Am 23.01.23 um 23:07 schrieb piorunz:
On 23/01/2023 21:59, Bernhard Übelacker wrote:
Hello Piotr,
if it is not clear it would be possible to take
the bug number from the email addresses and open
the bug page e.g. https://bugs.debian.org/1028208.
This was just about the plasma-discover crash part.
However, you can also iterate through past logs by
showing the different boots like this:
$ journalctl --user --list-boots
IDX BOOT ID FIRST ENTRY LAST
ENTRY
...
-14 Thu 2022-12-29 16:49:33 CET Mon
2023-01-09 12:05:55 CET
...
And then show the logging for your user at that boot and
scroll to the wanted time frame:
$ journalctl --user -b -14
Kind regards,
Bernhard
Thank you Bernhard,
I've extracted logs from that time, please see attached.
Hello Piotr,
ok, then I got distracted by this from coredumpctl, showing "GMT",
while it might be really localtime ... at least it is the same as in journalctl
...
Timestamp: Sun 2023-01-08 13:36:08 GMT (1h 29min ago)
Your log shows four times a crashing kded5 before, then the plasma-discover:
Jan 08 13:35:03 ryzen systemd-coredump[2835427]: Process 1997855 (kded5) of
user 1000 dumped core.
Jan 08 13:35:04 ryzen systemd-coredump[2835810]: Process 2835418 (kded5) of
user 1000 dumped core.
Jan 08 13:35:06 ryzen systemd-coredump[2836051]: Process 2835806 (kded5) of
user 1000 dumped core.
Jan 08 13:35:08 ryzen systemd-coredump[2836163]: Process 2836048 (kded5) of
user 1000 dumped core.
Jan 08 13:36:09 ryzen systemd-coredump[2842084]: Process 2832872
(plasma-discover) of user 1000 dumped core.
And before the plasma-discover PID 2832872 shows:
Jan 08 13:35:00 ryzen plasmashell[2832872]: adding empty sources model
QStandardItemModel(0x55cb965ccf70)
Jan 08 13:35:00 ryzen plasmashell[2832872]:
file:///usr/lib/x86_64-linux-gnu/qt5/qml/QtQuick/Controls.2/org.kde.desktop/private/TextFieldContextMenu.qml:159:5:
QML MenuItem: Binding loop detected for property "implicitWidth"
Jan 08 13:35:00 ryzen plasmashell[2832872]:
file:///usr/lib/x86_64-linux-gnu/qt5/qml/org/kde/kirigami.2/private/PrivateActionToolButton.qml:76:5:
QML Binding: Binding loop detected for property "value"
Jan 08 13:35:00 ryzen plasmashell[2832872]: packagekitqt.transaction: Unknown Transaction
property: "Sender" QVariant(QString, ":1.577")
Jan 08 13:35:01 ryzen plasmashell[2832872]: packagekitqt.transaction: Unknown Transaction
property: "Sender" QVariant(QString, ":1.577")
Jan 08 13:35:01 ryzen plasmashell[2832872]: packagekitqt.transaction: Unknown Transaction
property: "Sender" QVariant(QString, ":1.577")
Jan 08 13:35:03 ryzen plasmashell[2832872]: took really long to fetch
PackageKitBackend(0x55cb964b2e50)
Jan 08 13:35:03 ryzen plasmashell[2832872]: 29 -- exe=/usr/bin/plasma-discover
Jan 08 13:35:03 ryzen plasmashell[2832872]: KCrash: crashing...
crashRecursionCounter = 2
So I assume this plasma-discover crash is a result
of the crash of kded5 with PID 1997855 in the same second before:
#2
_ZN6KCrash19defaultCrashHandlerEi (libKF5Crash.so.5 + 0x586d)
#3 n/a (libc.so.6 + 0x3bf90)
#4
_ZNK10PackageKit11Transaction4roleEv (libpackagekitqt5.so.1 + 0x1aba4)
#5 n/a (kded_apperd.so +
0xeaae)
#6 n/a (kded_apperd.so +
0xeb99)
#7 n/a (libQt5Core.so.5 +
0x2e8fcf)
#8
_ZN10PackageKit6Daemon22transactionListChangedERK11QStringList
(libpackagekitqt5.so.1 + 0xe095)
Which looks like that from Debian bug #1026062.
The 60 seconds between the crashes might be the timeout
of KCrash to give the user the chance to click the icon
in the systray, after that time it "forwards" the crash
to get collected by systemd-coredump.
So in the end it looks like plasma-discover is not prepared
for kded5 crashing, not ideal, but if kded5 gets fixed this
plasma-discover crash might disappear too.
Kind regards,
Bernhard
