https://bugs.kde.org/show_bug.cgi?id=523065
Rob Kerns <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDSINFO |REPORTED Resolution|WAITINGFORINFO |--- --- Comment #5 from Rob Kerns <[email protected]> --- I attempted to reproduce the crash after installing the available updates and restoring NoMachine’s EGL preload. Current test environment: Fedora 44 KDE Plasma on Wayland Plasma Workspace 6.7.3 KPipeWire 6.7.3 KWin 6.7.3 Spectacle 6.7.3 NoMachine 9.7.3 EnableEGLCapture 1 LD_PRELOAD=/usr/NX/lib/libnxegl.so present in both the Plasma shell environment and the user systemd environment No active NoMachine viewer/client session I repeatedly exercised the activities I remembered doing around the time of the original crashes, including opening, closing, cancelling, and taking multiple types of screenshots with Spectacle. I was unable to reproduce the plasmashell crash. Several of the original crashes also occurred during ordinary local desktop use when the NoMachine viewer/client was not open. I was able to obtain the exact matching debug file for the original Fedora kpipewire-6.7.2-1.fc44 build using its build ID: f257da54fd91a2629a3272b4ef4904676150364a Resolving the three KPipeWire offsets from the original backtrace produced: 0x0000000000004f39 QScopedPointer<PipeWireSourceStreamPrivate, QScopedPointerDeleter<PipeWireSourceStreamPrivate> >::~QScopedPointer() inlined at /usr/src/debug/kpipewire-6.7.2-1.fc44.x86_64/src/pipewiresourcestream.cpp:438:1 in PipeWireSourceStream::~PipeWireSourceStream() /usr/include/qt6/QtCore/qscopedpointer.h:81:12 0x00000000000051c7 PipeWireSourceStream::~PipeWireSourceStream() inlined at /usr/include/c++/16/bits/unique_ptr.h:92:2 in PipeWireSourceItem::~PipeWireSourceItem() /usr/src/debug/kpipewire-6.7.2-1.fc44.x86_64/src/pipewiresourcestream.cpp:438:1 0x0000000000012ab1 QQmlPrivate::QQmlElement<PipeWireSourceItem>::~QQmlElement() [clone .lto_priv.0] /usr/include/qt6/QtQml/qqmlprivate.h:119:31 I could not obtain the requested new fully symbolicated backtrace because the crash did not recur. There are two significant differences between the original crashes and this reproduction attempt: 1. Plasma and KPipeWire were updated from 6.7.2 to 6.7.3. 2. The original crashes occurred with an AMD Radeon RX 5700 XT. The current system now has an AMD Radeon RX 9070 XT. I have restored the stable configuration by disabling NoMachine EGL capture again. The global LD_PRELOAD and active Plasma environment hook are no longer present. -- You are receiving this mail because: You are watching all bug changes.
