Re: [Python] Add methods reference_value and const_value to gdb.Value

[Doug Evans <dje@google.com>]

Siva Chandra writes:
  > This will help address libstdc++/65840:
  > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65840
  >
  > gdb/ChangeLog:
  >
  > 2015-04-27  Siva Chandra Reddy  <sivachandra@google.com>
  >
  >         * NEWS (Python Scripting): Mention the new gdb.Value methods.
  >         * python/py-value.c (valpy_reference_value): New function.
  >         (valpy_const_value): Likewise.
  >         (value_object_methods): Add new methods.
  >         * value.c (make_cv_value): New function.
  >         * value.h (make_cv_value): Declare.
  >
  > gdb/doc/ChangeLog:
  >
  > 2015-04-27  Siva Chandra Reddy  <sivachandra@google.com>
  >
  >         * python.texi (Values From Inferior): Add descriptions of new
  >         methods gdb.Value.reference_value and gdb.Value.const_value.
  >
  > gdb/testsuite/ChangeLog:
  >
  > 2015-04-27  Siva Chandra Reddy  <sivachandra@google.com>
  >
  >         * gdb.python/py-xmethods.cc: Enhance test case.
  >         * gdb.python/py-xmethods.exp: New tests.
  >         * gdb.python/py-xmethods.py (A_indexoper): New xmethod worker
  >         function.
  >         (B_indexoper): Likewise.
  >         (global_dm_list) : Add new xmethod worker functions.

Thanks again for the ping.

LGTM with two nits below.  grep for ====.

One more nit might be that we already have gdb.Value.referenced_value,
and now we'll also have gdb.Value.reference_value.
I'm ok with it.

  > diff --git a/gdb/NEWS b/gdb/NEWS
  > index b711553..627afd0 100644
  > --- a/gdb/NEWS
  > +++ b/gdb/NEWS
  > @@ -52,6 +52,9 @@
  >       which is the name of the objfile as specified by the user,
  >       without, for example, resolving symlinks.
  >    ** You can now write frame unwinders in Python.
  > +  ** gdb.Value objects have new methods "reference_value" and
  > +     "const_value" which return a reference to the value and a
  > +     "const" version of the value respectively.
  >
  >  * New commands
  >
  > diff --git a/gdb/doc/python.texi b/gdb/doc/python.texi
  > index 448fa8b2..1b1079d 100644
  > --- a/gdb/doc/python.texi
  > +++ b/gdb/doc/python.texi
  > @@ -781,6 +781,16 @@ The @code{gdb.Value} object @code{py_val} is  
identical to that
  >  corresponding to @code{val}.
  >  @end defun
  >
  > +@defun Value.reference_value ()
  > +Return a @code{gdb.Value} object which is a reference to the value
  > +encapsulated by this instance.
  > +@end defun
  > +
  > +@defun Value.const_value ()
  > +Return a @code{gdb.Value} object which is a @code{const} version of the
  > +value encapsulated by this instance.
  > +@end defun
  > +
  >  @defun Value.dynamic_cast (type)
  >  Like @code{Value.cast}, but works as if the C@t{++} @code{dynamic_cast}
  >  operator were used.  Consult a C@t{++} reference for details.
  > diff --git a/gdb/python/py-value.c b/gdb/python/py-value.c
  > index 6622d11..94ff9ba 100644
  > --- a/gdb/python/py-value.c
  > +++ b/gdb/python/py-value.c
  > @@ -236,6 +236,60 @@ valpy_referenced_value (PyObject *self, PyObject  
*args)
  >    return result;
  >  }
  >
  > +/* Return a value which is a reference to the value.  */
  > +
  > +static PyObject *
  > +valpy_reference_value (PyObject *self, PyObject *args)
  > +{
  > +  PyObject *result = NULL;
  > +
  > +  TRY
  > +    {
  > +      struct value *self_val;
  > +      struct cleanup *cleanup = make_cleanup_value_free_to_mark  
(value_mark ());
  > +
  > +      self_val = ((value_object *) self)->value;
  > +      result = value_to_value_object (value_ref (self_val));
  > +
  > +      do_cleanups (cleanup);
  > +    }
  > +  CATCH (except, RETURN_MASK_ALL)
  > +    {
  > +      GDB_PY_HANDLE_EXCEPTION (except);
  > +    }
  > +  END_CATCH
  > +
  > +  return result;
  > +}
  > +
  > +/* Return a "const" qualified version of the value.  */
  > +
  > +static PyObject *
  > +valpy_const_value (PyObject *self, PyObject *args)
  > +{
  > +  PyObject *result = NULL;
  > +
  > +  TRY
  > +    {
  > +      struct value *self_val, *res_val;
  > +      struct type *const_type;

====
const_type can be deleted.

  > +      struct cleanup *cleanup = make_cleanup_value_free_to_mark  
(value_mark ());
  > +
  > +      self_val = ((value_object *) self)->value;
  > +      res_val = make_cv_value (1, 0, self_val);
  > +      result = value_to_value_object (res_val);
  > +
  > +      do_cleanups (cleanup);
  > +    }
  > +  CATCH (except, RETURN_MASK_ALL)
  > +    {
  > +      GDB_PY_HANDLE_EXCEPTION (except);
  > +    }
  > +  END_CATCH
  > +
  > +  return result;
  > +}
  > +
  >  /* Return "&value".  */
  >  static PyObject *
  >  valpy_get_address (PyObject *self, void *closure)
  > @@ -1692,6 +1746,10 @@ reinterpret_cast operator."
  >    { "dereference", valpy_dereference, METH_NOARGS, "Dereferences the  
value." },
  >    { "referenced_value", valpy_referenced_value, METH_NOARGS,
  >      "Return the value referenced by a TYPE_CODE_REF or TYPE_CODE_PTR  
value." },
  > +  { "reference_value", valpy_reference_value, METH_NOARGS,
  > +    "Return a value of type TYPE_CODE_REF referencing this value." },
  > +  { "const_value", valpy_const_value, METH_NOARGS,
  > +    "Return a 'const' qualied version of the same value." },
  >    { "lazy_string", (PyCFunction) valpy_lazy_string,
  >      METH_VARARGS | METH_KEYWORDS,
  >      "lazy_string ([encoding]  [, length]) -> lazy_string\n\
  > diff --git a/gdb/testsuite/gdb.python/py-xmethods.cc  
b/gdb/testsuite/gdb.python/py-xmethods.cc
  > index aedd1de..98bdb72 100644
  > --- a/gdb/testsuite/gdb.python/py-xmethods.cc
  > +++ b/gdb/testsuite/gdb.python/py-xmethods.cc
  > @@ -173,7 +173,7 @@ int main(void)
  >
  >    for (int i = 0; i < 10; i++)
  >      {
  > -      a1.array[i] = a2.array[i] = i;
  > +      a1.array[i] = a2.array[i] = b1.array[i] = i;
  >      }
  >
  >    return 0; /* Break here.  */
  > diff --git a/gdb/testsuite/gdb.python/py-xmethods.exp  
b/gdb/testsuite/gdb.python/py-xmethods.exp
  > index a83b14d..eacfa0c 100644
  > --- a/gdb/testsuite/gdb.python/py-xmethods.exp
  > +++ b/gdb/testsuite/gdb.python/py-xmethods.exp
  > @@ -100,6 +100,8 @@ gdb_test "p a1.geta()" "From Python  
<A_geta>.*5" "After: a1.geta()"
  >  gdb_test "p ++a1" "From Python <plus_plus_A>.*6" "After: ++a1"
  >  gdb_test "p a1.getarrayind(5)" "From Python <A_getarrayind>.*5" \
  >    "After: a1.getarrayind(5)"
  > +gdb_test "P a1\[6\]" ".*int &.*6" "After a1\[\]"
  > +gdb_test "P b1\[7\]" ".*const int &.*7" "After b1\[\]"
  >  # Note the following test.  Xmethods on dynamc types are not looked up
  >  # currently.  Hence, even though a_ptr points to a B object, the xmethod
  >  # defined for A objects is invoked.
  > diff --git a/gdb/testsuite/gdb.python/py-xmethods.py  
b/gdb/testsuite/gdb.python/py-xmethods.py
  > index 78935e1..3194e89 100644
  > --- a/gdb/testsuite/gdb.python/py-xmethods.py
  > +++ b/gdb/testsuite/gdb.python/py-xmethods.py
  > @@ -43,6 +43,12 @@ def A_getarrayind(obj, index):
  >    print('From Python <A_getarrayind>:')
  >    return obj['array'][index]
  >
  > +def A_indexoper(obj, index):
  > +  return obj['array'][index].reference_value()
  > +
  > +def B_indexoper(obj, index):
  > +  return obj['array'][index].const_value().reference_value()
  > +
  >
  >  type_A = gdb.parse_and_eval('(dop::A *) 0').type.target()
  >  type_B = gdb.parse_and_eval('(dop::B *) 0').type.target()
  > @@ -208,6 +214,16 @@ global_dm_list = [
  >                           '^getarrayind$',
  >                           A_getarrayind,
  >                           type_int),
  > +    SimpleXMethodMatcher('A_indexoper',
  > +                         '^dop::A$',
  > +                         'operator\\[\\]',
  > +                         A_indexoper,
  > +                         type_int),
  > +    SimpleXMethodMatcher('B_indexoper',
  > +                         '^dop::B$',
  > +                         'operator\\[\\]',
  > +                         B_indexoper,
  > +                         type_int)
  >  ]
  >
  >  for matcher in global_dm_list:
  > diff --git a/gdb/value.c b/gdb/value.c
  > index cb56849..242b46c 100644
  > --- a/gdb/value.c
  > +++ b/gdb/value.c
  > @@ -1704,6 +1704,22 @@ value_copy (struct value *arg)
  >    return val;
  >  }
  >
  > +/* Return a "const" and/or "volatile" qualified version of the value V.
  > +   If CNST is true, then the returned value will be qualified with
  > +   "const".
  > +   if VOLTL is true, then the returned value will be qualified with
  > +   "volatile".  */
  > +
  > +struct value *
  > +make_cv_value (int cnst, int voltl, struct value *v)
  > +{
  > +  struct value *cv_val = value_copy (v);
  > +
  > +  cv_val->type = make_cv_type (cnst, voltl, value_type (v), NULL);
  > +
  > +  return cv_val;
  > +}

====
I'm not sure the call to value_copy will always produce the desired result,
but I don't have anything specific to say. So let's go with it.
I'd replace the assignment to cv_val->type with a call to
deprecated_set_value_type though.
Do we need to call set_value_enclosing_type too?
For consistency I'd think so.

  > +
  >  /* Return a version of ARG that is non-lvalue.  */
  >
  >  struct value *
  > diff --git a/gdb/value.h b/gdb/value.h
  > index 21baa32..9aeaa4a 100644
  > --- a/gdb/value.h
  > +++ b/gdb/value.h
  > @@ -1040,6 +1040,8 @@ extern struct value *value_non_lval (struct value  
*);
  >
  >  extern void value_force_lval (struct value *, CORE_ADDR);
  >
  > +extern struct value *make_cv_value (int, int, struct value *);
  > +
  >  extern void preserve_one_value (struct value *, struct objfile *,  
htab_t);
  >
  >  /* From valops.c */