momory fault on exit

[sjoerdmullender]

BPO	675341
Nosy	@loewis, @sjoerdmullender

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/loewis'
closed_at = <Date 2003-01-27.12:18:48.000>
created_at = <Date 2003-01-27.10:32:43.000>
labels = ['interpreter-core']
title = 'momory fault on exit'
updated_at = <Date 2003-01-27.12:18:48.000>
user = 'https://github.com/sjoerdmullender'

bugs.python.org fields:

activity = <Date 2003-01-27.12:18:48.000>
actor = 'sjoerd'
assignee = 'loewis'
closed = True
closed_date = None
closer = None
components = ['Interpreter Core']
creation = <Date 2003-01-27.10:32:43.000>
creator = 'sjoerd'
dependencies = []
files = []
hgrepos = []
issue_num = 675341
keywords = []
message_count = 3.0
messages = ['14226', '14227', '14228']
nosy_count = 2.0
nosy_names = ['loewis', 'sjoerd']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = None
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue675341'
versions = ['Python 2.3']

[sjoerdmullender]

Martin, I think this has to do with your changes to
iconv_codec.

After cvs update; configure; make, I get a seg fault on
exit of the interpreter after doing nothing other than
typing ctrl-D. This is on Linux (RedHat 8.0).

+ ./python
Python 2.3a1 (#1, Jan 27 2003, 11:06:58)
[GCC 3.2 2002090 (Red Hat Linux 8.0 3.2-7)] on linux2
Type "help", "copyright", "credits" or "license" for
more information.

>>
Memory fault

However:
+ ./python -S
Python 2.3a1 (#1, Jan 27 2003, 11:06:58)
[GCC 3.2 2002090 (Red Hat Linux 8.0 3.2-7)] on linux2
Type "help", "copyright", "credits" or "license" for
more information.

>>

Stack trace:

(gdb) bt
#0 0x0808606c in type_dealloc (type=0x4001d400)
at ../Objects/typeobject.c:1998
#1 0x08073179 in dict_dealloc (mp=0x400f4a44) at
../Objects/dictobject.c:686
#2 0x08073179 in dict_dealloc (mp=0x4009a8ac) at
../Objects/dictobject.c:686
#3 0x080bcb1d in _PyImport_Fini () at
../Python/import.c:224
#4 0x080c57b4 in Py_Finalize () at
../Python/pythonrun.c:232
#5 0x08054b1d in Py_Main (argc=1, argv=0xbffff5e4) at
../Modules/main.c:457
#6 0x0805431a in main (argc=1, argv=0xbffff5e4) at
../Modules/python.c:23
#7 0x420158d4 in __libc_start_main () from
/lib/i686/libc.so.6

And the reason I think Martin has something to do with
it (note the tp_name field):
(gdb) p *type
$2 = {ob_refcnt = 0, ob_type = 0x8104b20, ob_size = 0,
tp_name = 0x4001cbf2 "iconvcodec", tp_basicsize = 20,
tp_itemsize = 0,
tp_dealloc = 0x4001c81e <iconvcodec_dealloc>,
tp_print = 0, tp_getattr = 0,
tp_setattr = 0, tp_compare = 0, tp_repr = 0x4001c8ca
<iconvcodec_repr>,
tp_as_number = 0x0, tp_as_sequence = 0x0,
tp_as_mapping = 0x0, tp_hash = 0,
tp_call = 0, tp_str = 0, tp_getattro = 0x807786a
<PyObject_GenericGetAttr>,
tp_setattro = 0, tp_as_buffer = 0x0, tp_flags = 17899,
tp_doc = 0x4001d020 "iconvcodec object", tp_traverse
= 0, tp_clear = 0,
tp_richcompare = 0, tp_weaklistoffset = 0, tp_iter =
0, tp_iternext = 0,
tp_methods = 0x4001d3c0, tp_members = 0x0, tp_getset
= 0x0, tp_base = 0x0,
tp_dict = 0x0, tp_descr_get = 0, tp_descr_set = 0,
tp_dictoffset = 0,
tp_init = 0, tp_alloc = 0x80833e6 <PyType_GenericAlloc>,
tp_new = 0x4001c5c4 <iconvcodec_new>, tp_free =
0x80ccc0e <PyObject_GC_Del>,
tp_is_gc = 0, tp_bases = 0x0, tp_mro = 0x0, tp_cache
= 0x0,
tp_subclasses = 0x0, tp_weaklist = 0x0, tp_del = 0}

[loewis]

Logged In: YES
user_id=21627

This is fixed in _iconv_codec.c 1.4.

I'm surprised that it happens in this way for you, though,
since normally, importing site.py won't import _iconv_codec.
The only explanation is that you have changed site.py in
some way.

[sjoerdmullender]

Logged In: YES
user_id=43607

The reason _iconv_codec got imported is because I have
JapanesCodecs installed which is imported by addpackage in
site.py. And JapaneseCodecs imports encodings which in turn
imports iconv_codecs.

Things work now thanks to your fix. Thanks.