Apache Segmentation fault when trying to open a changeset

[dk@…]

Hello,

after upgrading my debian system I had severe troubles when trying to get trac running again.

Now - after mod_python segfaults right when I try to access a trac page I switched to wsgi which seems to work except when I click on a changeset inside the timeline. The I get a segfault.

The backtrace of the coredump only shows

[..]
Reading symbols from /usr/lib/python2.4/lib-dynload/sha.so...Reading symbols from /usr/lib/debug/usr/lib/python2.4/lib-dynload/sha.so...done.
done.
Loaded symbols for /usr/lib/python2.4/lib-dynload/sha.so
Reading symbols from /usr/lib/python2.4/lib-dynload/termios.so...Reading symbols from /usr/lib/debug/usr/lib/python2.4/lib-dynload/termios.so...done.
done.
Loaded symbols for /usr/lib/python2.4/lib-dynload/termios.so
Core was generated by `/usr/sbin/apache2 -k start'.
Program terminated with signal 11, Segmentation fault.
#0  0x0000000080d080a6 in ?? ()
(gdb) bt
#0  0x0000000080d080a6 in ?? ()
#1  0x0000000000000000 in ?? ()

Browsing the source works.

Opening the changeset with tracd works also.

Maybe this also leads to the cause of the mod_python segfault. I am completely lost. Everything seems to be broken.

[ThurnerRupert]

could you try to upgrade to python-2.5? with mod_python there was a segmentation fault problem, see ​here. as genshi uses xml templates it exposes this error immediatly.

for wsgi: what are you using there, is there mod_wsgi? and might it be a mod_wsgi issue?

[sid]

Sorry this hasn't gotten more attention. Were you able to resolve your problem? Did Thurner's suggestions help?

[Christian Boos]

As rupert said, if this was Trac trunk with Genshi, then it was most probably the expat issue. Which makes me think we should make that more apparent in the TracModPython and wiki:TracModWSGI documentation.

[Christian Boos]

Just noticed that in the description:

Browsing the source works.

… so this might rather have been a SVN Python binding issue, possibly triggered by incompatible apr libraries.

Also, as a general advice, when you get such unusable backtraces from gdb, best is to first attach to the running process, then trigger the bug. The backtrace is obtained more reliably this way.

[anonymous]

When i tried to run in cgi every thing is binds correctly and works fine when i tried the same with mod_python it gives a segmentation fault error to me..

Regards vela.velappan

[ebray]

Replying to anonymous:

When i tried to run in cgi every thing is binds correctly and works fine when i tried the same with mod_python it gives a segmentation fault error to me..

Regards vela.velappan

Yes, well, if you read the ticket you'd see that the likely source of this problem is known, and has even been documented here. Read the site linked to there, and come back if that doesn't seem to be it. Regardless, I don't think this is something that can be fixed in Trac.