Unfortunately, this does not work with the mesa-freeworld drivers from rpmfusion. The necessary builds were deleted from their servers: mesa-freeworld | Package Info | koji
I got the same with fedora 41 and 42, mainly after resuming from suspend or hibernate.
Increasing video memory in bios (auto ==> gaming) seems to help.
Iām using 6.14.2-300.fc42.x86_64 and mesa-freeworld
The fc41 freeworld packages from rpmfusion appear to work just fine on Fedora 42.
I combined @jkbachās instructions with some others I found and ended up doing the following:
mkdir temp-mesa-fix
cd temp-mesa-fix
koji download-build --arch=x86_64 mesa-24.3.4-1.fc42
koji -s https://koji.rpmfusion.org/kojihub --topurl=https://koji.rpmfusion.org/kojifiles download-build --arch=x86_64 mesa-freeworld-24.3.4-8.fc41
createrepo_c .
sudo dnf --repofrompath localmesa,. install mesa-vdpau-drivers-freeworld-24.3.4-8.fc41.x86_64 mesa-va-drivers-freeworld-24.3.4-8.fc41.x86_64 --no-gpgchecks
sudo dnf versionlock add mesa-va-drivers
sudo reboot
I used the download folder as a local repo, however sudo dnf downgrade *.rpm probably would have worked too.
I only did this a little while ago so I canāt say for sure whether it actually helped.
Next day update:
This did help significantly. I still get some glitchiness in the workspace switching animations, but thatās the only place Iāve noticed any issues since downgrading to the older mesa freeworld packages.
Thank you all, especially @jkbach, for your downgrade workaround for Fedora 42.
To summarize information for folks finding this thread:
-
If youāre on Fedora 41, you can downgrade to v24
mesa.
sudo dnf downgrade mesa-va-drivers-0:24.2.4-1.fc41.x86_64You can then use the version lock plugin to prevent upgrading to the faulty version, per reply #20.
sudo dnf versionlock add mesa-va-drivers -
If youāre on Fedora 42, stuck with v25 and unable to downgrade from the repositories, you can use koji. See the above reply #39.
-
For those of you waiting on a fix, per the
mesaissue (Gitlab #12809), it would seem to be fixed in v25.0.4 - as of yet not released on Fedora 42 See Fedora Packages. -
Setting the kernel parameter
amdgpu.sg_display=0per the above reply #27 or #33 did not resolve the issue for me, though some individuals have found it helpful.
2 Likes
Thanks for the tip, I havenāt seen any glitches in over a day after doing this. Not sure if the problem is totally gone but itās definitely a lot better. Now Iāve probably jinxed it and my screen will go nuts right after posting this 
Mesa 25.0.4 is now available on the stable release channel for Fedora 42.
2 Likes
I can say that it is not a perfect solution: it helps, but I still have some artefacts sometimes with firefox.
But it is much less frequent!
Unfortunately, it does not seem to have fixed the problem for me. Before seeing this thread I updated to Fedora 42 and brought my packages up to date, and still experience this problem.
| DNF Package | Version Installed |
|---|---|
| kernel.x86_64 | 6.13.9-200.fc41 |
| amd-gpu-firmware.noarch | 20250410-1.fc42 |
| amd-ucode-firmware.noarch | 20250410-1.fc42 |
| mesa-va-drivers.x86_64 | 25.0.4-2.fc42 |
Here are some screenshots. It seems highly sensitive to cursor movement, and can sometimes get quite extreme (I think the latter image was supposed to be of a Terminal).
1 Like
This is helpful, thank you.
Myself, I am on FW 13 AMD Ryzen 7040 Series, Fedora 42, BIOS 3.07:
| DNF Package | Version Installed |
|---|---|
| 6.14.4-300.fc42.x86_64 | |
| amd-gpu-firmware-20250410-1.fc42.noarch | |
| amd-ucode-firmware-20250410-1.fc42.noarch | |
| mesa-va-drivers-25.0.4-2.fc42.x86_64 | |
| In BIOS, iGPU setting was on auto. |
I installed and ran s-tui with stress enabled, then I tried bouncing my file manager around with my touchpad - same with Firefox windows. Additionally, I also did the same with Firefox, some 4K YouTube videos. Moving the window around while under s-tui stress with 4k video playing.
I have as of yet, not been able to reproduce. Can you share how long this usually takes and the workload you tried?
Current Fedora 42 kernel should be at 6.14.4-300.fc42, no additional boot parameters in use.
With the tests below, with stress running hard to tax the system. I was unable to create any oddness. I will leave the laptop on, stress running and test again later to see if time creates any new issues.
- Tested against Firefox, default settings. Various pages, including YouTube 4k videos.
- File browser, same result.
- System is stressed with it enabled in s-tui, I can hear the fans.
Same configuration and Iāve also had no issues since upgrading packages to the latest on Fedora 42 when the mesa updates were made available four days ago (per Jules, post #46).
| DNF Package | Version Installed |
|---|---|
| kernel | 6.14.3-300.fc42.x86_64 |
| amd-gpu-firmware.noarch | 20250410-1.fc42 |
| amd-ucode-firmware.noarch | 20250410-1.fc42 |
| mesa-va-drivers.x86_64 | 25.0.4-2.fc42 |
I do still have the kernel argument amdgpu.sg_display=0 set, however it did not solve this issue on the previous version of mesa (as I noted in post #44). I donāt believe this argument has had any effect on my particular issue, but I will remove and test it anyway.
@Max_Pearce_Basman, note that @Matt_Hartley and I cannot reproduce on kernel 6.14.x, whereas you are on 6.13.x. I donāt have time to dive in and figure out if this is true, but perhaps the kernel version interacts with the mesa update.
After updating again Iām at kernel version 6.14.4-300.fc42.x86_64 and itās not immediately happening.
Iāll reply again if the issues manifest once more. It hasnāt been super easy to intentionally reproduce, but typically happens within a few days of using the laptop regularly.
For other details:
Model: Framework Laptop 13 AMD Ryzen 7040 Series
Memory: 32 GiB
Processor: AMD Ryzenā¢ 5 7640U w/ Radeonā¢ 760M Graphics Ć 12
Firmware Version: 03.07
Edit: I should note that Iāve had a number of other freezing and crashing issues, in case they are linked in some way, some of which Iāve noted in replies to [RESPONDED] Laptop crashes and reboot sometimes after resuming from suspend, and some that seem slighly different but might have the same cause (like screen freezing while cursor keeps moving, or screen anc cursor freezing while audio keeps playing, or some combination of those before a black screen and then the login screen).
1 Like
Ohhh I only just found this thread through the ārelatedā or linked button on the forum - still didnt read all.
Just wanted to drop this here: [Issue] FW13 Fedora 41/42 (kernel 6.14) hard freeze - #15 by thedaniel
In my case, with iGPU memory set to just 0.5 GB and a decent amount of screen real estate, I can pretty easily reproduce the issue. Yesterday I had three boots in a row with glitches and freezes within ~2 minutes of boot each time - partially also freezing and gpu restarts from what it felt like - running Ubuntu with the latest mainline 6.14 kernel.
What i do after boot - you could consider a stress test maybe - start browsers up [3-4 profiles], started a youtube video [laptop and chrome is setup to use hardware decode] in one of the tabs from last session. Start some dev tools of which some take 10+ threads full cpu for some minutes, containers etc - system basically goes to ~10GB memory used very quickly.
Last update of mesa-va-drivers-freeworld (25.0.4-1) seems to resolve the issue for me.
I even set back video memory to auto in bios and I didnāt see artefacts for 24h.
Kernel version: 6.14.3-300.fc42.x86_64
Not sure if related but it feels like it is. I have the newer 13 inch with AMD AI 370. I get black screen randomly and i cant even recover from it. I have to hard kill it and reboot.
Linux mint wont even install and get stuck. Ubuntu had screen crash at the beginning of the installer.
Fedora 42 works but then again i get the black screen, can take 30min or 4hours. It happens specially when connected to external display. Ideas?
@Reza try this maybe Framework 13 amd gpu crash debian
1 Like
Thank you! Just to be sure you are refering to this right?
sudo grubby --update-kernel=ALL --args="amdgpu.dcdebugmask=0x10"
Through allot of testing using chatgpt i am right now trying this flag:
sudo nano /etc/default/grub
GRUB_CMDLINE_LINUX=āquiet splash amd_pmf.disable=1ā
sudo grub2-mkconfig -o /boot/efi/EFI/fedora/grub.cfg
reboot
the pmf flag somehow made the system much more stable, never been able to run almost a full day without blackscreen + display going to sleep several time. I just woke it up in the morning and still works.
Will try this for some more time and see if this alone help but if i get the black screen crash again i will try your suggestion on top it.
Update 1: seem the change did not stick so had to also add to blacklist. I had the crash just today 
echo āblacklist amd_pmfā | sudo tee /etc/modprobe.d/blacklist-amd_pmf.conf
sudo dracut -f
reboot