Refcount problem in ceval.c

[Berthold =?iso-8859-15?Q?H=F6llmann?=]

I have a problem with my debug Python 2.5.2 executable with Py_REF_DEBUG
and Py_TRACE_REFS set. With one of my scripts I get:

....
Fatal Python error: Python/ceval.c:947 object at 0x2aa3f4d840 has negative ref count -260424622217076 0230

Program received signal SIGABRT, Aborted.
[Switching to Thread 182905044128 (LWP 2595)]
0x0000002a95dde 479 in raise () from /lib64/tls/libc.so.6
(gdb) p ((PyObject*)0x2 aa3f4d840)->ob_type
$1 = (struct _typeobject *) 0xdbdbdbdbdbdbd bdb
(gdb) p ((PyObject*)0x2 aa3f4d840)->ob_refcnt
$2 = -260424622217076 0230
(gdb) p ((PyObject*)0x2 aa3f4d840)->_ob_next
$3 = (struct _object *) 0xdbdbdbdbdbdbd bdb
(gdb) p ((PyObject*)0x2 aa3f4d840)->_ob_prev
$4 = (struct _object *) 0xdbdbdbdbdbdbd bdb
(gdb)


Is there any "common" reason to for such a strange object on the command
stack, or is it more likely that any of my extension modules is causing
havoc?

Kind regards
Berthold
--
__ Address:
G / \ L Germanischer Lloyd
phone: +49-40-36149-7374 -+----+- Vorsetzen 35 P.O.Box 111606
fax : +49-40-36149-7320 \__/ D-20459 Hamburg D-20416 Hamburg

[Christian Heimes]

Re: Refcount problem in ceval.c

Berthold Höllmann wrote:

Is there any "common" reason to for such a strange object on the command
stack, or is it more likely that any of my extension modules is causing
havoc?

It's very likely that your extension has a reference counting bug. It
looks like you are either missing a Py_INCREF or you have a Py_DECREF
too much. Newly freed memory is filled with 0xDB (see Objects/obmalloc.c
DEADBYTE).

Wild guess: Are you using PyModule_AddObj ect with a PyTypeObject w/o
Py_INCREF()ing the type object first?

Christian

[Berthold =?iso-8859-15?Q?H=F6llmann?=]

Re: Refcount problem in ceval.c

Christian Heimes <lists@cheimes. dewrites:

Berthold Höllmann wrote:

>Is there any "common" reason to for such a strange object on the command
>stack, or is it more likely that any of my extension modules is causing
>havoc?

>
It's very likely that your extension has a reference counting bug. It
looks like you are either missing a Py_INCREF or you have a Py_DECREF
too much. Newly freed memory is filled with 0xDB (see
Objects/obmalloc.c DEADBYTE).

I was suspecting this, that's why I build the debugging version of
Python. I hoped I would get the error message somewhere near the code
causing the error, but I seems i have to continue the search.

Wild guess: Are you using PyModule_AddObj ect with a PyTypeObject w/o
Py_INCREF()ing the type object first?

That would have been easy :-) I have only one occurrence of
PyModule_AddObj ect, and its PyTypeObject is Py_INCREF()ed

Thanks
Berthold
--
__ Address:
G / \ L Germanischer Lloyd
phone: +49-40-36149-7374 -+----+- Vorsetzen 35 P.O.Box 111606
fax : +49-40-36149-7320 \__/ D-20459 Hamburg D-20416 Hamburg

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (GNU/Linux)

iD8DBQFIyMTEvnK LBupGT+8RArWiAK CBY40o5X7cfXEaJ tIE34aII67kEQCf dUte
ScgFJcxTta5fd3e VQcy7NZM=
=Z6Xr
-----END PGP SIGNATURE-----