https://bugs.kde.org/show_bug.cgi?id=526339
--- Comment #2 from [email protected] <[email protected]> --- I reproduced the issue again with DRM debugging enabled, this time using a much shorter DPMS timeout: * screen dimming after 10 seconds * display power-off after 30 seconds * external HDMI-A-1 display at 1920x1080@60 * Wayland / KWin 6.6.6 / Intel i915 The attached DRM debug log contains a particularly relevant sequence. At approximately 6448.641 seconds, the DRM atomic check reports: `[CRTC:110:pipe A] requires full modeset` Immediately afterwards, the DRM ioctl issued by `kwin_wayland` returns: `ret=-22` The log then shows further atomic operations involving CRTC 110 / HDMI-A-1, followed by HDMI hotplug activity. A few hundred milliseconds later, i915 reports an HPD event and HDMI-A-1 changes state. The log contains repeated connector state changes and subsequent atomic modeset activity. There are also GMBUS/HDMI detection messages during this sequence, including NAK responses while probing the HDMI connection. Importantly, I do not see an i915 GPU reset/hang in this capture. This seems to provide a kernel DRM trace corresponding to the `Atomic modeset test failed! Invalid argument` message reported by KWin in the original reproduction. I am attaching the complete DRM debug log from this reproduction. Could you please take a look at the sequence around the `ret=-22` / `requires full modeset` event and the following HDMI HPD/state changes? -- You are receiving this mail because: You are watching all bug changes.
