[reportlab-users] Reproducible crash of reportlab 3.5.50 on x86_64 with current Python3
hpj at urpla.net
Tue Sep 22 14:38:36 EDT 2020
while working on packaging the current version of reportlab on openSUSE, I
stumbled upon a reproducible crash in the tests:
[ 9s] + export PYTHONPATH=/home/abuild/rpmbuild/BUILDROOT/python-reportlab-3.5.50-0.x86_64/usr/lib64/python3.8/site-packages
[ 9s] + PYTHONPATH=/home/abuild/rpmbuild/BUILDROOT/python-reportlab-3.5.50-0.x86_64/usr/lib64/python3.8/site-packages
[ 9s] + python3 runAll.py
[ 11s] .............*** Error in `/usr/bin/python3': free(): invalid pointer: 0x0000562662d76b20 ***
[ 12s] F.........................*** Error in `python3': free(): invalid pointer: 0x0000560d03b48d20 ***
[ 12s] /var/tmp/rpm-tmp.FHfzwC: line 45: 24177 Aborted (core dumped) python3 runAll.py
[ 12s] error: Bad exit status from /var/tmp/rpm-tmp.FHfzwC (%check)
Interestingly, the crash didn't happen on i586 builds, although it's using
the same compiler, python, etc.
This is reproduced on several targets Leap 15.1, 15.2 and Tumbleweed (a
pretty current rolling release), and Python 3.8.5.
Does this ring a bell for anybody?
The build can be accessed here with full logs:
Note, that I'm disabling the tests for now, but they can be switched on easily.
I've attempted to file a an issued on bitbucket, but they completely removed
the mercurial repos.
More information about the reportlab-users