https://bugs.kde.org/show_bug.cgi?id=527823
--- Comment #2 from [email protected] --- The description above was taken from an early draft and contains incorrect numbers. Corrected facts: Crash count: 4 total, all with an identical symbolized stack. 2026-10-19 03:45:34 kwin 6.8.0-1 popFramebuffer +0x319884 2026-10-19 20:12:41 kwin 6.8.0-1 popFramebuffer +0x319884 2026-10-19 20:13:42 kwin 6.8.0-1 popFramebuffer +0x319884 2026-10-22 04:18:59 kwin 6.8.1-1 popFramebuffer +0x31a704 So 6.8.0 crashed three times and 6.8.1 once, not 1 + 2. No crashes on 6.7.5 during ~1 month of the same workload. Distro: EndeavourOS (Arch-based, packages from Arch repos), not plain Arch. Environment: Mesa 26.2.5, Qt 6.12.0, KF6 6.31.0, kernel 7.2.11-zen1, Intel Iris Xe (8086:9a49), OBS Studio (Flatpak). Stack trace of the crashing thread (systemd-coredump): #0 KWin::EglContext::popFramebuffer() #1 KWin::ItemRendererOpenGL::endFrame() #2 KWin::ItemTreeView::paint(RenderTarget const&, QPoint const&, Region const&) #3 KWin::Application::cursorImage() const #4 KWin::ScreenCastStream::addCursorMetadata(spa_buffer*, KWin::Cursor*) #5 KWin::ScreenCastStream::record(std::chrono::nanoseconds, QFlags<ScreenCastStream::Content>) #8 KWin::VsyncMonitor::vblankOccurred(std::chrono::nanoseconds) Signal 11 (SEGV), si_code SEGV_MAPERR, fault address 0x1. The symbol "VsyncSource" in the description should read "VsyncMonitor"; grabCursorOpenGL is inlined and has no separate frame. Journal checks: no "Could not make egl context current" and no GPU reset/hang line before any of the four crashes. All details are in the attachment kwin-screencast-crash-attachment.txt. -- You are receiving this mail because: You are watching all bug changes.
