https://bugs.kde.org/show_bug.cgi?id=524933
Engin Karahan <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |[email protected] --- Comment #3 from Engin Karahan <[email protected]> --- Confirming on newer versions: Fedora 44, Plasma 6.7.5, KF 6.30.0 (kf6-networkmanager-qt-6.30.0), Qt 6.11.2. Scenario: laptop undocked and lid closed (no suspend, no external display), Wi-Fi connected, ~36 access points in range. plasmashell's main thread hung shortly afterwards and stayed hung for ~2.5 days until I restarted plasma-plasmashell.service. Panel, tray and notifications were dead, KWin and applications kept working. Main thread from the core dump (SIGABRT on restart), identical to the report: #8 QWaitCondition::wait #9 QDBusPendingCallPrivate::waitForFinished #10 QDBusConnectionPrivate::sendWithReply #11 QDBusConnection::call #12 NetworkManager::NetworkManagerPrivate::retrieveInitialProperties #13 NetworkManager::AccessPoint::AccessPoint #14 NetworkManager::WirelessDevicePrivate::accessPointAdded ... #18 QDBusConnectionPrivate::deliverCall Additional observation: the QDBusConnectionManager thread was idle at the same time (QThread::exec -> QEventDispatcherGlib::processEvents -> ppoll), so this does not look like a lock-order deadlock between the two threads. The reply, or the 25 s timeout, is simply never delivered to the waiting pending call. Knock-on effects while hung: every notification client blocked ("Failed to notify ... Did not receive a reply" from kded6, drkonqi, DiscoverNotifier), notify-send calls from system services hung until killed, and after ~1.5 days dbus-broker started disconnecting peers ("UID 1000 exceeded its 'bytes' quota" / "does not have the resources to receive a signal it subscribed to"), which took down akonadi, xdg-document-portal and flatpak-session-helper. Workaround in use: a user timer that calls org.freedesktop.Notifications.GetServerInformation every 2 minutes and restarts plasma-plasmashell.service after 3 consecutive timeouts. I can attach the full backtrace of all threads if useful. -- You are receiving this mail because: You are watching all bug changes.
