*** If you're not sure this is actually a bug, instead post about it at https://discuss.kde.org If you're reporting a crash, attach a backtrace with debug symbols; see https://community.kde.org/Guidelines_and_HOWTOs/Debugging/How_to_create_useful_crash_reports Please remove this comment after reading and before submitting - thanks! *** SUMMARY In the arch update 6.4.4-4, the change at https://invent.kde.org/plasma/kwin/-/commit/7d36003c.patch was backported. As a result of this change, if you have a dGPU that takes a moment to wake from d3cold/d3hot, opening windows will take considerable time if in this state. This is seen on the Framework 16 with the dedicated AMD GPU bay. While windows are not spawned on the dGPU, it still waits for the wakeup. Calling `echo on | sudo tee /sys/bus/pci/devices/0000:03:00.0/power/control` completely stops the hangs, by forcing the gpu to never reach D3cold (auto being the default). This can also be further tuned by writing to `/sys/bus/pci/devices/0000:03:00.0/power/autosuspend_delay_ms` which affects how long it takes to hit a low power state. STEPS TO REPRODUCE 1. Have a machine with an AMD GPU that correctly goes into d3cold 2. Open a new window for any program, see a delay as it starts up 3. Start another before it hits cold, see window opens immediately. 4. Wait to hit d3 again, try again, delay. 5. Turn off autosuspend, see the problem go away. OBSERVED RESULT Delay in window appearing EXPECTED RESULT Window appears immediately, as it did before this patch SOFTWARE/OS VERSIONS Operating System: Arch Linux KDE Plasma Version: 6.4.4 KDE Frameworks Version: 6.17.0 Qt Version: 6.9.1 Kernel Version: 6.16.3-zen1-1-zen (64-bit) Graphics Platform: Wayland Processors: 16 × AMD Ryzen 7 7840HS w/ Radeon 780M Graphics Memory: 96 GiB of RAM (86.2 GiB usable) Graphics Processor 1: AMD Radeon RX 7700S Graphics Processor 2: AMD Radeon 780M Graphics ADDITIONAL INFORMATION Machine is a Framework 16. State when suspended: ``` ❯ cat /sys/bus/pci/devices/0000:03:00.0/power_state D3cold ```
Oops, can I not edit the comment here? Forgot to remove the header, apologies.
That commit is definitely not related to the issue. Did you try downgrading that one package and testing the downgraded version?
I did try downgrading just now to the "known good" version I had before - seems it persists. I wonder what the issue is, but it seems it's not in kwin then. My bad, apologies!
I'd recommend checking Mesa and the kernel, it's almost certainly one of these two.