Not feeling like a beta tester.
Due to several issues reported, I’m blacklisting some packages in /etc/dnf/dnf.conf before scheduled dnf upgrade tomorrow. Last upgrade was 8 days ago.
excludepkgs=mesa-*25.0.7*, amd-gpu-firmware, amd-ucode-firmware, \
kernel-6.15.3*, kernel-core-6.15.3*, kernel-headers-6.15.3*, kernel-modules-6.15.3*, kernel-modules-core-6.15.3*, \
kernel-modules-extra-6.15.3*, kernel-tools-6.15.3*, kernel-tools-libs-6.15.3*
Understood.
I updated my rx 7700 amd system and it’s working with no issues.
My guess is that the problems are gpu model specific.
People with 7800xt and 7900xt amd gpus are affected on 6.15*
(Weird cursor issues on Fedora KDE - #3 by deafhobbit)
Good to know it’s GPU specific. I also suspect it may be wayland specific. That thing is as stable as windows 95 was. Only an opinion, but I’ve seen reports.
I’m a developer and Debian sysadmin and sometimes I prefer doing real work and using my computer instead of trying to figure out for hours what’s wrong and get things the way they were before an ‘upgrade’
Will test the next 6.15, in the mean time, happily running 6.12.34-longterm, upgraded a few minutes ago. Next upgrade in 8 days.
System is A10-7850K (R7 GPU), Xorg.
About the kernel/amdgpu/radeon development, there’s nothing that could possibly make my hardware go better. Quite the contrary, it’s bug after bug. The developers don’t have the hardware to test something that old. Hence my idea to freeze mesa, amd firmware, etc.
You know, if it ain’t broken, don’t fix it.
Not a rant, not complaining, only facts.
Im a software engineer and im at this point convinced that modern hardware is not a right fit for me and potentially not well fit for modern linux altogether yet. I find that the linux experience is so unstable for me, and its headache after headache encountering issues, albeit mostly minor, that take forever to diagnose the cause of. i dont know whether an issue is caused by a kernel bug affecting user space or some critical library maintained by just 2 guys working in open source is causing the issue. It doesnt matter to me that the kernel is purportedly stable, when in reality what the end user experiences is constant pain from kernel modules that break, and take ages to fix because only 1 guy works on it or is skilled enough to maintain that driver or whatever the thing might be.
Worst part is the time and effort it takes to track down whether i broke the entire thing by just adding in that one package that now cascades into causing issues elsewhere and with other packages, or if some update caused the thing to break and its outside my hands. Regardless, the experience FEELS very unstable, because things break with very minor changes, and correcting the issue is like trying to make changes to a legacy microservice application.
maybe i just need to become a sys admin to understand things better
I don’t mean to be harsh, but if you don’t want problems with the latest versions of software, avoid running bleeding-edge distributions like the latest Fedora (or Arch, Gentoo, or similar distros).
Something like Ubuntu LTS might suit your needs much better. Alternatively, you could run a previous version of Fedora—Fedora is supported for two cycles, so you could still use Fedora 41, which only receives security and minor updates now.
pls explain how this would have worked for you in this case?
https://bodhi.fedoraproject.org/updates/FEDORA-2025-e30e0c2127
F41 gets all the firmware and kernel updates just like f42 does.
It seems you have completely missed the point I was making.
First of all, I was speaking in general terms. If you want a more stable experience, run an LTS distribution. My first recommendation was Ubuntu LTS, and I suggested the previous Fedora only as an alternative for those who wished to stick with Fedora.
Secondly, I never said that the experience would be completely bug-free; I only mentioned that bleeding-edge distributions are more likely to have issues.
Also there are at least two amdgpu issues, and one of them can be fixed by downgrading the firmware or upgrading to an upstream firmware version. I don’t know which issue the poster I was replying to was affected by, but their post made it sound like a general issue, which I addressed.
Just use previous kernel by grub menu.
This is the way on Fedora.
Good advice generally, but recently mesa has been a cause of problems and now firmware is implecated.
Most of the issues at the moment are firmware related. as the firmware ends up in the initramfs, you have to either downgrade the firmware and then rebuild the initramfs, or wait for the next update (6.15.4 is being built today with an updated firmware which should fix these issues.
Upgraded succesfully. Enhanced exclusions though. System is A10-7850k. Xorg. No wayland!
excludepkgs=sublime-text, mesa-*, \
brave-browser, vivaldi-stable, \
amd-gpu-firmware, amd-ucode-firmware, \
kernel*6.14*, kernel*6.15.3*, kernel*6.15.4*
Available kernels:
$ ll /lib/modules
drwxr-xr-x 1 root root 742 9 mai 12:13 6.12.25-200.fc42.x86_64
drwxr-xr-x 1 root root 764 4 jui 08:53 6.12.35-200.fc42.x86_64
drwxr-xr-x 1 root root 772 29 mai 22:13 6.14.5-300.fc42.x86_64
I might test 6.15. Not a priority. Will wait for negative reports which will soon arrive.
6.12 is unconditionally stable.
6.14.11 resulted in crashes in the terminal! Requiring a REISUB. Back to 6.14.5.
Released hold on mesa only because wine depends on it. Upgraded again. Currently mesa 25.1.4-2.