Open eldersixpack13 opened 11 months ago
So the steps are: update via gui, and find a kodi crash log in /storage/.kodi/temp? I can't reproduce with an RPi3. I used the November 24 -> November 25 nightly RPi2 builds as in your logs.
Would you try logging in via ssh and running systemctl --no-block reboot
and see if it produces a crash log? This is the reboot command being used when the addon performs an update.
Thanks for looking into it.
"systemctl --no-block reboot" (without quotes) does not produce a crash log.
I'm no developer (just an IT-Monkey (Support, Admin) retired, using ~20 years Linux (Redhat 6.x, Fedora since FC1) To me it seems /usr/lib/libpython is running wild.
Box was new installed ~3 month ago, LE11, upgrade to LE12, ...
just did a update to 505b95f and with "Network => now with Maximum Wait Time: 10 sec" => no crash log
will retest with Maximum Wait Time: 5 sec again and report
@antonlacon
It may not crash if you are going directly to update. I can reproduce the issue by selecting all LE-Settings tabs and then performing the update.
We are using TimeoutStopSec=30
for kodi. On weaker devices this may not be enough time to create a crash log. Consider increasing this value with a systemd drop in file.
You can use the persistent journal setting to look for crashes without log in shutdown e.g.:
Dez 02 11:38:20.768693 marsle12 kernel: LanguageInvoker[1043]: segfault at 6 ip 0000000000000006 sp 00007f56257f9218 error 14 in kodi.bin[400000+1b56000] likely on CPU 2 (core 2, socket 0)
Dez 02 11:38:20.768798 marsle12 kernel: Code: Unable to access opcode bytes at 0xffffffffffffffdc.
Dez 02 11:38:21.355624 marsle12 kodi.sh[1002]: Segmentation fault (core dumped)
Finally here is a crash dump of a debug enabled image: kodi_crashlog_20231202113821.log
If you have any idea how to proceed ... ;-)
I added a drop-in at /storage/.config/system.d/kodi.service
on the December 4 RPi2 nightly image.
[Service]
TimeoutStopSec=60
I can provoke a crash by restarting the Kodi service. Journalctl shows:
Dec 04 03:12:05.077097 LibreELEC systemd[1]: Starting kodi.service...
Dec 04 03:12:05.104307 LibreELEC systemd[1]: Started kodi.service.
Dec 04 03:12:05.106640 LibreELEC (kodi.sh)[867]: kodi.service: Referenced but unset environment variable evaluates to an empty string: KODI_DEBUG
Dec 04 03:12:12.457985 LibreELEC kodi.sh[871]: double free or corruption (out)
Dec 04 03:12:30.372162 LibreELEC kodi.sh[867]: Aborted (core dumped)
Dec 04 03:12:34.112408 LibreELEC kodi.sh[867]: Crash report available at /storage/.kodi/temp/kodi_crashlog_20231204031230.log
Dec 04 03:12:34.191708 LibreELEC systemd[1]: kodi.service: Main process exited, code=exited, status=134/n/a
Dec 04 03:12:34.228954 LibreELEC systemd[1]: kodi.service: Failed with result 'exit-code'.
Dec 04 03:12:34.229797 LibreELEC systemd[1]: kodi.service: Consumed 14.643s CPU time.
The error shows up somewhere around 20% of the time. Same error can show up from rebooting the system.
Is this still happening with current nightlies.
Yes !
seems to be fixed since nightly (18.01.2024) Thanks !
It crashes most of the time but not at any update.
Reopening.
@mglae - is this still an issue, or did it get fixed in the current LE12b1?
Still an issue here. In January I made a wrong assumption after 3 in row without crashlog
Status request: Is this still occurring in LE12 (released)? Is this occurring in LE13 nightlies?
yes on LE13 nightly 738b75b
@eldersixpack13 are you still experiencing this?
Yes
Describe the bug
see Bug #7987
LE 12 nightly update from 52b450b to 8b39055, but it happend earlier too
How to reproduce
Steps to reproduce the behavior:
Information
Log file
log during update: kodi_crash.log
log after update: kodi.log
Context
regarding the logs: I guess the crash happens during the update (see nightly git hashes)
Network => Maximum Wait Time: 5 sec