2005-02-03 Andrew Cagney <cagney@gnu.org>

[deliverable/binutils-gdb.git] / gdb / value.c

/* Low level packing and unpacking of values for GDB, the GNU Debugger.

   Copyright 1986, 1987, 1988, 1989, 1990, 1991, 1992, 1993, 1994,
   1995, 1996, 1997, 1998, 1999, 2000, 2002, 2003, 2004, 2005 Free
   Software Foundation, Inc.

   This file is part of GDB.

   This program is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
   the Free Software Foundation; either version 2 of the License, or
   (at your option) any later version.

   This program is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   GNU General Public License for more details.

   You should have received a copy of the GNU General Public License
   along with this program; if not, write to the Free Software
   Foundation, Inc., 59 Temple Place - Suite 330,
   Boston, MA 02111-1307, USA.  */

#include "defs.h"
#include "gdb_string.h"
#include "symtab.h"
#include "gdbtypes.h"
#include "value.h"
#include "gdbcore.h"
#include "command.h"
#include "gdbcmd.h"
#include "target.h"
#include "language.h"
#include "scm-lang.h"
#include "demangle.h"
#include "doublest.h"
#include "gdb_assert.h"
#include "regcache.h"
#include "block.h"

/* Prototypes for exported functions. */

void _initialize_values (void);

/* Prototypes for local functions. */

static void show_values (char *, int);

static void show_convenience (char *, int);


/* The value-history records all the values printed
   by print commands during this session.  Each chunk
   records 60 consecutive values.  The first chunk on
   the chain records the most recent values.
   The total number of values is in value_history_count.  */

#define VALUE_HISTORY_CHUNK 60

struct value_history_chunk
  {
    struct value_history_chunk *next;
    struct value *values[VALUE_HISTORY_CHUNK];
  };

/* Chain of chunks now in use.  */

static struct value_history_chunk *value_history_chain;

static int value_history_count;	/* Abs number of last entry stored */
\f
/* List of all value objects currently allocated
   (except for those released by calls to release_value)
   This is so they can be freed after each command.  */

static struct value *all_values;

/* Allocate a  value  that has the correct length for type TYPE.  */

struct value *
allocate_value (struct type *type)
{
  struct value *val;
  struct type *atype = check_typedef (type);

  val = (struct value *) xzalloc (sizeof (struct value) + TYPE_LENGTH (atype));
  val->next = all_values;
  all_values = val;
  val->type = type;
  val->enclosing_type = type;
  VALUE_LVAL (val) = not_lval;
  VALUE_ADDRESS (val) = 0;
  VALUE_FRAME_ID (val) = null_frame_id;
  val->offset = 0;
  val->bitpos = 0;
  val->bitsize = 0;
  VALUE_REGNUM (val) = -1;
  val->lazy = 0;
  VALUE_OPTIMIZED_OUT (val) = 0;
  VALUE_EMBEDDED_OFFSET (val) = 0;
  VALUE_POINTED_TO_OFFSET (val) = 0;
  val->modifiable = 1;
  return val;
}

/* Allocate a  value  that has the correct length
   for COUNT repetitions type TYPE.  */

struct value *
allocate_repeat_value (struct type *type, int count)
{
  int low_bound = current_language->string_lower_bound;		/* ??? */
  /* FIXME-type-allocation: need a way to free this type when we are
     done with it.  */
  struct type *range_type
  = create_range_type ((struct type *) NULL, builtin_type_int,
		       low_bound, count + low_bound - 1);
  /* FIXME-type-allocation: need a way to free this type when we are
     done with it.  */
  return allocate_value (create_array_type ((struct type *) NULL,
					    type, range_type));
}

/* Accessor methods.  */

struct type *
value_type (struct value *value)
{
  return value->type;
}

int
value_offset (struct value *value)
{
  return value->offset;
}

int
value_bitpos (struct value *value)
{
  return value->bitpos;
}

int
value_bitsize (struct value *value)
{
  return value->bitsize;
}

bfd_byte *
value_contents_raw (struct value *value)
{
  return value->aligner.contents + value->embedded_offset;
}

bfd_byte *
value_contents_all_raw (struct value *value)
{
  return value->aligner.contents;
}

struct type *
value_enclosing_type (struct value *value)
{
  return value->enclosing_type;
}

const bfd_byte *
value_contents_all (struct value *value)
{
  if (value->lazy)
    value_fetch_lazy (value);
  return value->aligner.contents;
}

int
value_lazy (struct value *value)
{
  return value->lazy;
}

\f
/* Return a mark in the value chain.  All values allocated after the
   mark is obtained (except for those released) are subject to being freed
   if a subsequent value_free_to_mark is passed the mark.  */
struct value *
value_mark (void)
{
  return all_values;
}

/* Free all values allocated since MARK was obtained by value_mark
   (except for those released).  */
void
value_free_to_mark (struct value *mark)
{
  struct value *val;
  struct value *next;

  for (val = all_values; val && val != mark; val = next)
    {
      next = val->next;
      value_free (val);
    }
  all_values = val;
}

/* Free all the values that have been allocated (except for those released).
   Called after each command, successful or not.  */

void
free_all_values (void)
{
  struct value *val;
  struct value *next;

  for (val = all_values; val; val = next)
    {
      next = val->next;
      value_free (val);
    }

  all_values = 0;
}

/* Remove VAL from the chain all_values
   so it will not be freed automatically.  */

void
release_value (struct value *val)
{
  struct value *v;

  if (all_values == val)
    {
      all_values = val->next;
      return;
    }

  for (v = all_values; v; v = v->next)
    {
      if (v->next == val)
	{
	  v->next = val->next;
	  break;
	}
    }
}

/* Release all values up to mark  */
struct value *
value_release_to_mark (struct value *mark)
{
  struct value *val;
  struct value *next;

  for (val = next = all_values; next; next = next->next)
    if (next->next == mark)
      {
	all_values = next->next;
	next->next = NULL;
	return val;
      }
  all_values = 0;
  return val;
}

/* Return a copy of the value ARG.
   It contains the same contents, for same memory address,
   but it's a different block of storage.  */

struct value *
value_copy (struct value *arg)
{
  struct type *encl_type = value_enclosing_type (arg);
  struct value *val = allocate_value (encl_type);
  val->type = arg->type;
  VALUE_LVAL (val) = VALUE_LVAL (arg);
  VALUE_ADDRESS (val) = VALUE_ADDRESS (arg);
  val->offset = arg->offset;
  val->bitpos = arg->bitpos;
  val->bitsize = arg->bitsize;
  VALUE_FRAME_ID (val) = VALUE_FRAME_ID (arg);
  VALUE_REGNUM (val) = VALUE_REGNUM (arg);
  val->lazy = arg->lazy;
  VALUE_OPTIMIZED_OUT (val) = VALUE_OPTIMIZED_OUT (arg);
  VALUE_EMBEDDED_OFFSET (val) = VALUE_EMBEDDED_OFFSET (arg);
  VALUE_POINTED_TO_OFFSET (val) = VALUE_POINTED_TO_OFFSET (arg);
  val->modifiable = arg->modifiable;
  if (!value_lazy (val))
    {
      memcpy (value_contents_all_raw (val), value_contents_all_raw (arg),
	      TYPE_LENGTH (value_enclosing_type (arg)));

    }
  return val;
}
\f
/* Access to the value history.  */

/* Record a new value in the value history.
   Returns the absolute history index of the entry.
   Result of -1 indicates the value was not saved; otherwise it is the
   value history index of this new item.  */

int
record_latest_value (struct value *val)
{
  int i;

  /* We don't want this value to have anything to do with the inferior anymore.
     In particular, "set $1 = 50" should not affect the variable from which
     the value was taken, and fast watchpoints should be able to assume that
     a value on the value history never changes.  */
  if (value_lazy (val))
    value_fetch_lazy (val);
  /* We preserve VALUE_LVAL so that the user can find out where it was fetched
     from.  This is a bit dubious, because then *&$1 does not just return $1
     but the current contents of that location.  c'est la vie...  */
  val->modifiable = 0;
  release_value (val);

  /* Here we treat value_history_count as origin-zero
     and applying to the value being stored now.  */

  i = value_history_count % VALUE_HISTORY_CHUNK;
  if (i == 0)
    {
      struct value_history_chunk *new
      = (struct value_history_chunk *)
      xmalloc (sizeof (struct value_history_chunk));
      memset (new->values, 0, sizeof new->values);
      new->next = value_history_chain;
      value_history_chain = new;
    }

  value_history_chain->values[i] = val;

  /* Now we regard value_history_count as origin-one
     and applying to the value just stored.  */

  return ++value_history_count;
}

/* Return a copy of the value in the history with sequence number NUM.  */

struct value *
access_value_history (int num)
{
  struct value_history_chunk *chunk;
  int i;
  int absnum = num;

  if (absnum <= 0)
    absnum += value_history_count;

  if (absnum <= 0)
    {
      if (num == 0)
	error ("The history is empty.");
      else if (num == 1)
	error ("There is only one value in the history.");
      else
	error ("History does not go back to $$%d.", -num);
    }
  if (absnum > value_history_count)
    error ("History has not yet reached $%d.", absnum);

  absnum--;

  /* Now absnum is always absolute and origin zero.  */

  chunk = value_history_chain;
  for (i = (value_history_count - 1) / VALUE_HISTORY_CHUNK - absnum / VALUE_HISTORY_CHUNK;
       i > 0; i--)
    chunk = chunk->next;

  return value_copy (chunk->values[absnum % VALUE_HISTORY_CHUNK]);
}

/* Clear the value history entirely.
   Must be done when new symbol tables are loaded,
   because the type pointers become invalid.  */

void
clear_value_history (void)
{
  struct value_history_chunk *next;
  int i;
  struct value *val;

  while (value_history_chain)
    {
      for (i = 0; i < VALUE_HISTORY_CHUNK; i++)
	if ((val = value_history_chain->values[i]) != NULL)
	  xfree (val);
      next = value_history_chain->next;
      xfree (value_history_chain);
      value_history_chain = next;
    }
  value_history_count = 0;
}

static void
show_values (char *num_exp, int from_tty)
{
  int i;
  struct value *val;
  static int num = 1;

  if (num_exp)
    {
      /* "info history +" should print from the stored position.
         "info history <exp>" should print around value number <exp>.  */
      if (num_exp[0] != '+' || num_exp[1] != '\0')
	num = parse_and_eval_long (num_exp) - 5;
    }
  else
    {
      /* "info history" means print the last 10 values.  */
      num = value_history_count - 9;
    }

  if (num <= 0)
    num = 1;

  for (i = num; i < num + 10 && i <= value_history_count; i++)
    {
      val = access_value_history (i);
      printf_filtered ("$%d = ", i);
      value_print (val, gdb_stdout, 0, Val_pretty_default);
      printf_filtered ("\n");
    }

  /* The next "info history +" should start after what we just printed.  */
  num += 10;

  /* Hitting just return after this command should do the same thing as
     "info history +".  If num_exp is null, this is unnecessary, since
     "info history +" is not useful after "info history".  */
  if (from_tty && num_exp)
    {
      num_exp[0] = '+';
      num_exp[1] = '\0';
    }
}
\f
/* Internal variables.  These are variables within the debugger
   that hold values assigned by debugger commands.
   The user refers to them with a '$' prefix
   that does not appear in the variable names stored internally.  */

static struct internalvar *internalvars;

/* Look up an internal variable with name NAME.  NAME should not
   normally include a dollar sign.

   If the specified internal variable does not exist,
   one is created, with a void value.  */

struct internalvar *
lookup_internalvar (char *name)
{
  struct internalvar *var;

  for (var = internalvars; var; var = var->next)
    if (strcmp (var->name, name) == 0)
      return var;

  var = (struct internalvar *) xmalloc (sizeof (struct internalvar));
  var->name = concat (name, NULL);
  var->value = allocate_value (builtin_type_void);
  release_value (var->value);
  var->next = internalvars;
  internalvars = var;
  return var;
}

struct value *
value_of_internalvar (struct internalvar *var)
{
  struct value *val;

  val = value_copy (var->value);
  if (value_lazy (val))
    value_fetch_lazy (val);
  VALUE_LVAL (val) = lval_internalvar;
  VALUE_INTERNALVAR (val) = var;
  return val;
}

void
set_internalvar_component (struct internalvar *var, int offset, int bitpos,
			   int bitsize, struct value *newval)
{
  char *addr = VALUE_CONTENTS (var->value) + offset;

  if (bitsize)
    modify_field (addr, value_as_long (newval),
		  bitpos, bitsize);
  else
    memcpy (addr, VALUE_CONTENTS (newval), TYPE_LENGTH (value_type (newval)));
}

void
set_internalvar (struct internalvar *var, struct value *val)
{
  struct value *newval;

  newval = value_copy (val);
  newval->modifiable = 1;

  /* Force the value to be fetched from the target now, to avoid problems
     later when this internalvar is referenced and the target is gone or
     has changed.  */
  if (value_lazy (newval))
    value_fetch_lazy (newval);

  /* Begin code which must not call error().  If var->value points to
     something free'd, an error() obviously leaves a dangling pointer.
     But we also get a danling pointer if var->value points to
     something in the value chain (i.e., before release_value is
     called), because after the error free_all_values will get called before
     long.  */
  xfree (var->value);
  var->value = newval;
  release_value (newval);
  /* End code which must not call error().  */
}

char *
internalvar_name (struct internalvar *var)
{
  return var->name;
}

/* Free all internalvars.  Done when new symtabs are loaded,
   because that makes the values invalid.  */

void
clear_internalvars (void)
{
  struct internalvar *var;

  while (internalvars)
    {
      var = internalvars;
      internalvars = var->next;
      xfree (var->name);
      xfree (var->value);
      xfree (var);
    }
}

static void
show_convenience (char *ignore, int from_tty)
{
  struct internalvar *var;
  int varseen = 0;

  for (var = internalvars; var; var = var->next)
    {
      if (!varseen)
	{
	  varseen = 1;
	}
      printf_filtered ("$%s = ", var->name);
      value_print (var->value, gdb_stdout, 0, Val_pretty_default);
      printf_filtered ("\n");
    }
  if (!varseen)
    printf_unfiltered ("No debugger convenience variables now defined.\n\
Convenience variables have names starting with \"$\";\n\
use \"set\" as in \"set $foo = 5\" to define them.\n");
}
\f
/* Extract a value as a C number (either long or double).
   Knows how to convert fixed values to double, or
   floating values to long.
   Does not deallocate the value.  */

LONGEST
value_as_long (struct value *val)
{
  /* This coerces arrays and functions, which is necessary (e.g.
     in disassemble_command).  It also dereferences references, which
     I suspect is the most logical thing to do.  */
  val = coerce_array (val);
  return unpack_long (value_type (val), VALUE_CONTENTS (val));
}

DOUBLEST
value_as_double (struct value *val)
{
  DOUBLEST foo;
  int inv;

  foo = unpack_double (value_type (val), VALUE_CONTENTS (val), &inv);
  if (inv)
    error ("Invalid floating value found in program.");
  return foo;
}
/* Extract a value as a C pointer. Does not deallocate the value.  
   Note that val's type may not actually be a pointer; value_as_long
   handles all the cases.  */
CORE_ADDR
value_as_address (struct value *val)
{
  /* Assume a CORE_ADDR can fit in a LONGEST (for now).  Not sure
     whether we want this to be true eventually.  */
#if 0
  /* ADDR_BITS_REMOVE is wrong if we are being called for a
     non-address (e.g. argument to "signal", "info break", etc.), or
     for pointers to char, in which the low bits *are* significant.  */
  return ADDR_BITS_REMOVE (value_as_long (val));
#else

  /* There are several targets (IA-64, PowerPC, and others) which
     don't represent pointers to functions as simply the address of
     the function's entry point.  For example, on the IA-64, a
     function pointer points to a two-word descriptor, generated by
     the linker, which contains the function's entry point, and the
     value the IA-64 "global pointer" register should have --- to
     support position-independent code.  The linker generates
     descriptors only for those functions whose addresses are taken.

     On such targets, it's difficult for GDB to convert an arbitrary
     function address into a function pointer; it has to either find
     an existing descriptor for that function, or call malloc and
     build its own.  On some targets, it is impossible for GDB to
     build a descriptor at all: the descriptor must contain a jump
     instruction; data memory cannot be executed; and code memory
     cannot be modified.

     Upon entry to this function, if VAL is a value of type `function'
     (that is, TYPE_CODE (VALUE_TYPE (val)) == TYPE_CODE_FUNC), then
     VALUE_ADDRESS (val) is the address of the function.  This is what
     you'll get if you evaluate an expression like `main'.  The call
     to COERCE_ARRAY below actually does all the usual unary
     conversions, which includes converting values of type `function'
     to `pointer to function'.  This is the challenging conversion
     discussed above.  Then, `unpack_long' will convert that pointer
     back into an address.

     So, suppose the user types `disassemble foo' on an architecture
     with a strange function pointer representation, on which GDB
     cannot build its own descriptors, and suppose further that `foo'
     has no linker-built descriptor.  The address->pointer conversion
     will signal an error and prevent the command from running, even
     though the next step would have been to convert the pointer
     directly back into the same address.

     The following shortcut avoids this whole mess.  If VAL is a
     function, just return its address directly.  */
  if (TYPE_CODE (value_type (val)) == TYPE_CODE_FUNC
      || TYPE_CODE (value_type (val)) == TYPE_CODE_METHOD)
    return VALUE_ADDRESS (val);

  val = coerce_array (val);

  /* Some architectures (e.g. Harvard), map instruction and data
     addresses onto a single large unified address space.  For
     instance: An architecture may consider a large integer in the
     range 0x10000000 .. 0x1000ffff to already represent a data
     addresses (hence not need a pointer to address conversion) while
     a small integer would still need to be converted integer to
     pointer to address.  Just assume such architectures handle all
     integer conversions in a single function.  */

  /* JimB writes:

     I think INTEGER_TO_ADDRESS is a good idea as proposed --- but we
     must admonish GDB hackers to make sure its behavior matches the
     compiler's, whenever possible.

     In general, I think GDB should evaluate expressions the same way
     the compiler does.  When the user copies an expression out of
     their source code and hands it to a `print' command, they should
     get the same value the compiler would have computed.  Any
     deviation from this rule can cause major confusion and annoyance,
     and needs to be justified carefully.  In other words, GDB doesn't
     really have the freedom to do these conversions in clever and
     useful ways.

     AndrewC pointed out that users aren't complaining about how GDB
     casts integers to pointers; they are complaining that they can't
     take an address from a disassembly listing and give it to `x/i'.
     This is certainly important.

     Adding an architecture method like integer_to_address() certainly
     makes it possible for GDB to "get it right" in all circumstances
     --- the target has complete control over how things get done, so
     people can Do The Right Thing for their target without breaking
     anyone else.  The standard doesn't specify how integers get
     converted to pointers; usually, the ABI doesn't either, but
     ABI-specific code is a more reasonable place to handle it.  */

  if (TYPE_CODE (value_type (val)) != TYPE_CODE_PTR
      && TYPE_CODE (value_type (val)) != TYPE_CODE_REF
      && gdbarch_integer_to_address_p (current_gdbarch))
    return gdbarch_integer_to_address (current_gdbarch, value_type (val),
				       VALUE_CONTENTS (val));

  return unpack_long (value_type (val), VALUE_CONTENTS (val));
#endif
}
\f
/* Unpack raw data (copied from debugee, target byte order) at VALADDR
   as a long, or as a double, assuming the raw data is described
   by type TYPE.  Knows how to convert different sizes of values
   and can convert between fixed and floating point.  We don't assume
   any alignment for the raw data.  Return value is in host byte order.

   If you want functions and arrays to be coerced to pointers, and
   references to be dereferenced, call value_as_long() instead.

   C++: It is assumed that the front-end has taken care of
   all matters concerning pointers to members.  A pointer
   to member which reaches here is considered to be equivalent
   to an INT (or some size).  After all, it is only an offset.  */

LONGEST
unpack_long (struct type *type, const char *valaddr)
{
  enum type_code code = TYPE_CODE (type);
  int len = TYPE_LENGTH (type);
  int nosign = TYPE_UNSIGNED (type);

  if (current_language->la_language == language_scm
      && is_scmvalue_type (type))
    return scm_unpack (type, valaddr, TYPE_CODE_INT);

  switch (code)
    {
    case TYPE_CODE_TYPEDEF:
      return unpack_long (check_typedef (type), valaddr);
    case TYPE_CODE_ENUM:
    case TYPE_CODE_BOOL:
    case TYPE_CODE_INT:
    case TYPE_CODE_CHAR:
    case TYPE_CODE_RANGE:
      if (nosign)
	return extract_unsigned_integer (valaddr, len);
      else
	return extract_signed_integer (valaddr, len);

    case TYPE_CODE_FLT:
      return extract_typed_floating (valaddr, type);

    case TYPE_CODE_PTR:
    case TYPE_CODE_REF:
      /* Assume a CORE_ADDR can fit in a LONGEST (for now).  Not sure
         whether we want this to be true eventually.  */
      return extract_typed_address (valaddr, type);

    case TYPE_CODE_MEMBER:
      error ("not implemented: member types in unpack_long");

    default:
      error ("Value can't be converted to integer.");
    }
  return 0;			/* Placate lint.  */
}

/* Return a double value from the specified type and address.
   INVP points to an int which is set to 0 for valid value,
   1 for invalid value (bad float format).  In either case,
   the returned double is OK to use.  Argument is in target
   format, result is in host format.  */

DOUBLEST
unpack_double (struct type *type, const char *valaddr, int *invp)
{
  enum type_code code;
  int len;
  int nosign;

  *invp = 0;			/* Assume valid.   */
  CHECK_TYPEDEF (type);
  code = TYPE_CODE (type);
  len = TYPE_LENGTH (type);
  nosign = TYPE_UNSIGNED (type);
  if (code == TYPE_CODE_FLT)
    {
      /* NOTE: cagney/2002-02-19: There was a test here to see if the
	 floating-point value was valid (using the macro
	 INVALID_FLOAT).  That test/macro have been removed.

	 It turns out that only the VAX defined this macro and then
	 only in a non-portable way.  Fixing the portability problem
	 wouldn't help since the VAX floating-point code is also badly
	 bit-rotten.  The target needs to add definitions for the
	 methods TARGET_FLOAT_FORMAT and TARGET_DOUBLE_FORMAT - these
	 exactly describe the target floating-point format.  The
	 problem here is that the corresponding floatformat_vax_f and
	 floatformat_vax_d values these methods should be set to are
	 also not defined either.  Oops!

         Hopefully someone will add both the missing floatformat
         definitions and the new cases for floatformat_is_valid ().  */

      if (!floatformat_is_valid (floatformat_from_type (type), valaddr))
	{
	  *invp = 1;
	  return 0.0;
	}

      return extract_typed_floating (valaddr, type);
    }
  else if (nosign)
    {
      /* Unsigned -- be sure we compensate for signed LONGEST.  */
      return (ULONGEST) unpack_long (type, valaddr);
    }
  else
    {
      /* Signed -- we are OK with unpack_long.  */
      return unpack_long (type, valaddr);
    }
}

/* Unpack raw data (copied from debugee, target byte order) at VALADDR
   as a CORE_ADDR, assuming the raw data is described by type TYPE.
   We don't assume any alignment for the raw data.  Return value is in
   host byte order.

   If you want functions and arrays to be coerced to pointers, and
   references to be dereferenced, call value_as_address() instead.

   C++: It is assumed that the front-end has taken care of
   all matters concerning pointers to members.  A pointer
   to member which reaches here is considered to be equivalent
   to an INT (or some size).  After all, it is only an offset.  */

CORE_ADDR
unpack_pointer (struct type *type, const char *valaddr)
{
  /* Assume a CORE_ADDR can fit in a LONGEST (for now).  Not sure
     whether we want this to be true eventually.  */
  return unpack_long (type, valaddr);
}

\f
/* Get the value of the FIELDN'th field (which must be static) of
   TYPE.  Return NULL if the field doesn't exist or has been
   optimized out. */

struct value *
value_static_field (struct type *type, int fieldno)
{
  struct value *retval;

  if (TYPE_FIELD_STATIC_HAS_ADDR (type, fieldno))
    {
      retval = value_at (TYPE_FIELD_TYPE (type, fieldno),
			 TYPE_FIELD_STATIC_PHYSADDR (type, fieldno));
    }
  else
    {
      char *phys_name = TYPE_FIELD_STATIC_PHYSNAME (type, fieldno);
      struct symbol *sym = lookup_symbol (phys_name, 0, VAR_DOMAIN, 0, NULL);
      if (sym == NULL)
	{
	  /* With some compilers, e.g. HP aCC, static data members are reported
	     as non-debuggable symbols */
	  struct minimal_symbol *msym = lookup_minimal_symbol (phys_name, NULL, NULL);
	  if (!msym)
	    return NULL;
	  else
	    {
	      retval = value_at (TYPE_FIELD_TYPE (type, fieldno),
				 SYMBOL_VALUE_ADDRESS (msym));
	    }
	}
      else
	{
	  /* SYM should never have a SYMBOL_CLASS which will require
	     read_var_value to use the FRAME parameter.  */
	  if (symbol_read_needs_frame (sym))
	    warning ("static field's value depends on the current "
		     "frame - bad debug info?");
	  retval = read_var_value (sym, NULL);
 	}
      if (retval && VALUE_LVAL (retval) == lval_memory)
	SET_FIELD_PHYSADDR (TYPE_FIELD (type, fieldno),
			    VALUE_ADDRESS (retval));
    }
  return retval;
}

/* Change the enclosing type of a value object VAL to NEW_ENCL_TYPE.  
   You have to be careful here, since the size of the data area for the value 
   is set by the length of the enclosing type.  So if NEW_ENCL_TYPE is bigger 
   than the old enclosing type, you have to allocate more space for the data.  
   The return value is a pointer to the new version of this value structure. */

struct value *
value_change_enclosing_type (struct value *val, struct type *new_encl_type)
{
  if (TYPE_LENGTH (new_encl_type) <= TYPE_LENGTH (value_enclosing_type (val))) 
    {
      val->enclosing_type = new_encl_type;
      return val;
    }
  else
    {
      struct value *new_val;
      struct value *prev;
      
      new_val = (struct value *) xrealloc (val, sizeof (struct value) + TYPE_LENGTH (new_encl_type));

      new_val->enclosing_type = new_encl_type;
 
      /* We have to make sure this ends up in the same place in the value
	 chain as the original copy, so it's clean-up behavior is the same. 
	 If the value has been released, this is a waste of time, but there
	 is no way to tell that in advance, so... */
      
      if (val != all_values) 
	{
	  for (prev = all_values; prev != NULL; prev = prev->next)
	    {
	      if (prev->next == val) 
		{
		  prev->next = new_val;
		  break;
		}
	    }
	}
      
      return new_val;
    }
}

/* Given a value ARG1 (offset by OFFSET bytes)
   of a struct or union type ARG_TYPE,
   extract and return the value of one of its (non-static) fields.
   FIELDNO says which field. */

struct value *
value_primitive_field (struct value *arg1, int offset,
		       int fieldno, struct type *arg_type)
{
  struct value *v;
  struct type *type;

  CHECK_TYPEDEF (arg_type);
  type = TYPE_FIELD_TYPE (arg_type, fieldno);

  /* Handle packed fields */

  if (TYPE_FIELD_BITSIZE (arg_type, fieldno))
    {
      v = value_from_longest (type,
			      unpack_field_as_long (arg_type,
						    VALUE_CONTENTS (arg1)
						    + offset,
						    fieldno));
      v->bitpos = TYPE_FIELD_BITPOS (arg_type, fieldno) % 8;
      v->bitsize = TYPE_FIELD_BITSIZE (arg_type, fieldno);
      v->offset = value_offset (arg1) + offset
	+ TYPE_FIELD_BITPOS (arg_type, fieldno) / 8;
    }
  else if (fieldno < TYPE_N_BASECLASSES (arg_type))
    {
      /* This field is actually a base subobject, so preserve the
         entire object's contents for later references to virtual
         bases, etc.  */
      v = allocate_value (value_enclosing_type (arg1));
      v->type = type;
      if (value_lazy (arg1))
	VALUE_LAZY (v) = 1;
      else
	memcpy (value_contents_all_raw (v), value_contents_all_raw (arg1),
		TYPE_LENGTH (value_enclosing_type (arg1)));
      v->offset = value_offset (arg1);
      VALUE_EMBEDDED_OFFSET (v)
	= offset +
	VALUE_EMBEDDED_OFFSET (arg1) +
	TYPE_FIELD_BITPOS (arg_type, fieldno) / 8;
    }
  else
    {
      /* Plain old data member */
      offset += TYPE_FIELD_BITPOS (arg_type, fieldno) / 8;
      v = allocate_value (type);
      if (value_lazy (arg1))
	VALUE_LAZY (v) = 1;
      else
	memcpy (value_contents_raw (v),
		value_contents_raw (arg1) + offset,
		TYPE_LENGTH (type));
      v->offset = (value_offset (arg1) + offset
		   + VALUE_EMBEDDED_OFFSET (arg1));
    }
  VALUE_LVAL (v) = VALUE_LVAL (arg1);
  if (VALUE_LVAL (arg1) == lval_internalvar)
    VALUE_LVAL (v) = lval_internalvar_component;
  VALUE_ADDRESS (v) = VALUE_ADDRESS (arg1);
  VALUE_REGNUM (v) = VALUE_REGNUM (arg1);
  VALUE_FRAME_ID (v) = VALUE_FRAME_ID (arg1);
/*  VALUE_OFFSET (v) = VALUE_OFFSET (arg1) + offset
   + TYPE_FIELD_BITPOS (arg_type, fieldno) / 8; */
  return v;
}

/* Given a value ARG1 of a struct or union type,
   extract and return the value of one of its (non-static) fields.
   FIELDNO says which field. */

struct value *
value_field (struct value *arg1, int fieldno)
{
  return value_primitive_field (arg1, 0, fieldno, value_type (arg1));
}

/* Return a non-virtual function as a value.
   F is the list of member functions which contains the desired method.
   J is an index into F which provides the desired method.

   We only use the symbol for its address, so be happy with either a
   full symbol or a minimal symbol.
 */

struct value *
value_fn_field (struct value **arg1p, struct fn_field *f, int j, struct type *type,
		int offset)
{
  struct value *v;
  struct type *ftype = TYPE_FN_FIELD_TYPE (f, j);
  char *physname = TYPE_FN_FIELD_PHYSNAME (f, j);
  struct symbol *sym;
  struct minimal_symbol *msym;

  sym = lookup_symbol (physname, 0, VAR_DOMAIN, 0, NULL);
  if (sym != NULL)
    {
      msym = NULL;
    }
  else
    {
      gdb_assert (sym == NULL);
      msym = lookup_minimal_symbol (physname, NULL, NULL);
      if (msym == NULL)
	return NULL;
    }

  v = allocate_value (ftype);
  if (sym)
    {
      VALUE_ADDRESS (v) = BLOCK_START (SYMBOL_BLOCK_VALUE (sym));
    }
  else
    {
      VALUE_ADDRESS (v) = SYMBOL_VALUE_ADDRESS (msym);
    }

  if (arg1p)
    {
      if (type != value_type (*arg1p))
	*arg1p = value_ind (value_cast (lookup_pointer_type (type),
					value_addr (*arg1p)));

      /* Move the `this' pointer according to the offset.
         VALUE_OFFSET (*arg1p) += offset;
       */
    }

  return v;
}

\f
/* Unpack a field FIELDNO of the specified TYPE, from the anonymous object at
   VALADDR.

   Extracting bits depends on endianness of the machine.  Compute the
   number of least significant bits to discard.  For big endian machines,
   we compute the total number of bits in the anonymous object, subtract
   off the bit count from the MSB of the object to the MSB of the
   bitfield, then the size of the bitfield, which leaves the LSB discard
   count.  For little endian machines, the discard count is simply the
   number of bits from the LSB of the anonymous object to the LSB of the
   bitfield.

   If the field is signed, we also do sign extension. */

LONGEST
unpack_field_as_long (struct type *type, const char *valaddr, int fieldno)
{
  ULONGEST val;
  ULONGEST valmask;
  int bitpos = TYPE_FIELD_BITPOS (type, fieldno);
  int bitsize = TYPE_FIELD_BITSIZE (type, fieldno);
  int lsbcount;
  struct type *field_type;

  val = extract_unsigned_integer (valaddr + bitpos / 8, sizeof (val));
  field_type = TYPE_FIELD_TYPE (type, fieldno);
  CHECK_TYPEDEF (field_type);

  /* Extract bits.  See comment above. */

  if (BITS_BIG_ENDIAN)
    lsbcount = (sizeof val * 8 - bitpos % 8 - bitsize);
  else
    lsbcount = (bitpos % 8);
  val >>= lsbcount;

  /* If the field does not entirely fill a LONGEST, then zero the sign bits.
     If the field is signed, and is negative, then sign extend. */

  if ((bitsize > 0) && (bitsize < 8 * (int) sizeof (val)))
    {
      valmask = (((ULONGEST) 1) << bitsize) - 1;
      val &= valmask;
      if (!TYPE_UNSIGNED (field_type))
	{
	  if (val & (valmask ^ (valmask >> 1)))
	    {
	      val |= ~valmask;
	    }
	}
    }
  return (val);
}

/* Modify the value of a bitfield.  ADDR points to a block of memory in
   target byte order; the bitfield starts in the byte pointed to.  FIELDVAL
   is the desired value of the field, in host byte order.  BITPOS and BITSIZE
   indicate which bits (in target bit order) comprise the bitfield.  
   Requires 0 < BITSIZE <= lbits, 0 <= BITPOS+BITSIZE <= lbits, and
   0 <= BITPOS, where lbits is the size of a LONGEST in bits.  */

void
modify_field (char *addr, LONGEST fieldval, int bitpos, int bitsize)
{
  ULONGEST oword;
  ULONGEST mask = (ULONGEST) -1 >> (8 * sizeof (ULONGEST) - bitsize);

  /* If a negative fieldval fits in the field in question, chop
     off the sign extension bits.  */
  if ((~fieldval & ~(mask >> 1)) == 0)
    fieldval &= mask;

  /* Warn if value is too big to fit in the field in question.  */
  if (0 != (fieldval & ~mask))
    {
      /* FIXME: would like to include fieldval in the message, but
         we don't have a sprintf_longest.  */
      warning ("Value does not fit in %d bits.", bitsize);

      /* Truncate it, otherwise adjoining fields may be corrupted.  */
      fieldval &= mask;
    }

  oword = extract_unsigned_integer (addr, sizeof oword);

  /* Shifting for bit field depends on endianness of the target machine.  */
  if (BITS_BIG_ENDIAN)
    bitpos = sizeof (oword) * 8 - bitpos - bitsize;

  oword &= ~(mask << bitpos);
  oword |= fieldval << bitpos;

  store_unsigned_integer (addr, sizeof oword, oword);
}
\f
/* Convert C numbers into newly allocated values */

struct value *
value_from_longest (struct type *type, LONGEST num)
{
  struct value *val = allocate_value (type);
  enum type_code code;
  int len;
retry:
  code = TYPE_CODE (type);
  len = TYPE_LENGTH (type);

  switch (code)
    {
    case TYPE_CODE_TYPEDEF:
      type = check_typedef (type);
      goto retry;
    case TYPE_CODE_INT:
    case TYPE_CODE_CHAR:
    case TYPE_CODE_ENUM:
    case TYPE_CODE_BOOL:
    case TYPE_CODE_RANGE:
      store_signed_integer (value_contents_raw (val), len, num);
      break;

    case TYPE_CODE_REF:
    case TYPE_CODE_PTR:
      store_typed_address (value_contents_raw (val), type, (CORE_ADDR) num);
      break;

    default:
      error ("Unexpected type (%d) encountered for integer constant.", code);
    }
  return val;
}


/* Create a value representing a pointer of type TYPE to the address
   ADDR.  */
struct value *
value_from_pointer (struct type *type, CORE_ADDR addr)
{
  struct value *val = allocate_value (type);
  store_typed_address (value_contents_raw (val), type, addr);
  return val;
}


/* Create a value for a string constant to be stored locally
   (not in the inferior's memory space, but in GDB memory).
   This is analogous to value_from_longest, which also does not
   use inferior memory.  String shall NOT contain embedded nulls.  */

struct value *
value_from_string (char *ptr)
{
  struct value *val;
  int len = strlen (ptr);
  int lowbound = current_language->string_lower_bound;
  struct type *string_char_type;
  struct type *rangetype;
  struct type *stringtype;

  rangetype = create_range_type ((struct type *) NULL,
				 builtin_type_int,
				 lowbound, len + lowbound - 1);
  string_char_type = language_string_char_type (current_language,
						current_gdbarch);
  stringtype = create_array_type ((struct type *) NULL,
				  string_char_type,
				  rangetype);
  val = allocate_value (stringtype);
  memcpy (value_contents_raw (val), ptr, len);
  return val;
}

struct value *
value_from_double (struct type *type, DOUBLEST num)
{
  struct value *val = allocate_value (type);
  struct type *base_type = check_typedef (type);
  enum type_code code = TYPE_CODE (base_type);
  int len = TYPE_LENGTH (base_type);

  if (code == TYPE_CODE_FLT)
    {
      store_typed_floating (value_contents_raw (val), base_type, num);
    }
  else
    error ("Unexpected type encountered for floating constant.");

  return val;
}

struct value *
coerce_ref (struct value *arg)
{
  struct type *value_type_arg_tmp = check_typedef (value_type (arg));
  if (TYPE_CODE (value_type_arg_tmp) == TYPE_CODE_REF)
    arg = value_at_lazy (TYPE_TARGET_TYPE (value_type_arg_tmp),
			 unpack_pointer (value_type (arg),		
					 VALUE_CONTENTS (arg)));
  return arg;
}

struct value *
coerce_array (struct value *arg)
{
  arg = coerce_ref (arg);
  if (current_language->c_style_arrays
      && TYPE_CODE (value_type (arg)) == TYPE_CODE_ARRAY)
    arg = value_coerce_array (arg);
  if (TYPE_CODE (value_type (arg)) == TYPE_CODE_FUNC)
    arg = value_coerce_function (arg);
  return arg;
}

struct value *
coerce_number (struct value *arg)
{
  arg = coerce_array (arg);
  arg = coerce_enum (arg);
  return arg;
}

struct value *
coerce_enum (struct value *arg)
{
  if (TYPE_CODE (check_typedef (value_type (arg))) == TYPE_CODE_ENUM)
    arg = value_cast (builtin_type_unsigned_int, arg);
  return arg;
}
\f

/* Should we use DEPRECATED_EXTRACT_STRUCT_VALUE_ADDRESS instead of
   EXTRACT_RETURN_VALUE?  GCC_P is true if compiled with gcc and TYPE
   is the type (which is known to be struct, union or array).

   On most machines, the struct convention is used unless we are
   using gcc and the type is of a special size.  */
/* As of about 31 Mar 93, GCC was changed to be compatible with the
   native compiler.  GCC 2.3.3 was the last release that did it the
   old way.  Since gcc2_compiled was not changed, we have no
   way to correctly win in all cases, so we just do the right thing
   for gcc1 and for gcc2 after this change.  Thus it loses for gcc
   2.0-2.3.3.  This is somewhat unfortunate, but changing gcc2_compiled
   would cause more chaos than dealing with some struct returns being
   handled wrong.  */
/* NOTE: cagney/2004-06-13: Deleted check for "gcc_p".  GCC 1.x is
   dead.  */

int
generic_use_struct_convention (int gcc_p, struct type *value_type)
{
  return !(TYPE_LENGTH (value_type) == 1
	   || TYPE_LENGTH (value_type) == 2
	   || TYPE_LENGTH (value_type) == 4
	   || TYPE_LENGTH (value_type) == 8);
}

/* Return true if the function returning the specified type is using
   the convention of returning structures in memory (passing in the
   address as a hidden first parameter).  GCC_P is nonzero if compiled
   with GCC.  */

int
using_struct_return (struct type *value_type, int gcc_p)
{
  enum type_code code = TYPE_CODE (value_type);

  if (code == TYPE_CODE_ERROR)
    error ("Function return type unknown.");

  if (code == TYPE_CODE_VOID)
    /* A void return value is never in memory.  See also corresponding
       code in "print_return_value".  */
    return 0;

  /* Probe the architecture for the return-value convention.  */
  return (gdbarch_return_value (current_gdbarch, value_type,
				NULL, NULL, NULL)
	  != RETURN_VALUE_REGISTER_CONVENTION);
}

void
_initialize_values (void)
{
  add_cmd ("convenience", no_class, show_convenience,
	   "Debugger convenience (\"$foo\") variables.\n\
These variables are created when you assign them values;\n\
thus, \"print $foo=1\" gives \"$foo\" the value 1.  Values may be any type.\n\n\
A few convenience variables are given values automatically:\n\
\"$_\"holds the last address examined with \"x\" or \"info lines\",\n\
\"$__\" holds the contents of the last address examined with \"x\".",
	   &showlist);

  add_cmd ("values", no_class, show_values,
	   "Elements of value history around item number IDX (or last ten).",
	   &showlist);
}