My mistake. I was getting the same error message but it was for Flatpak apps using the GNOME 47 runtime. I’m on F40 with KDE Plasma, and my laptop is an AMD + Nvidia system. The above workarounds do work for me, but are imperfect on displays using fractional scaling. I hope this issue gets solved soon.
thanks worked
I’ve implemented the workaround, and it does address the issue for now. Nevertheless, it’s frustrating that such a workaround is necessary. These temporary fixes add mental overhead as we have to track and remove them once an official fix is available. How are you all managing this?
While it works, it seems like everything’s slower now. Not unexpected though, and I’m just happy that I can update stuff again. (This feels like a release blocker. I’m hoping that the QA team notices.)
Unfortunately, the bugzilla ticket for this issue has been closed and tagged as CANTFIX, as apparently it’s a bug in the NVIDIA proprietary driver’s explicit sync implementation for versions 560+, but has already been fixed for the next driver series so just gonna have to wait for that.
1 Like
I have the same problem (`bluefin-dx-nvidia:stable` : can't open some native apps anymore - Bluefin and Aurora - Universal Blue)
I am on bluefin-dx-stable:stable (Universal Blue image on top of Fedora Silverblue) which is on Fedora 40 as of now
Is ~/.config/environment.d/gsk.conf a new file in Fedora 41 ?
1 Like
Does that mean the same bug could be happening on Fedora 40 ?
Yes it’s a new file, you need to create the file and folder by yourself
Even on Fedora 40 ?
I have found no one else saying that this bug is happening on Fedora 40 yet
I’m not sure about the fedora 40
Patch (fix)
Comand
This command fixes this issue (Fedora 40 + 41) :
mkdir -p `~/.config/environment.d/`
echo "GSK_RENDERER=gl" >> ~/.config/environment.d/gsk.conf
The system might need to reboot to load that config file properly :
systemctl reboot
Revert
If you ever need to revert it back :
sed -i '/GSK_RENDERER=gl/d' ~/.config/environment.d/gsk.conf
1 Like
Can confirm that this occurs on fedora 40, for myself and other secureblue users.
Your workaround fixes the issue and also fixes the issue I’ve been experiencing for months with external monitors being completely broken. If there are any ongoing threads about external monitors breaking on nvidia optimus laptops, this thread and the fix may be relevant.
1 Like
I’m trying to package the patch but I do not succeed at all
If someone can help me :
https://copr.fedorainfracloud.org/coprs/malix-off/gsk-renderer-gl-patch/
I maintain a whole website full of temp fixes 
and just un-do em occasionally.
I haven’t experienced this specific issue (don’t use NVIDIA nor Flatpaks), but at a guess on how I’d troubleshoot it:
- If I saw that dispatching to Wayland error or had any other reason to suspect Wayland, I’d be on Xorg (been using it F40-older)
- I heard about
GSK_RENDERERchanges and manually forcedvulkanF40/F41 beta; I later switched to gl after hearingvulkanwasn’t ideal (not fully-optimized yet, missing stuff on Xorg, andglwas described more-simple and easier on low-end HW; basicallyglsounded best andngldidn’t sound exciting/worth potentially debugging) - If I later saw this thread, I’d be feeling pretty good about choosing Xorg/GSK
gl
Thanks for sharing this, this did the trick for me. For anyone searching, I was encountering this error with Mission Control and Bottles, under Fedora 40 KDE. I have a hybrid laptop (AMD Ryzen + nvidia with proprietary drivers).
Also FYI, the above commands didn’t quite work. mkdir errored out because of the ` characters, and echo wouldn’t write, but I just used nano to add that line after creating the file and rebooted. But I’m not complaining since this is working now!
1 Like
Indeed !
I copy pasted it too fast, thanks for the catch
(Can’t edit though)
1 Like
That’s ok, I appreciate you sharing and it did fix the issue!
Also I think someone else pointed this out in this thread, but it will make your mouse cursor huge when hovering over the apps in question if you’re on a high DPI display (assuming you’re not at 100% scale). A small price to pay for now!
Yes, this just started happening on Bluefin 40 NVIDIA on my Acer Nitro a day or so ago! So technically it is an NVIDIA bug that propagated into the RPM Fusion repositories and thence to Universal Blue’s build system, which didn’t detect it.
When it started happening, I rebased from Bluefin 40 NVIDIA to Fedora 41 Silverblue, which fixed it. When I then installed the NVIDIA drivers, it came back. I found this message stream and applied the workaround.
If I get a chance I’ll bring this up in the Universal Blue forums; they just released beta images of Bluefin 41. But I’m staying with Silverblue 41 for now and the next month or so; I have a couple of deadlines and don’t have much time to beta test NVIDIA configurations. If I get any more issues I’ll take out the NVIDIA drivers; nouveau is good enough on my GTX 1650 Ti for now.
I have already done it I think.
I’m probably the most active member in their discourse too so I’ll try to spread the news
I notified in their discord awhile back: Discord
I’ve also added it to the secureblue FAQ: secureblue/FAQ.md at live · secureblue/secureblue · GitHub
nouveauis good enough on my GTX 1650 Ti for now.
Out of curiosity, are you able to use external monitors with Nouveau? Unfortunately for Turing Optimus GPUs, the only way to get proper power management (D3cold) is to use Nouveau. However, when I use external monitors with Nouveau, I get a white screen.