Created attachment 137892 [details] `bt -full` output of the crash dump SUMMARY Kwin_wayland segfaults occasionally when right-click on the title bar of a window. I remenber that occured with Firefox(xwayland) and Zim(gtk wayland). The backtrace may be different for different ocurrance. I capture this one that crashed with Zim. STEPS TO REPRODUCE 1. 2. 3. OBSERVED RESULT kwin_wayland segfaults. EXPECTED RESULT kwin_wayland doesn't crash. SOFTWARE/OS VERSIONS Windows: macOS: Linux/KDE Plasma: Fedora 34 x86_64 (available in About System) KDE Plasma Version: 5.21.3 KDE Frameworks Version: 5.80.0 Qt Version: 5.15.2 ADDITIONAL INFORMATION
#0 0x00007effb4efcc7b in __memmove_avx_unaligned_erms () from /lib64/libc.so.6 No symbol table info available. #1 0x0000564c6d4461a6 in std::__copy_move<false, true, std::random_access_iterator_tag>::__copy_m<unsigned int> (__result=<optimized out>, __last=0x7eff895db290, __first=0x7eff895daa10) at /usr/include/c++/11/bits/stl_algobase.h:431 _Num = 544 _Num = <optimized out> #2 std::__copy_move_a2<false, unsigned int const*, unsigned int*> (__result=<optimized out>, __last=0x7eff895db290, __first=0x7eff895daa10) at /usr/include/c++/11/bits/stl_algobase.h:495 No locals. #3 std::__copy_move_a1<false, unsigned int const*, unsigned int*> (__result=<optimized out>, __last=0x7eff895db290, __first=0x7eff895daa10) at /usr/include/c++/11/bits/stl_algobase.h:522 No locals. #4 std::__copy_move_a<false, unsigned int const*, unsigned int*> (__result=<optimized out>, __last=0x7eff895db290, __first=0x7eff895daa10) at /usr/include/c++/11/bits/stl_algobase.h:529 No locals. #5 std::copy<unsigned int const*, unsigned int*> (__result=<optimized out>, __last=0x7eff895db290, __first=0x7eff895daa10) at /usr/include/c++/11/bits/stl_algobase.h:620 No locals. #6 KWin::QPA::blitImage (rect=..., target=..., source=...) at /usr/src/debug/kwin-5.21.4-1.fc34.x86_64/plugins/qpa/backingstore.cpp:65 in = 0x7eff895daa10 out = <optimized out> i = 168 devicePixelRatio = <optimized out> x = <optimized out> y = <optimized out> width = <optimized out> height = <optimized out> #7 KWin::QPA::blitImage (region=..., target=..., source=...) at /usr/src/debug/kwin-5.21.4-1.fc34.x86_64/plugins/qpa/backingstore.cpp:72 rect = @0x564c704afd60: {x1 = 0, y1 = 0, x2 = 271, y2 = 167} __for_range = @0x7fffa0a262d8: {d = 0x564c70475e40} __for_begin = 0x564c704afd60 __for_end = 0x564c704afd70 #8 KWin::QPA::BackingStore::flush (this=0x564c6f37d850, window=<optimized out>, region=..., offset=...) at /usr/src/debug/kwin-5.21.4-1.fc34.x86_64/plugins/qpa/backingstore.cpp:86 platformWindow = <optimized out> client = 0x564c70389cb0 #9 0x00007effb6def9b7 in QBackingStore::flush(QRegion const&, QWindow*, QPoint const&) () from /lib64/libQt5Gui.so.5 No symbol table info available. #10 0x00007effb588ef8d in QWidgetRepaintManager::flush(QWidget*, QRegion const&, QPlatformTextureList*) () from /lib64/libQt5Widgets.so.5 No symbol table info available. #11 0x00007effb589099b in QWidgetRepaintManager::flush() () from /lib64/libQt5Widgets.so.5
Unfortunately, the backtrace has a lot of important details missing, e.g. devicePixelRatio = <optimized out> x = <optimized out> y = <optimized out> width = <optimized out> height = <optimized out>
i = 168 devicePixelRatio = <optimized out> if you don't have a hidpi monitor, it does look like a compiler bug
(In reply to Vlad Zahorodnii from comment #3) > i = 168 > devicePixelRatio = <optimized out> > > if you don't have a hidpi monitor, it does look like a compiler bug I was using two monitors, an internal one with 1920x1080 resolution and 125% DPI scale and an external one with 3840x2160 resolution and 200% DPI scale.
There were some decoration rendering fixes, reopen this bug report if the issue is still reproducible in 5.26 or 5.27