Product: Fedora
https://bugzilla.redhat.com/show_bug.cgi?id=889131
Bug ID: 889131
Summary: [abrt] jpackage-utils-1.7.5-18.1.fc17:
zipfile.py:781:_RealGetContents:IOError: [Errno 22]
Invalid argument
Product: Fedora
Version: 18
Component: javapackages-tools
Severity: unspecified
Priority: unspecified
Reporter: mizdebsk(a)redhat.com
+++ This bug was initially created as a clone of Bug #853759 +++
Version-Release number of selected component:
jpackage-utils-1.7.5-18.1.fc17
Additional info:
libreport version: 2.0.13
abrt_version: 2.0.12
cmdline: /usr/bin/python /usr/lib/rpm/osgi.prov
kernel: 3.5.2-3.fc17.x86_64
--- Additional comment from Ilyes Gouta on 2012-09-02 19:45:38 CEST ---
Created attachment 609141
File: backtrace
--- Additional comment from Ilyes Gouta on 2012-09-02 19:45:40 CEST ---
Created attachment 609142
File: core_backtrace
--- Additional comment from Przemek Klosowski on 2012-09-05 17:42:24 CEST ---
rebooted after update, logged into standard Gnome session, got black background
error announcement
Package: jpackage-utils-1.7.5-18.1.fc17
OS Release: Fedora release 17 (Beefy Miracle)
--- Additional comment from Mikolaj Izdebski on 2012-12-19 19:01:20 CET ---
For me this looks like a bug in python - _RealGetContents from zipfile.py is
calling seek() with invalid argument.
Reassigning to python package.
--- Additional comment from Bohuslav "Slavek" Kabrda on 2012-12-20 09:30:37 CET
---
Notes:
- Although the error is raised in _RealGetContents, the actual problem seems to
be in zipfile.is_zipfile, which wrongly returns True for /usr/bin/fastjar
(normal binary).
- Reproduce:
-- yum install libgcj, that provides /usr/bin/fastjar
-- # the following line should print False, but prints True
-- python -c "import zipfile; print
zipfile.is_zipfile('/usr/bin/fastjar')"
- The core of the problem seems to be in fact, that stringEndArchive
("PK\005\006") is found in the file.
Reported upstream as
http://bugs.python.org/issue16735
--- Additional comment from Stanislav Ochotnicky on 2012-12-20 09:38:41 CET ---
Part of the problem is that osgi.prov should have never been called on that
binary because it doesn't end with ".jar". Quick look doesn't tell me
why it
was called because we have this in osgi.attr (rpm auto-requires/provides
generator):
%__osgi_provides %{_rpmconfigdir}/osgi.prov
#%__osgi_requires %{_rpmconfigdir}/osgi.req
%__osgi_path ^(.*\.jar|((%{_libdir}|%{_datadir}).*/MANIFEST.MF))
Note that third line has correctly escaped dot before jar
--- Additional comment from Mikolaj Izdebski on 2012-12-20 10:39:24 CET ---
(In reply to comment #7)
Note that third line has correctly escaped dot before jar
I think there needs to be double escaping - one for rpm and one for glibc.
--- Additional comment from Mikolaj Izdebski on 2012-12-20 10:57:39 CET ---
Created attachment 666574
Reproducer
Reproduce with: rpmbuild -ba foo.spec
--- Additional comment from Mikolaj Izdebski on 2012-12-20 11:08:07 CET ---
Fixed upstream:
http://git.fedorahosted.org/cgit/javapackages.git/commit/?id=5a855c8b189a...
--
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug
https://bugzilla.redhat.com/token.cgi?t=F2JO9FI4ut&a=cc_unsubscribe