Bug 513812 - Desktop locks up or "The driver rejected the output configuration" whenever a 4th output is enabled from my GPU. Only happens on Wayland.
Summary: Desktop locks up or "The driver rejected the output configuration" whenever a...
Status: NEEDSINFO WAITINGFORINFO
Alias: None
Product: kwin
Classification: Plasma
Component: platform-drm (other bugs)
Version First Reported In: 6.5.3
Platform: Fedora RPMs Linux
: NOR normal
Target Milestone: ---
Assignee: KWin default assignee
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2025-12-25 19:07 UTC by emilia
Modified: 2026-01-07 20:51 UTC (History)
1 user (show)

See Also:
Latest Commit:
Version Fixed/Implemented In:
Sentry Crash Report:


Attachments
Combined logs from dmesg at boot, dmesg during the error and journalctl (2.32 MB, text/x-log)
2025-12-25 19:07 UTC, emilia
Details
After enabling display dmesg (2.21 MB, text/x-log)
2025-12-25 19:08 UTC, emilia
Details
After enabling display journalctl (63.24 KB, text/x-log)
2025-12-25 19:09 UTC, emilia
Details
Boot logs (omitting early boot stages that didn't seem relevant) (44.57 KB, text/x-log)
2025-12-25 19:09 UTC, emilia
Details

Note You need to log in before you can comment on or make changes to this bug.
Description emilia 2025-12-25 19:07:31 UTC
Created attachment 187978 [details]
Combined logs from dmesg at boot, dmesg during the error and journalctl

SUMMARY
Whenever I try to enable a 4th monitor from my GPU, it either hangs (through kscreen-doctor) until i blindly type the command to disable it again. When done through system settings, it just says "Driver rejected the output configuration".

It does not matter which monitor or on which output it is. I've tried every combination I could think of and 3 monitors always works fine, but the second I enable the 4th, the entire desktop graphically freezes (it's still running under that, I can up arrow, ctrl backspace to replace the .enable with .disable to get it fixed) or it just refuses to apply. It doesn't matter whether the 4th monitor is DP or HDMI, it's just when it gets to four.

I've tried it on Arch Linux, Kubuntu 25.10 and Fedora 43. Sadly I don't know the exact version of kwin where this broke, as I reinstalled instead of updating and didn't take note of what package versions I had installed previously... But I'm very sure I was already on Plasma 6.5. My educated guess is 6.5.3 but take that with a margin of error of 6.5.2 through 6.5.4.. Apologies for that. I've tried reverting to previous versions with little luck bcs of dependency errors, couldn't find every single RPM it wanted for dependencies. I've also tried this in a development environment with every KDE package compiled myself, but there were many more issues that got in the way first, so I couldn't capture the logs on there, but it did seem to exhibit the same general behavior.  


Only happens on Wayland. Xorg session works fine with all 4 (and the 2 on mobo, for a total of 6) monitors. But I have an HDR monitor and require many more Wayland features...

I've also tried the legacy modesetting (KWIN_DRM_NO_AMS) in kwin and the same behaviour occurs, although it just says modesetting failed and not atomic modeset..

This started happening around the start of this month, but I've just disabled my 4th GPU monitor for now.

STEPS TO REPRODUCE
1. Enable a 4th monitor from the GPU, either through system settings or kscreen-doctor


My logs were obtained through the following sequence of commands:

```
echo "BEGIN LOGS" | sudo tee /dev/kmsg && echo 0x1FF | sudo tee /sys/module/drm/parameters/debug && kscreen-doctor output.HDMI-A-1.enable && sleep 1 && kscreen-doctor output.HDMI-A-1.disable && sleep 0.5 && echo 0 | sudo tee /sys/module/drm/parameters/debug && echo "END LOGS" | sudo tee /dev/kmsg
```
My session was also started with the following in .config/plasma-workspace/env: (taken from https://invent.kde.org/plasma/kwin/-/wikis/Debugging/Debugging-DRM-issues)

export QT_LOGGING_RULES="kwin_*.debug=true"
export LC_ALL=C


With the logs then being obtained through `journalctl -xe --user  -f | grep --ignore-case --extended-regexp "kwin|kde|plasma"` and `sudo dmesg -w`. Since I don't see how to add multiple attachments I concatenated these logs (alongside the dmesg at boot) with:
```
echo "=== boot-dmesg.log: ===" > all.log; cat boot-dmesg.log >> all.log; echo "=== after-kscreen-journalctl.log: ===" >> all.log; cat after-kscreen-journalct.log >> all.log; echo "=== after-kscreen-dmesg.log: ===" >> all.log; cat after-kscreen-absurdly-long.log >> all.log
```

OBSERVED RESULT
"applying config failed! The driver rejected the output configuration", along with a lot of "Atomic modeset test failed! Invalid argument" in journalctl (without debug on this is all it shows)


EXPECTED RESULT
I can enable a 4th monitor.

SOFTWARE/OS VERSIONS

Linux/KDE Plasma: 6.17.12-300.fc43-.x86_64
KDE Plasma Version: 6.5.4-2
KDE Frameworks Version: 6.21.0
Qt Version: 6.10.1

ADDITIONAL INFORMATION
I've tested this on an Intel Arc B580 and an AMD Radeon 7900 XT, both show the exact same behaviour. 
I have two monitors plugged into my motherboards running on integrated graphics, but even without these the same happens. 

Also, importantly, this doesn't happen in other desktop environments, just KDE Plasma and just in Wayland. It also did seem to work in the Kubuntu 25.10 live image from what I recall, but not after installation (which would have newer package versions)

Thanks in advance! Please let me know if there's anything I can do to help provide more information or troubleshoot!
Comment 1 emilia 2025-12-25 19:08:50 UTC
Created attachment 187979 [details]
After enabling display dmesg
Comment 2 emilia 2025-12-25 19:09:06 UTC
Created attachment 187980 [details]
After enabling display journalctl
Comment 3 emilia 2025-12-25 19:09:41 UTC
Created attachment 187981 [details]
Boot logs (omitting early boot stages that didn't seem relevant)
Comment 4 emilia 2025-12-25 19:10:03 UTC
Noticed I could add more attachments after submitting so I've uploaded them separately as well!
Comment 5 Zamundaaa 2026-01-07 20:51:50 UTC
The driver is hitting some sort of resource limits:
> [  240.868304] amdgpu 0000:03:00.0: [drm] Adding stream 000000009e4d87f2 to context failed with err 2!
> [  240.868306] amdgpu 0000:03:00.0: [drm:amdgpu_dm_atomic_check [amdgpu]] ENABLE: dm_update_crtc_state() failed
> [  240.868448] amdgpu 0000:03:00.0: [drm:amdgpu_dm_atomic_check [amdgpu]] Atomic check failed with err: -22
"err 2" is DC_NO_CONTROLLER_RESOURCE, I've never seen that before though.

It also has another warning, which I'm more familiar with:
> DC global validation failure: Bandwidth validation failure (BW and Watermark) (13)
To reduce bandwidth requirements, you could try reducing resolution and refresh rate on one or more screens.
Just to be certain, also check if you're using the exact same modes on Xorg. KWin goes for the highest refresh rate by default, which can cause increased bandwidth requirements.

One more thing you can test: is your "color accuracy" setting on any of the screens set to "prefer color accuracy"? In theory, KWin should be doing automatic fallbacks, but it's worth trying with all screens set to "prefer efficiency" anyways.

Last but not least, the fastest way to find out what to do about this would be a bisect. If you're sure it worked earlier in 6.5, then you could compile KWin 6.5.0 and see if that can enable all screens with the same settings.