2 < TYPE_LENGTH (builtin_type_int))
 871         {
 872           is_unsigned2 = 0;
 873           promoted_len2 = TYPE_LENGTH (builtin_type_int);
 874         }
 875
 876       /* Determine type length of the result, and if the operation should
 877          be done unsigned.
 878          Use the signedness of the operand with the greater length.
 879          If both operands are of equal length, use unsigned operation
 880          if one of the operands is unsigned.  */
 881       if (promoted_len1 > promoted_len2)
 882         {
 883           unsigned_operation = is_unsigned1;
 884           result_len = promoted_len1;
 885         }
 886       else if (promoted_len2 > promoted_len1)
 887         {
 888           unsigned_operation = is_unsigned2;
 889           result_len = promoted_len2;
 890         }
 891       else
 892         {
 893           unsigned_operation = is_unsigned1 || is_unsigned2;
 894           result_len = promoted_len1;
 895         }
 896
 897       if (unsigned_operation)
 898         {
 899           ULONGEST v1, v2, v = 0;
 900           v1 = (ULONGEST) value_as_long (arg1);
 901           v2 = (ULONGEST) value_as_long (arg2);
 902
 903           /* Truncate values to the type length of the result.  */
 904           if (result_len < sizeof (ULONGEST))
 905             {
 906               v1 &= ((LONGEST) 1 << HOST_CHAR_BIT * result_len) - 1;
 907               v2 &= ((LONGEST) 1 << HOST_CHAR_BIT * result_len) - 1;
 908             }
 909
 910           switch (op)
 911             {
 912             case BINOP_ADD:
 913               v = v1 + v2;
 914               break;
 915
 916             case BINOP_SUB:
 917               v = v1 - v2;
 918               break;
 919
 920             case BINOP_MUL:
 921               v = v1 * v2;
 922               break;
 923
 924             case BINOP_DIV:
 925               v = v1 / v2;
 926               break;
 927
 928             case BINOP_EXP:
 929               v = pow (v1, v2);
 930               if (errno)
 931                 error ("Cannot perform exponentiation: %s", safe_strerror (errno));
 932               break;
 933
 934             case BINOP_REM:
 935               v = v1 % v2;
 936               break;
 937
 938             case BINOP_MOD:
 939               /* Knuth 1.2.4, integer only.  Note that unlike the C '%' op,
 940                  v1 mod 0 has a defined value, v1. */
 941               if (v2 == 0)
 942                 {
 943                   v = v1;
 944                 }
 945               else
 946                 {
 947                   v = v1 / v2;
 948                   /* Note floor(v1/v2) == v1/v2 for unsigned. */
 949                   v = v1 - (v2 * v);
 950                 }
 951               break;
 952
 953             case BINOP_LSH:
 954               v = v1 << v2;
 955               break;
 956
 957             case BINOP_RSH:
 958               v = v1 >> v2;
 959               break;
 960
 961             case BINOP_BITWISE_AND:
 962               v = v1 & v2;
 963               break;
 964
 965             case BINOP_BITWISE_IOR:
 966               v = v1 | v2;
 967               break;
 968
 969             case BINOP_BITWISE_XOR:
 970               v = v1 ^ v2;
 971               break;
 972
 973             case BINOP_LOGICAL_AND:
 974               v = v1 && v2;
 975               break;
 976
 977             case BINOP_LOGICAL_OR:
 978               v = v1 || v2;
 979               break;
 980
 981             case BINOP_MIN:
 982               v = v1 < v2 ? v1 : v2;
 983               break;
 984
 985             case BINOP_MAX:
 986               v = v1 > v2 ? v1 : v2;
 987               break;
 988
 989             case BINOP_EQUAL:
 990               v = v1 == v2;
 991               break;
 992
 993             case BINOP_NOTEQUAL:
 994               v = v1 != v2;
 995               break;
 996
 997             case BINOP_LESS:
 998               v = v1 < v2;
 999               break;
1000
1001             default:
1002               error ("Invalid binary operation on numbers.");
1003             }
1004
1005           /* This is a kludge to get around the fact that we don't
1006              know how to determine the result type from the types of
1007              the operands.  (I'm not really sure how much we feel the
1008              need to duplicate the exact rules of the current
1009              language.  They can get really hairy.  But not to do so
1010              makes it hard to document just what we *do* do).  */
1011
1012           /* Can't just call init_type because we wouldn't know what
1013              name to give the type.  */
1014           val = allocate_value
1015             (result_len > TARGET_LONG_BIT / HOST_CHAR_BIT
1016              ? builtin_type_unsigned_long_long
1017              : builtin_type_unsigned_long);
1018           store_unsigned_integer (VALUE_CONTENTS_RAW (val),
1019                                   TYPE_LENGTH (VALUE_TYPE (val)),
1020                                   v);
1021         }
1022       else
1023         {
1024           LONGEST v1, v2, v = 0;
1025           v1 = value_as_long (arg1);
1026           v2 = value_as_long (arg2);
1027
1028           switch (op)
1029             {
1030             case BINOP_ADD:
1031               v = v1 + v2;
1032               break;
1033
1034             case BINOP_SUB:
1035               v = v1 - v2;
1036               break;
1037
1038             case BINOP_MUL:
1039               v = v1 * v2;
1040               break;
1041
1042             case BINOP_DIV:
1043               if (v2 != 0)
1044                 v = v1 / v2;
1045               else
1046                 error ("Division by zero");
1047               break;
1048
1049             case BINOP_EXP:
1050               v = pow (v1, v2);
1051               if (errno)
1052                 error ("Cannot perform exponentiation: %s", safe_strerror (errno));
1053               break;
1054
1055             case BINOP_REM:
1056               if (v2 != 0)
1057                 v = v1 % v2;
1058               else
1059                 error ("Division by zero");
1060               break;
1061
1062             case BINOP_MOD:
1063               /* Knuth 1.2.4, integer only.  Note that unlike the C '%' op,
1064                  X mod 0 has a defined value, X. */
1065               if (v2 == 0)
1066                 {
1067                   v = v1;
1068                 }
1069               else
1070                 {
1071                   v = v1 / v2;
1072                   /* Compute floor. */
1073                   if (TRUNCATION_TOWARDS_ZERO && (v < 0) && ((v1 % v2) != 0))
1074                     {
1075                       v--;
1076                     }
1077                   v = v1 - (v2 * v);
1078                 }
1079               break;
1080
1081             case BINOP_LSH:
1082               v = v1 << v2;
1083               break;
1084
1085             case BINOP_RSH:
1086               v = v1 >> v2;
1087               break;
1088
1089             case BINOP_BITWISE_AND:
1090               v = v1 & v2;
1091               break;
1092
1093             case BINOP_BITWISE_IOR:
1094               v = v1 | v2;
1095               break;
1096
1097             case BINOP_BITWISE_XOR:
1098               v = v1 ^ v2;
1099               break;
1100
1101             case BINOP_LOGICAL_AND:
1102               v = v1 && v2;
1103               break;
1104
1105             case BINOP_LOGICAL_OR:
1106               v = v1 || v2;
1107               break;
1108
1109             case BINOP_MIN:
1110               v = v1 < v2 ? v1 : v2;
1111               break;
1112
1113             case BINOP_MAX:
1114               v = v1 > v2 ? v1 : v2;
1115               break;
1116
1117             case BINOP_EQUAL:
1118               v = v1 == v2;
1119               break;
1120
1121             case BINOP_LESS:
1122               v = v1 < v2;
1123               break;
1124
1125             default:
1126               error ("Invalid binary operation on numbers.");
1127             }
1128
1129           /* This is a kludge to get around the fact that we don't
1130              know how to determine the result type from the types of
1131              the operands.  (I'm not really sure how much we feel the
1132              need to duplicate the exact rules of the current
1133              language.  They can get really hairy.  But not to do so
1134              makes it hard to document just what we *do* do).  */
1135
1136           /* Can't just call init_type because we wouldn't know what
1137              name to give the type.  */
1138           val = allocate_value
1139             (result_len > TARGET_LONG_BIT / HOST_CHAR_BIT
1140              ? builtin_type_long_long
1141              : builtin_type_long);
1142           store_signed_integer (VALUE_CONTENTS_RAW (val),
1143                                 TYPE_LENGTH (VALUE_TYPE (val)),
1144                                 v);
1145         }
1146     }
1147
1148   return val;
1149 }
1150 \f
1151 /* Simulate the C operator ! -- return 1 if ARG1 contains zero.  */
1152
1153 int
1154 value_logical_not (struct value *arg1)
1155 {
1156   int len;
1157   char *p;
1158   struct type *type1;
1159
1160   COERCE_NUMBER (arg1);
1161   type1 = check_typedef (VALUE_TYPE (arg1));
1162
1163   if (TYPE_CODE (type1) == TYPE_CODE_FLT)
1164     return 0 == value_as_double (arg1);
1165
1166   len = TYPE_LENGTH (type1);
1167   p = VALUE_CONTENTS (arg1);
1168
1169   while (--len >= 0)
1170     {
1171       if (*p++)
1172         break;
1173     }
1174
1175   return len < 0;
1176 }
1177
1178 /* Perform a comparison on two string values (whose content are not
1179    necessarily null terminated) based on their length */
1180
1181 static int
1182 value_strcmp (struct value *arg1, struct value *arg2)
1183 {
1184   int len1 = TYPE_LENGTH (VALUE_TYPE (arg1));
1185   int len2 = TYPE_LENGTH (VALUE_TYPE (arg2));
1186   char *s1 = VALUE_CONTENTS (arg1);
1187   char *s2 = VALUE_CONTENTS (arg2);
1188   int i, len = len1 < len2 ? len1 : len2;
1189
1190   for (i = 0; i < len; i++)
1191     {
1192       if (s1[i] < s2[i])
1193         return -1;
1194       else if (s1[i] > s2[i])
1195         return 1;
1196       else
1197         continue;
1198     }
1199
1200   if (len1 < len2)
1201     return -1;
1202   else if (len1 > len2)
1203     return 1;
1204   else
1205     return 0;
1206 }
1207
1208 /* Simulate the C operator == by returning a 1
1209    iff ARG1 and ARG2 have equal contents.  */
1210
1211 int
1212 value_equal (struct value *arg1, struct value *arg2)
1213 {
1214   int len;
1215   char *p1, *p2;
1216   struct type *type1, *type2;
1217   enum type_code code1;
1218   enum type_code code2;
1219   int is_int1, is_int2;
1220
1221   COERCE_ARRAY (arg1);
1222   COERCE_ARRAY (arg2);
1223
1224   type1 = check_typedef (VALUE_TYPE (arg1));
1225   type2 = check_typedef (VALUE_TYPE (arg2));
1226   code1 = TYPE_CODE (type1);
1227   code2 = TYPE_CODE (type2);
1228   is_int1 = is_integral_type (type1);
1229   is_int2 = is_integral_type (type2);
1230
1231   if (is_int1 && is_int2)
1232     return longest_to_int (value_as_long (value_binop (arg1, arg2,
1233                                                        BINOP_EQUAL)));
1234   else if ((code1 == TYPE_CODE_FLT || is_int1)
1235            && (code2 == TYPE_CODE_FLT || is_int2))
1236     return value_as_double (arg1) == value_as_double (arg2);
1237
1238   /* FIXME: Need to promote to either CORE_ADDR or LONGEST, whichever
1239      is bigger.  */
1240   else if (code1 == TYPE_CODE_PTR && is_int2)
1241     return value_as_address (arg1) == (CORE_ADDR) value_as_long (arg2);
1242   else if (code2 == TYPE_CODE_PTR && is_int1)
1243     return (CORE_ADDR) value_as_long (arg1) == value_as_address (arg2);
1244
1245   else if (code1 == code2
1246            && ((len = (int) TYPE_LENGTH (type1))
1247                == (int) TYPE_LENGTH (type2)))
1248     {
1249       p1 = VALUE_CONTENTS (arg1);
1250       p2 = VALUE_CONTENTS (arg2);
1251       while (--len >= 0)
1252         {
1253           if (*p1++ != *p2++)
1254             break;
1255         }
1256       return len < 0;
1257     }
1258   else if (code1 == TYPE_CODE_STRING && code2 == TYPE_CODE_STRING)
1259     {
1260       return value_strcmp (arg1, arg2) == 0;
1261     }
1262   else
1263     {
1264       error ("Invalid type combination in equality test.");
1265       return 0;                 /* For lint -- never reached */
1266     }
1267 }
1268
1269 /* Simulate the C operator < by returning 1
1270    iff ARG1's contents are less than ARG2's.  */
1271
1272 int
1273 value_less (struct value *arg1, struct value *arg2)
1274 {
1275   enum type_code code1;
1276   enum type_code code2;
1277   struct type *type1, *type2;
1278   int is_int1, is_int2;
1279
1280   COERCE_ARRAY (arg1);
1281   COERCE_ARRAY (arg2);
1282
1283   type1 = check_typedef (VALUE_TYPE (arg1));
1284   type2 = check_typedef (VALUE_TYPE (arg2));
1285   code1 = TYPE_CODE (type1);
1286   code2 = TYPE_CODE (type2);
1287   is_int1 = is_integral_type (type1);
1288   is_int2 = is_integral_type (type2);
1289
1290   if (is_int1 && is_int2)
1291     return longest_to_int (value_as_long (value_binop (arg1, arg2,
1292                                                        BINOP_LESS)));
1293   else if ((code1 == TYPE_CODE_FLT || is_int1)
1294            && (code2 == TYPE_CODE_FLT || is_int2))
1295     return value_as_double (arg1) < value_as_double (arg2);
1296   else if (code1 == TYPE_CODE_PTR && code2 == TYPE_CODE_PTR)
1297     return value_as_address (arg1) < value_as_address (arg2);
1298
1299   /* FIXME: Need to promote to either CORE_ADDR or LONGEST, whichever
1300      is bigger.  */
1301   else if (code1 == TYPE_CODE_PTR && is_int2)
1302     return value_as_address (arg1) < (CORE_ADDR) value_as_long (arg2);
1303   else if (code2 == TYPE_CODE_PTR && is_int1)
1304     return (CORE_ADDR) value_as_long (arg1) < value_as_address (arg2);
1305   else if (code1 == TYPE_CODE_STRING && code2 == TYPE_CODE_STRING)
1306     return value_strcmp (arg1, arg2) < 0;
1307   else
1308     {
1309       error ("Invalid type combination in ordering comparison.");
1310       return 0;
1311     }
1312 }
1313 \f
1314 /* The unary operators - and ~.  Both free the argument ARG1.  */
1315
1316 struct value *
1317 value_neg (struct value *arg1)
1318 {
1319   struct type *type;
1320   struct type *result_type = VALUE_TYPE (arg1);
1321
1322   COERCE_REF (arg1);
1323
1324   type = check_typedef (VALUE_TYPE (arg1));
1325
1326   if (TYPE_CODE (type) == TYPE_CODE_FLT)
1327     return value_from_double (result_type, -value_as_double (arg1));
1328   else if (is_integral_type (type))
1329     {
1330       /* Perform integral promotion for ANSI C/C++.  FIXME: What about
1331          FORTRAN and (the deleted) chill ?  */
1332       if (TYPE_LENGTH (type) < TYPE_LENGTH (builtin_type_int))
1333         result_type = builtin_type_int;
1334
1335       return value_from_longest (result_type, -value_as_long (arg1));
1336     }
1337   else
1338     {
1339       error ("Argument to negate operation not a number.");
1340       return 0;                 /* For lint -- never reached */
1341     }
1342 }
1343
1344 struct value *
1345 value_complement (struct value *arg1)
1346 {
1347   struct type *type;
1348   struct type *result_type = VALUE_TYPE (arg1);
1349
1350   COERCE_REF (arg1);
1351
1352   type = check_typedef (VALUE_TYPE (arg1));
1353
1354   if (!is_integral_type (type))
1355     error ("Argument to complement operation not an integer or boolean.");
1356
1357   /* Perform integral promotion for ANSI C/C++.
1358      FIXME: What about FORTRAN ?  */
1359   if (TYPE_LENGTH (type) < TYPE_LENGTH (builtin_type_int))
1360     result_type = builtin_type_int;
1361
1362   return value_from_longest (result_type, ~value_as_long (arg1));
1363 }
1364 \f
1365 /* The INDEX'th bit of SET value whose VALUE_TYPE is TYPE,
1366    and whose VALUE_CONTENTS is valaddr.
1367    Return -1 if out of range, -2 other error. */
1368
1369 int
1370 value_bit_index (struct type *type, char *valaddr, int index)
1371 {
1372   LONGEST low_bound, high_bound;
1373   LONGEST word;
1374   unsigned rel_index;
1375   struct type *range = TYPE_FIELD_TYPE (type, 0);
1376   if (get_discrete_bounds (range, &low_bound, &high_bound) < 0)
1377     return -2;
1378   if (index < low_bound || index > high_bound)
1379     return -1;
1380   rel_index = index - low_bound;
1381   word = unpack_long (builtin_type_unsigned_char,
1382                       valaddr + (rel_index / TARGET_CHAR_BIT));
1383   rel_index %= TARGET_CHAR_BIT;
1384   if (BITS_BIG_ENDIAN)
1385     rel_index = TARGET_CHAR_BIT - 1 - rel_index;
1386   return (word >> rel_index) & 1;
1387 }
1388
1389 struct value *
1390 value_in (struct value *element, struct value *set)
1391 {
1392   int member;
1393   struct type *settype = check_typedef (VALUE_TYPE (set));
1394   struct type *eltype = check_typedef (VALUE_TYPE (element));
1395   if (TYPE_CODE (eltype) == TYPE_CODE_RANGE)
1396     eltype = TYPE_TARGET_TYPE (eltype);
1397   if (TYPE_CODE (settype) != TYPE_CODE_SET)
1398     error ("Second argument of 'IN' has wrong type");
1399   if (TYPE_CODE (eltype) != TYPE_CODE_INT
1400       && TYPE_CODE (eltype) != TYPE_CODE_CHAR
1401       && TYPE_CODE (eltype) != TYPE_CODE_ENUM
1402       && TYPE_CODE (eltype) != TYPE_CODE_BOOL)
1403     error ("First argument of 'IN' has wrong type");
1404   member = value_bit_index (settype, VALUE_CONTENTS (set),
1405                             value_as_long (element));
1406   if (member < 0)
1407     error ("First argument of 'IN' not in range");
1408   return value_from_longest (LA_BOOL_TYPE, member);
1409 }
1410
1411 void
1412 _initialize_valarith (void)
1413 {
1414 }