Kde is fine, the bug is relatively rare and I do not have a definitive way to reproduce. It seems, on a coincidence of 2 times that I can attest (others being forgotten) that clicking on a grouped sest of windows in the task manager can lead to a runaway plasmashell main thread (other slave threads to not run away). I have a core dump and I have several backtraces. Sometimes I catch a dbus request being made as evidenced by a dbus connection thread, but I generally just catch main in: #0 0x00007fd404d892b5 in QV4::PersistentValueStorage::allocate() () at /usr/lib64/libQt5Qml.so.5 #1 0x00007fd404d898f2 in QV4::PersistentValue::set(QV4::ExecutionEngine*, QV4::Value const&) () at /usr/lib64/libQt5Qml.so.5 #2 0x00007fd404e0027a in QV4::QObjectWrapper::method_connect(QV4::CallContext*) () at /usr/lib64/libQt5Qml.so.5 #3 0x00007fd404dae32b in QV4::BuiltinFunction::call(QV4::Managed const*, QV4::CallData*) () at /usr/lib64/libQt5Qml.so.5 #4 0x00007fd404e1baeb in QV4::Runtime::callProperty(QV4::ExecutionEngine*, int, QV4::CallData*) () at /usr/lib64/libQt5Qml.so.5 Sometimes I catch main in a QML garbage collection. All other threads are waiting on the poll sys call and not using CPU.
Created attachment 102712 [details] btfull
compositor is disabled The 2 instances it has crashed I mentioned is once on selecting one in a group of konsole windows. I'm unsure if the other instances (probably on the order of 10 crashes to date) had other circumstances - wasn't paying attention or forgot as time goes on. The 2nd is when I selected 1 in a group of ROX-Filer windows. I did not see the tooltip displayed in this latest case, with ROX-Filer. I'm not sure about prior cases. Each time, nothing out of the ordinary was going on - I was just trying to select another window from likely multitasking with chrome. That is no popup windows/dialogs/drag and drops were happening. Memory normally starts up somewhere in 120-180MiB, a couple of hours ago, as it's ran for a few days now, was at 300-400MiB. It is presently 1.6GiB and will grow fast after I unfreeze it. Bug has OOM'd me before if I don't kill it or stop it in time.
fresh start of plasmashell - 134MiB. Not a crash, just a reference from normal operation - started with kstart5 plasmashell and then I attached gdb afterwards. Notice less threads - absense of threaded-ml for instance. And that main landed in poll. * 1 Thread 0x7f7ab3af6900 (LWP 13389) "plasmashell" 0x00007f7aad2b149d in poll () from /lib64/libc.so.6 2 Thread 0x7f7aa05fe700 (LWP 13392) "QDBusConnection" 0x00007f7aad2b149d in poll () from /lib64/libc.so.6 3 Thread 0x7f7a9778c700 (LWP 13393) "QXcbEventReader" 0x00007f7aad2b149d in poll () from /lib64/libc.so.6 4 Thread 0x7f7a8ffff700 (LWP 13394) "QQmlThread" 0x00007f7aad2b149d in poll () from /lib64/libc.so.6 5 Thread 0x7f7a8ebbe700 (LWP 13395) "QQmlThread" 0x00007f7aad2b149d in poll () from /lib64/libc.so.6 6 Thread 0x7f7a8cdee700 (LWP 13396) "plasmashell" 0x00007f7aacabf0af in pthread_cond_wait@@GLIBC_2.3.2 () from /lib64/libpthread.so.0 7 Thread 0x7f79f7732700 (LWP 13397) "QQuickPixmapRea" 0x00007f7aad2b149d in poll () from /lib64/libc.so.6 8 Thread 0x7f79f5e7c700 (LWP 13398) "QSGRenderThread" 0x00007f7aacabf0af in pthread_cond_wait@@GLIBC_2.3.2 () from /lib64/libpthread.so.0 9 Thread 0x7f79edffe700 (LWP 13399) "QSGRenderThread" 0x00007f7aacabf0af in pthread_cond_wait@@GLIBC_2.3.2 () from /lib64/libpthread.so.0 10 Thread 0x7f79d513a700 (LWP 13400) "KCupsConnection" 0x00007f7aad2b149d in poll () from /lib64/libc.so.6 11 Thread 0x7f79cf2d1700 (LWP 13406) "QSGRenderThread" 0x00007f7aacabf0af in pthread_cond_wait@@GLIBC_2.3.2 () from /lib64/libpthread.so.0 12 Thread 0x7f79ce171700 (LWP 13407) "QSGRenderThread" 0x00007f7aacabf0af in pthread_cond_wait@@GLIBC_2.3.2 () from /lib64/libpthread.so.0 13 Thread 0x7f79c7fff700 (LWP 13408) "QSGRenderThread" 0x00007f7aacabf0af in pthread_cond_wait@@GLIBC_2.3.2 () from /lib64/libpthread.so.0
Thanks, but I'm afraid those traces don't really help so I'm going to close this, but please see the linked reports. We have multiple high CPU bugs, some caused by us, some not, some fixed, some unknown. It's hard to say what the cause of each one is, and bug reports get convuluted. I've made a list of some of the bigger more specific ones: For high CPU caused by reloading icons over the SNI protocols see: 356479 For high CPU caused by infinite spurious X events see: 373368 For others, search other bug reports here. Hopefully your cause will be fixed in the next or future releases.