I downgraded and it seems to be working for now but will report back. As reference for others that find this thread, I ran the following commands to downgrade mesa and versionlock it so it doesn’t get upgraded by accident until the bug in mesa gets fixed
sudo dnf downgrade 'mesa*'
sudo dnf versionlock add 'mesa*'
To unblock when bug in mesa eventually gets fixed:
sudo dnf versionlock delete 'mesa*'
It seems with FW there is now a fix that looks like it does fully eliminate these glitches, so only the VRR-related glitches seem to be left over: [amdgpu] Heavy artifacting in desktop applications since bump from 24.3.4 to 25.0.1 (#12809) · Issues · Mesa / mesa · GitLab
Install Linux 6.14 with Mesa 25.0.2 (no idea if that’s specific to that 25.x version) and in the BIOS settings, set the video memory to the game optimized 4 GB option.
I still had issues with crashing in gaming mode. It was a bit different, with everything just locking up.
Update: it has been working well for me since downgrading
I’m guessing you’re experiencing a different issue, crashing or freezing was never part of this.
I agree completely but downgrading Mesa also fixes it.
Mesa downgrading doesn’t work for me.
On my Fedora 41 I have plug/unplug sound every 3 minutes after these steps.
sudo dnf downgrade 'mesa*'
Actualizando y cargando repositorios:
Repositorios cargados.
The lowest available version of the "mesa-dri-drivers.x86_64" package is already installed, cannot downgrade it.
The lowest available version of the "mesa-filesystem.x86_64" package is already installed, cannot downgrade it.
The lowest available version of the "mesa-libEGL.x86_64" package is already installed, cannot downgrade it.
The lowest available version of the "mesa-libGL.x86_64" package is already installed, cannot downgrade it.
The lowest available version of the "mesa-libGLU.x86_64" package is already installed, cannot downgrade it.
The lowest available version of the "mesa-libgbm.x86_64" package is already installed, cannot downgrade it.
The lowest available version of the "mesa-libglapi.x86_64" package is already installed, cannot downgrade it.
The lowest available version of the "mesa-va-drivers-freeworld.x86_64" package is already installed, cannot downgrade it.
The lowest available version of the "mesa-vdpau-drivers-freeworld.x86_64" package is already installed, cannot downgrade it.
The lowest available version of the "mesa-vulkan-drivers.x86_64" package is already installed, cannot downgrade it.
The mesa downgrade helps with the pixels on the screen. The plug/unplug sound is I think unrelated cause it’s not helped by the downgrade and also it appeared more recently (several weeks after the pixel issue). I noticed that the USB peripherals connected through my external monitor, actually get disconnected for a very brief moment of time (which would explain this sound but don’t know what’s causing it).
For folks troubleshooting based on this thread:
… so don’t be surprised if they don’t work for you either. No solution presented on this thread has worked for everyone consistently. The base problem is that the Linux video stack simply doesn’t support amdgpu, and usability is just going to get worse.
In addition to the pixel art, the other issues I’m having include my Framework not recognizing external displays without a reboot. Since I plug & unplug my laptop from an external display more than once a day, this is unusable.
Well, for me at least this did the trick:
sudo dnf downgrade 'mesa*'
Absolutely no problem since.
And I plug and unplug multiple times per day external monitors, works (nearly) like a charm on Fedora with the Framework 13 AMD. When I say nearly: sometimes it picks up the wrong resolution (I’d say one on ten times). I then unplug it, replug it, and picks up the right one. But no need to reboot.
This is not true.
Yes there is “a lot” of regressions in the amdgpu kernel driver and then add the mesa issues with the Mesa 25 branch.
Downgrading to mesa 24.3 did not work for me either, still got Kernel Warnings, graphical glitches, and then ultimately system lockup on ring gfx timeout.
Tried kernels from 6.6 and up to 6.14 aswell as suggested kernel params
Still couldn’t reliably find a fix to the issue(s), I am interested in helping debug the issue(s) if possible
After upgrading to the latest 6.14 the issue seems to have disappeared for me (Archlinux)
Inffy,
The last five months of continuous video problems I’ve had would argue that you’re wrong about this. I have literally tried every single recommended fix and none of them have worked. At this point, I’m looking at getting a new laptop.
Beyond community support, have you considered opening a support ticket directly with Framework?
Tim,
I have. It was not a good experience or productive.
If none of the fixes worked for you, it sounds very much like defective hardware. Because with the debugmask in place, I don’t see any issue whatsoever. Not the glitchy lines at the bottom of the display and not the checkerboard glitches. And the same seems to be true for at least most people if not everyone expect you.
Yeah it’s still been working flawlessly for me for the past month since downgrading mesa and it sounds like that was the same issue for at least some people on the thread. For now I’m refusing to update any package with the word ‘mesa’ in it but I guess we should keep an eye out for fixes made to mesa that address this. Does anyone know if there’s an open bug on the mesa project or know if they’ve made any changes to fix this yet?
Does kernel 6.14.4 help?
This might have helped:
commit 56daad28d6c6e1157ad9021d3dab50e733f22f58
commit faa91baf373547e1509ffe2425acb7a4e51ecd34
commit 1ed4287bb452eeeb148d6ceda45aefeb2befb27f
I’ll try that out, but it’s quite stressing finding that out while the other glitch (the kinda bands that glitch over the whole width of the screen due to issues with PSR) is probably still present and still extremely prone.