https://bugs.kde.org/show_bug.cgi?id=527314
Raayed <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- Attachment #197376|0 |1 is obsolete| | --- Comment #5 from Raayed <[email protected]> --- Created attachment 197532 --> https://bugs.kde.org/attachment.cgi?id=197532&action=edit Better backtraces for this crash. This crash has been difficult to reproduce, but I've found that killing/restarting plasmashell helps. This backtrace is still missing a few debug symbols, but I've checked this time that there are enough to make it useful, unlike last time. I tried to install all the debug symbols needed, but some debug info is not there on the servers on Arch's side as of when I did this. I do use an LLM (Claude) to do troubleshooting like this, so it has some extra info regarding the traces and the few symbols that are still missing that could be useful. LLM's stuff: Backtrace attached (plasma-bt-full.txt: thread 1 with bt full, plus thread apply all bt), captured from a core file of the hung plasmashell with symbols from debuginfod.archlinux.org. Frames in Qt Core, Qt Widgets and Kirigami are fully symbolized. libQt6Qml, libQt6Quick and several other Qt Declarative libraries (QmlMeta, QmlModels, QuickControls2, QuickLayouts, etc.) still have no debug info, so frame #0 (where the main thread is spinning), frame #1 and frames from #12 onward show only library names. I checked that this is on the server side: with an empty debuginfod cache, debuginfod-find downloaded the Qt Core debug info from https://debuginfod.archlinux.org/ without problems, but a lookup for libQt6Qml's build ID (587adbdbee6487acdfde7fababce61d8212d5baf, qt6-declarative 6.12.0-1) fails with "Server query failed: No such file or directory". The same goes for libQt6Quick. The stack shows Kirigami::Platform::PlatformTheme::update() sending an event to itself, then PlatformThemeChangeTracker's destructor emitting the theme changes (changes=127) into QML, triggered by a QQuickTextEdit being attached to a window during layout. I can't tell whether that's the loop or only where it was sitting when I captured it. If the missing frames matter, tell me what you'd like me to try (for example a specific debug package or repository) and I'll capture again. I still have the core file and can reproduce the hang, though it's intermittent. -- You are receiving this mail because: You are watching all bug changes.
