python2.6 (python-visual) crashed with SIGSEGV in __cxa_allocate_exception()

Bug #408663 reported by C Anthony Risinger
28
This bug affects 4 people
Affects Status Importance Assigned to Milestone
boost1.38 (Ubuntu)
Invalid
Undecided
Unassigned
python-visual (Ubuntu)
New
Medium
Unassigned

Bug Description

Binary package hint: python2.6

Description: Ubuntu karmic (development branch)
Release: 9.10

python-visual:
  Installed: 1:5.11-1
  Candidate: 1:5.11-1
  Version table:
 *** 1:5.11-1 0
        500 http://us.archive.ubuntu.com karmic/universe Packages
        100 /var/lib/dpkg/status

EXPECTED:

$ python
>>> from visual import *
>>>

ACTUAL:

$ python
>>> from visual import *
Segmentation fault (core dumped)
$

This is a fresh installation of Karmic, full updates as of ten minutes ago.

ProblemType: Crash
Architecture: i386
Date: Mon Aug 3 20:33:59 2009
Dependencies:
 findutils 4.4.2-1
 gcc-4.4-base 4.4.1-1ubuntu1
 libc6 2.10.1-0ubuntu1
 libgcc1 1:4.4.1-1ubuntu1
 zlib1g 1:1.2.3.3.dfsg-13ubuntu1
DistroRelease: Ubuntu 9.10
ExecutablePath: /usr/bin/python2.6
Package: python2.6-minimal 2.6.2-0ubuntu3
ProcCmdline: python
ProcEnviron:
 SHELL=/bin/bash
 LANG=en_US.UTF-8
ProcVersionSignature: Ubuntu 2.6.31-3.19-generic
SegvAnalysis:
 Segfault happened at: 0x1c69fa7 <__cxa_allocate_exception+55>: addl $0x1,0x4(%eax)
 PC (0x01c69fa7) ok
 source "$0x1" ok
 destination "0x4(%eax)" (0x0000000c) not located in a known VMA region (needed writable region)!
SegvReason: writing NULL VMA
Signal: 11
SourcePackage: python2.6
StacktraceTop:
 __cxa_allocate_exception () from /usr/lib/libstdc++.so.6
 boost::python::throw_error_already_set() ()
 boost::python::objects::class_base::add_static_property(char const*, boost::python::api::object const&, boost::python::api::object const&)
 ?? () from /usr/lib/python2.6/dist-packages/cvisualmodule.so
 ?? () from /usr/lib/python2.6/dist-packages/cvisualmodule.so
Title: python2.6 crashed with SIGSEGV in __cxa_allocate_exception()
Uname: Linux 2.6.31-3-generic i686
UserGroups: adm admin cdrom dialout lpadmin plugdev sambashare

Revision history for this message
C Anthony Risinger (extofme) wrote :
visibility: private → public
Revision history for this message
Apport retracing service (apport) wrote : Stacktrace.txt (retraced)

StacktraceTop:__cxa_allocate_exception (thrown_size=4)
boost::python::throw_error_already_set ()
boost::python::objects::class_base::add_static_property (
cvisual::wrap_display_kernel ()
cvisual::init_module_cvisual ()

Revision history for this message
Apport retracing service (apport) wrote : ThreadStacktrace.txt (retraced)
Changed in python2.6 (Ubuntu):
importance: Undecided → Medium
tags: removed: need-i386-retrace
Revision history for this message
C Anthony Risinger (extofme) wrote :

also according to people on the mailing list this package:

https://gutefee.massey.ac.nz/moin/Python/3D?action=AttachFile&do=view&target=python-visual_5.11-1_i386.deb

was built from karmic sources with build-dep/buildpackage'ed into a working jaunty package for himself, and he claims it works perfectly. however, i could not get that package to work with a vanilla jaunty install, nor the official package on a vanilla karmic install.

same result/sigsegv on both.

tried the build-deps thinking maybe the package was simply missing something in the deps, even tried rebuilding the package that way, same result, segfault.

since the guy that made those packages got it to work, it seems there is just a missing dependency or something.

Revision history for this message
C Anthony Risinger (extofme) wrote :

probably not a bug against python itself

affects: python2.6 (Ubuntu) → python-visual (Ubuntu)
Revision history for this message
C Anthony Risinger (extofme) wrote :

since no deps had been intoduced, i force installed the package from Debian (the Debian package does not install to python 2.6 dirs), and ran visual-python under python2.5

worked perfectly on ubuntu karmic

i will rebuild the package from debian to mirror the dependencies in the ubuntu package, and make sure it works under python2.6; the ubuntu package currently segfaults in 2.5 also

will probably compare the package contents as well

Revision history for this message
gs (gs-orst) wrote :

Please consider raising importance from medium to high. Not having this package working is a deal breaker in many educational environments.

Revision history for this message
C Anthony Risinger (extofme) wrote :

same problem, fresh installation on both an x86 laptop and an x86_64 desktop

still segfaults

Revision history for this message
gs (gs-orst) wrote :

I tried again with karmic beta . same result: seg fault.

suggested solution: Go back to previous version. There is absolutely no point in shipping a package which does nothing but crash.

Revision history for this message
Guy K. Kloss (guy.kloss) wrote :

I was the one who has backported the 5.11 karmic package to jaunty back then. Now being with one box on a kubuntu karmic beta gives me quite a headache as well.

I'm getting the same results, both when I'm installing the distribution provided binary package as well as when I'm using the dpkg-buildpackage method on the karmic sources myself. The build and install works flawlessly, but when executing a sample shows some trouble, which is different from the symptoms mentioned above:

guy@mountpaku:/usr/share/doc/python-visual/examples$ python orbit.py
Traceback (most recent call last):
  File "orbit.py", line 1, in <module>
    from visual import *
  File "/usr/lib/python2.6/dist-packages/visual/__init__.py", line 59, in <module>
    import cvisual
AttributeError: 'Boost.Python.StaticProperty' object attribute '__doc__' is read-only

Any clues? Has maybe the boost version changed in an incompatible way, or are some dependencies not listed in the source package, which just happened to be satisfied on my old Jaunty backport due to other pre-conditions of my system?

Revision history for this message
Guy K. Kloss (guy.kloss) wrote :

In an attempt to narrow down the problem with the python-visual 5.11 source package I have taken the python-visual 5.12 source package from Debian Sid, modified it's build rules (particularly to use libboost 1.38 rather 1.39) and built a package of it.

Unfortunately the effect is the same as described in my comment #10 just above.

However, it seems like there may be something else wrong, rather than the source version of python-visual (5.11, 5.12 or the recent 5.13). Maybe it's got something to do with the version of Boost.Python used. Reading the INSTALL.txt from the original python-visual sources indicates that different Boost versions may be the cause for certain troubles.

Revision history for this message
Guy K. Kloss (guy.kloss) wrote :

Trying to track down the problem further I've hit some leads that seem to
indicate that there seem to be quite some problems on Ubuntu Karmic with
Boost.Python (version 1.38.0-6ubuntu6). The PySide team has had also some
similar problems, as well as someone else. Here's a thread on the Py++ mailing
list on this

http://ur1.ca/du0m

It seems like Visual is not using .def_readonly(...), but I guess then that
this is only one of the ways to trigger this bug/behaviour.

Maybe some Boost specialist knows an approach to tackle this problem? Maybe this is even a bug in the Boost libs?

Revision history for this message
Guy K. Kloss (guy.kloss) wrote :

Further collaborative investigation has shown that Python >= 2.6.3 has implemented the __doc__ property differently. Many projects that are binding native libraries to Python through Boost.Python are suffering from this. Apparently there is a patch in the SVN version of Boost.Python that addresses this bug, according to this post from the Python Ogre crowd:

http://groups.google.com/group/python-ogre-developers/browse_thread/thread/130617f8c69f1bcb#

So I guess the only solution for Py++, PyOgre, python-visual, etc. is to hope soon for an updated version of Boost in Karmic ...

Revision history for this message
Guy K. Kloss (guy.kloss) wrote :

New bug opened for this in the more suitable area for boost1.38:

https://bugs.launchpad.net/ubuntu/+source/boost1.38/+bug/457688

Revision history for this message
Andrew Mitchell (ajmitch) wrote :

Split the boost part into a separate bug (https://bugs.launchpad.net/ubuntu/+source/boost1.38/+bug/457688)

Changed in boost1.38 (Ubuntu):
status: New → Invalid
Revision history for this message
Scott Howard (showard314) wrote :

Thank you for taking the time to report this bug and helping to make Ubuntu better. This particular bug has already been reported and is a duplicate of bug 457688, so it is being marked as such. Please look at the other bug report to see if there is any missing information that you can provide, or to see if there is a workaround for the bug. Additionally, any further discussion regarding the bug should occur in the other report. Please continue to report any other bugs you may find.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.