Hi,
I want to make a suggestion regarding the NVIDIA driver situation on openSUSE Tumbleweed and Slowroll.
Right now, the NVIDIA drivers get out of sync with the open kernel module (nvidia-open-driver-G06) about 90% of the time with every new release. It’s a mess, and it’s hard to fix manually because the packages in the repos don’t match up.
# The Situation
My system is currently broken because the userspace drivers updated, but the kernel module is lagging behind. Look at this zypper output:
`Bash# Userspace is at 580.142
i | nvidia-gl-G06 | package | 580.142-46.1
i | nvidia-compute-G06 | package | 580.142-46.1
BUT the Kernel Module (KMP) is stuck at 580.126
i+ | nvidia-open-driver-G06-signed-cuda-kmp-default | package | 580.126.20_k6.19.6…`
Because Userspace (580.142) > Kernel Module (580.126), the driver stack is out of sync and fails.
# The Suggestion
Can we make the open driver module package (nvidia-open-driver-G06-signed-cuda-kmp-default) request and block the install of NVIDIA userspace drivers if their version is higher than the module version?
If the KMP isn’t ready for 580.142 yet, zypper shouldn’t be allowed to install the 580.142 userspace. It should stay locked at 580.126 until the matching KMP hits the repo.
# Why this is needed
The repo is full of old versions, but the one version I actually need to match my drivers is missing. Here is how fragmented the update-slowroll repo looks right now:
- Userspace available: 580.142
- KMP versions available: 580.126, 580.105, 580.95, 580.82, 580.65… (and many more)
There is a huge list of old modules, but nothing that matches the latest userspace being pushed. This forces a broken state on the user every time they zypper dup.
# Summary of Logic
If we add a version check/conflict to the KMP package:
- KMP Version 580.126 → Blocks any NVIDIA-Userspace > 580.126.
This would stop the “out of sync” nightmare and make Slowroll/TW much more stable for NVIDIA users.



