https://bugs.kde.org/show_bug.cgi?id=528002
--- Comment #8 from jajaX <[email protected]> --- (In reply to kfranko from comment #7) > Created attachment 198084 [details] > Horizontal bar display corruption on iGPU monitors Thank you for the suggestions and feedback. I performed additional tests on my second affected laptop (ASUS, Intel HD Graphics 3000 + NVIDIA GF108M), running KDE Plasma 6.8.2 under Wayland. **Graphics configuration:** - Intel HD Graphics 3000: i915 driver, DRM Atomic Mode Setting enabled. - NVIDIA GF108M (Fermi): nouveau driver, DRM legacy mode. - HDMI-A-1 is physically connected to the NVIDIA GPU (`/dev/dri/card2`). - KWin uses OpenGL ES on the Intel GPU (Mesa 25.3.6). The following messages appear in the KWin startup logs: ``` drmSetClientCap for Atomic Mode Setting failed. Using legacy mode on GPU "/dev/dri/card2" Invalid framebuffer status: "GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT" ``` The framebuffer error appears only once during KWin startup. **Test of the suggested workaround:** I tested `KWIN_DISABLE_VULKAN=1` by creating an environment configuration file in `~/.config/environment.d/` and rebooting the laptop. Unfortunately, this did not resolve the issue. The HDMI output still displayed severe image corruption, including a strong red tint and horizontal artifacts. During these tests, RustDesk also stopped detecting the internal laptop display and only detected the HDMI display, even though the internal display remained active. This issue persisted after removing the environment variable and rebooting. **Additional observations regarding the HDMI output:** An important detail is that the desktop remains functional even when the physical HDMI display is black or severely corrupted. Windows can be moved normally from the internal laptop display to the affected HDMI display. Although the windows are not visible on the physical HDMI screen, I can still interact with them using the mouse. The mouse pointer changes shape depending on the elements underneath it (text fields, clickable links, window controls, etc.). Furthermore, screenshots taken with KDE Spectacle correctly capture the entire desktop, including the windows and their contents on the affected HDMI display. In other words, the desktop contents are correctly rendered in the screenshots, even though they are not displayed correctly on the physical HDMI monitor. A Spectacle screenshot demonstrating this behavior has already been attached to this bug report. These observations may suggest an issue with the final display output, scanout, or multi-GPU buffer handling rather than missing desktop contents. **Regarding the proprietary NVIDIA driver:** I have not tested the proprietary NVIDIA driver. The GF108M belongs to the Fermi generation, whose last supported proprietary driver branch was 390.xx. This legacy driver is no longer officially maintained and is not a practical option for my current kernel and Wayland environment. Both affected laptops use the open-source nouveau driver. HDMI output worked correctly under Wayland before Plasma 6.8.0, and the regression persists with Plasma 6.8.2. I can perform further diagnostic tests if needed. -- You are receiving this mail because: You are watching all bug changes.
