Hello!
I have a problem what is maybe a bug in the amdgpu driver related to DisplayPort MST. The thing is that I could not really identify the cause of the issue. Thus, I’m posting it here in the search of help. And also Plasma behaves strangely in my case (so maybe also something to report as bug for Plasma and co).
What is the Problem?
In my company we just got new monitor setups. We switch from having docks to just to monitors, which are daisy chained by DisplayPort. The main monitor is connected to with USB-C. Initially, connecting the monitors resulted in what seemed to be a freeze of Plasma, but the output was just really, really slow. When disconnecting the USB-C cable or also disconnecting the second (daisy chained) monitor, Plasma recovered and continued normally, without actually crashing. The monitors itself worked, as I could switch into a terminal (e.g. Ctrl + Alt + F3) and all 3 monitors (including the laptop display) would show the output.
Things I found out so far
I tried to look into the log with journalctl, but I could not really isolate the what the source of the problem is. The only thing I could find, is that kwin repeatedly spams
kwin_wayland[22297]: Atomic modeset test failed! No space left on device
with some
kwin_wayland[22297]: Applying output configuration failed!
in between, when in the “pseudo freeze” state. I could not really relate other messages to the issue, as I’m not that proficient in searching for such issues.
My workaround and consisting problems
One of my co-workers identified that the problem may be a bandwidth problem, as he managed to set both external monitors to 60 Hz (they are 75 Hz by default) and it works like this. With an additional HDMI cable, I also managed to set both monitors to 60 Hz and connected them by daisy chain again. As Plasma remembers the settings, it now also works without the extra cable, using DisplayPort MST again.
One problem that still consists: SDDM is not working like this. When rebooting the device, SDDM is repeatedly sending the same message as kwin before
sddm-helper-start-wayland[1821]: "kwin_wayland_drm: Atomic modeset test failed! No space left on device\n"
and the screen is black, until I disconnect the external monitors. A few seconds after which SDDM appears, I can login as normal and re-plug the monitors as soon as I am logged in. I assume SDDM again tries to use the monitors with 75 Hz and fails to do so, but I could not confirm this theory.
Other strange behavior
In the process of trying to get both monitors running, I also stumbled across something strange in the System Settings → Display Configuration. I had the case that I tried setting up the monitors, but I ended up with the menu telling me “Gaps between displays are not supported”, while not displaying any monitors to drag around. I found that there was one of the setups with an x-value of 2147481087 in ~/.config/kwinoutputconfig.json. The number looks familiar, but it is not INT_MAX, it is actually (INT_MAX - 2560). With 2560 being the width of the monitor. But I have no idea, why Plasma would put the monitor on the very edge of what is possible.
Sadly, I do not have any screenshots of this or snippets of the config file. And I was not able to reproduce this so I can take screenshots, etc.
System/Setup specs
Operating System: Fedora Linux 43
KDE Plasma Version: 6.5.4
KDE Frameworks Version: 6.21.0
Qt Version: 6.10.1
Kernel Version: 6.17.11-300.fc43.x86_64 (64-bit)
Graphics Platform: Wayland
Processors: 16 × AMD Ryzen 7 8845HS w/ Radeon 780M Graphics
Memory: 64 GiB of RAM (60,6 GiB usable)
Graphics Processor: AMD Radeon 780M Graphics
Manufacturer: TUXEDO
Product Name: TUXEDO InfinityBook Pro AMD Gen9
Monitors: Lenovo P27h-30
I’m happy to provide any additional logs, test things with the monitor setup if it helps or provide additional information, if needed.