Re: [PATCH 6/6] gdb/python: Add gdb.InferiorThread.__dict__ attribute

[Tom Tromey <tom@tromey.com>]

>>>>> "Andrew" == Andrew Burgess <aburgess@redhat.com> writes:

Andrew> +  thread_obj->dict = PyDict_New ();
Andrew> +  if (thread_obj->dict == nullptr)
Andrew> +    return nullptr;

If this fails...

Andrew>  static void
Andrew>  thpy_dealloc (PyObject *self)
Andrew>  {
Andrew> -  Py_DECREF (((thread_object *) self)->inf_obj);
Andrew> +  thread_object *thr_obj = (thread_object *) self;
Andrew> +
Andrew> +  gdb_assert (thr_obj->inf_obj != nullptr);
Andrew> +  gdb_assert (thr_obj->dict != nullptr);
Andrew> +
Andrew> +  Py_DECREF (thr_obj->inf_obj);
Andrew> +  Py_DECREF (thr_obj->dict);

... won't this be called and then crash because dict is NULL?

I am not really sure TBH but other code, like your change in
py-inferior.c, seems to think this.

I suspect Py_XDECREF should be used instead.

The current call to Py_DECREF isn't a bug because create_thread_object
avoids creating the thread object when 'inf_obj' would be null.  But,
the new patch doesn't do this.

Tom