OpenMandrivaLx Rome hangs after update (08-12-2025)

[n1ckla]

I’ve just updated Rome (Dec 8th, 2025). Now, when rebooting, it hangs at “Loading initial ramdisk”. Selecting the advanced boot menu, testing all variations of kernel and boot mode (normal, console, etc) leads to the same result.

After some searching, it seems that the cause is this line in the Grub:

linux   /@boot/vmlinuz-6.18.0-desktop-1omv2590 root=UUID=b4dcbab4-030c-4754-b054-2370d180dee6 ro rootflags=subvol=@  nouveau.modeset=0 nvidia-drm.modeset=1 nvidia-drm.fbdev=1 quiet splash resume=UUID=20358ba0-7bcd-480f-9e30-29cd535bb0ca logo.nologo audit=0 rd.timeout=120 dm_mod.use_blk_mq=1 rd.systemd.show_status=0 systemd.show_status=0

The offending options are these:

nouveau.modeset=0 nvidia-drm.modeset=1 nvidia-drm.fbdev=1

Particularly, the nouveau.modeset=0. If I remove nvidia-drm.modeset=1 nvidia-drm.fbdev=1, but keep nouveau.modeset=0, it will hang at “Loading initial ramdisk”

However, if I reverse that, keeping nvidia-drm.modeset=1 nvidia-drm.fbdev=1 but removing nouveau.modeset=0, the system boot normally, and I can login to the system.

However, by doing that, the system uses nouveau drivers instead:

$ lsmod | grep nvidia
$ nvidia-smi
NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.

$ lspci -k | grep -A 3 VGA
01:00.0 VGA compatible controller: NVIDIA Corporation GP107 [GeForce GTX 1050 Ti] (rev a1)
        Subsystem: ASUSTeK Computer Inc. Device 85d1
        Kernel driver in use: nouveau
        Kernel modules: nouveau, nvidia_drm, nvidia

$ glxinfo | grep -i "opengl renderer"
OpenGL renderer string: NV137

$ lsmod | grep nouveau
nouveau              3297280  64
drm_gpuvm              49152  1 nouveau
gpu_sched              53248  1 nouveau
drm_exec               16384  2 drm_gpuvm,nouveau
drm_client_lib         16384  1 nouveau
drm_display_helper    278528  1 nouveau
i2c_algo_bit           20480  1 nouveau
drm_ttm_helper         16384  1 nouveau
drm_kms_helper        253952  4 drm_display_helper,drm_ttm_helper,drm_client_lib,nouveau
mxm_wmi                12288  1 nouveau
ttm                   126976  2 drm_ttm_helper,nouveau
agpgart                53248  2 ttm,nouveau
drm                   831488  26 gpu_sched,drm_kms_helper,drm_exec,drm_gpuvm,drm_display_helper,drm_ttm_helper,drm_client_lib,ttm,nouveau
video                  73728  2 asus_wmi,nouveau
wmi                    32768  5 video,asus_wmi,wmi_bmof,mxm_wmi,nouveau
button                 20480  1 nouveau
i2c_core               94208  10 i2c_mux,i2c_algo_bit,at24,drm_display_helper,i2c_smbus,i2c_i801,regmap_i2c,i2c_dev,nouveau,drm

Checking if the NVidia drivers are installed:

$ ls -la /usr/lib/modules/6.18.0-desktop-1omv2590/kernel/drivers/video/nvidia/
total 180836
drwxr-xr-x 1 root root       138 Dec  7 23:21 .
drwxr-xr-x 1 root root        30 Dec  7 23:21 ..
-rw-r--r-- 1 root root   2576576 Dec  4 07:11 nvidia-drm.ko
-rw-r--r-- 1 root root 141251960 Dec  4 07:11 nvidia.ko
-rw-r--r-- 1 root root   3564456 Dec  4 07:11 nvidia-modeset.ko
-rw-r--r-- 1 root root    227488 Dec  4 07:11 nvidia-peermem.ko
-rw-r--r-- 1 root root  37541344 Dec  4 07:11 nvidia-uvm.ko

$ modinfo nvidia | grep -i version
version:        590.44.01
srcversion:     87C3F96C838262732FB449A
vermagic:       6.18.0-desktop-1omv2590 SMP preempt mod_unload modversions

$ rpm -qa | grep nvidia 
lib64nvidia-egl-wayland-1.1.20-1.x86_64
lib64nvidia-egl-gbm-1.1.2.1-1.x86_64
lib64nvidia-egl-x11-1.0.3-1.x86_64
libnvidia-egl-wayland-1.1.20-1.x86_64
nvidia-32bit-wayland-590.44.01-1.x86_64
nvidia-32bit-590.44.01-1.x86_64
nvidia-modprobe-590.44.01-1.x86_64
nvidia-persistenced-590.44.01-1.x86_64
nvidia-wayland-590.44.01-1.x86_64
nvidia-x11-590.44.01-1.x86_64
nvidia-590.44.01-1.x86_64
nvidia-kmod-desktop-590.44.01-1_6.18.0_1.x86_64
nvidia-kmod-common-590.44.01-1.noarch
nvidia-settings-590.44.01-1.x86_64
nvidia-firmware-20251125-1.noarch

Any idea how to resolve the issue, making the system boots normally while the NVidia drivers work?

Is it possible that NVidia drivers v590.x do not work with my GTX 1050 Ti? Can I downgrade to the last working version? It works fine before the update.

[zeroability]

Sounds like the nvidia drivers are broken. We saw another report in the newcomer chat about this. Could you try sudo dnf rm nvidia and then reboot? You should get back to graphics with the nouveau driver. Then try sudo dnf in nvidia and reboot again?

[zeroability]

@n1ckla

Can I downgrade to the last working version? It works fine before the update.

Unfortunately not. We only build one driver version against the kernel.

[n1ckla]

@zeroability

Sounds like the nvidia drivers are broken. We saw another report in the newcomer chat about this. Could you try sudo dnf rm nvidia and then reboot? You should get back to graphics with the nouveau driver. Then try sudo dnf in nvidia and reboot again?

I tried your suggestion:

sudo dnf rm nvidia: This removes the nvidia drivers, as well removes these parameters nouveau.modeset=0 nvidia-drm.modeset=1 nvidia-drm.fbdev=1 from the grub menu entry. The system boots fine.

sudo dnf in nvidia: Re-install the nvidia drivers and re-add those drivers, boot hangs at “Loading initial Ramdisk”

Now, I don’t really know what happened, as removing all other options from the grub menu entries (quiet, splash, *.show_status) do not result on boot messages printed on screen. I do suspect that the boot hangs at initrdfor some reason.

Here’s my hypothesis. My NVidia card is GTX 1050 Ti. Doing some research, NVidia just drop the support for it in the v590 branch of their drivers recently.

So, I think what happened is that with nouveau.modeset=0, the kernel drops the nouveau drivers, but when nvidia-drm.modeset=1 kicks in, my hardware isn’t supported, so it fails to load. Thus, the system is now without any display driver and do not know what to do, thus it hangs.

@zeroability

@n1ckla

Can I downgrade to the last working version? It works fine before the update.

Unfortunately not. We only build one driver version against the kernel.

Could you please reconsider? I don’t think the problem lies in “nvidia drivers are broken”. If it did, I’d think you have much more report of it. The problem is NVidia is notorious in dropping support for older cards. This is Win 10 end-of-life problem all over again, chucking out still perfectly working hardware to upgrade to Win 11.

I did see nvidia-legacy package in the repo that is nvidia drivers v470x, but seems unmaintained because it is compiled for kernel v6.7.1. Why don’t support/maintain those, so people with legacy cards could use them? The last working nvidia version in my system, just before I update, is v580.95. Why not update nvidia-legacy package to at least a newer package, or if not that, compile it to the latest kernel? Last time on windows, I used NVidia drivers version 390.x.

May I suggest at the very least, give warning and notifications about possible breaking changes when Nvidia driver updates, cos NVidia is notorious in dropping support.

I do hope there’s the last nvidia driver version still cached somewhere, as this problem with nvidia drivers is new.

[zeroability]

@n1ckla

Here’s my hypothesis. My NVidia card is GTX 1050 Ti. Doing some research, NVidia just drop the support for it in the v590 branch of their drivers recently.

We have seen this with the 960, as well. It’s probably the case.

@n1ckla

Could you please reconsider? I don’t think the problem lies in “nvidia drivers are broken”. If it did, I’d think you have much more report of it. The problem is NVidia is notorious in dropping support for older cards. This is Win 10 end-of-life problem all over again, chucking out still perfectly working hardware to upgrade to Win 11.

No. Their drivers are proprietary for those cards and they have already stated they will not backport fixes to newer kernels. It has been explained here:

Nvidia Open kernel module (Call for testing, ROME or Cooker)

Since the availability of the Open Source Linux kernel driver for nvidia, we have been interested in how it would work in OMLx, but have not had compatible cards to test it with. Based on recent guidance from the company, they will be suggesting this Open driver be the default: They did not include the compatible cards in this article, but the list is here: We package two versions of the proprietary and open drivers, one built with the installed kernel, and one with dkms. The latter is rec…

@n1ckla

I did see nvidia-legacy package in the repo that is nvidia drivers v470x, but seems unmaintained because it is compiled for kernel v6.7.1. Why don’t support/maintain those, so people with legacy cards could use them?

Because nvidia does not support those drivers being built on newer kernels. So, neither can we.

@n1ckla

May I suggest at the very least, give warning and notifications about possible breaking changes when Nvidia driver updates, cos NVidia is notorious in dropping support.

I do hope there’s the last nvidia driver version still cached somewhere, as this problem with nvidia drivers is new.

This is a topic we have covered at great length in all of our communication platforms. It is why the nvidia drivers are in the non-free repository:

1. We provide it as a convenience for users, and supporting hardware and troubleshooting will ultimately be up to the user and the vendor. The driver is closed source.
2. We cannot guarantee the driver will work for everyone, or be available to everyone forever. Please see # 1 for clarification.
3. Nothing stops a user from downloading the driver from nvidia directly and not using the package we provide at all. Especially in scenarios where the driver is considered “Legacy”. We will not be able to make the resulting driver you build work with the kernel if nvidia has pulled the rug out.

@n1ckla

sudo dnf rm nvidia: This removes the nvidia drivers, as well removes these parameters nouveau.modeset=0 nvidia-drm.modeset=1 nvidia-drm.fbdev=1 from the grub menu entry. The system boots fine.

This is honestly going to be the best long term solution unless you buy a non-nvidia graphics card.

[n1ckla]

@zeroability

@n1ckla

I did see nvidia-legacy package in the repo that is nvidia drivers v470x, but seems unmaintained because it is compiled for kernel v6.7.1. Why don’t support/maintain those, so people with legacy cards could use them?

Because nvidia does not support those drivers being built on newer kernels. So, neither can we.

Ah, I didn’t know that. I thought it’s just a matter of compiling. But if that include actual incompatibilities with the newer kernel, Well, what can I say…

Could you clarify, please, whether the nvidia package is the proprietary binaries, and nvidia-kmod-open-desktop is the open source one, and both has their versions in sync with each other (i.e. both is 590x), and has no relation to nouveau?

[zeroability]

@n1ckla

Could you clarify, please, whether the nvidia package is the proprietary binaries, and nvidia-kmod-open-desktop is the open source one, and both has their versions in sync with each other (i.e. both is 590x), and has no relation to nouveau?

nvidia-kmod-open-desktop is the open source kernel module only (for the desktop kernel flavor). We have been packaging that for a while now. It is built when the proprietary drivers are built. They are part of the same package.

nvidia.spec

nvidia will include the proprietary software like cuda, and some other closed source userland programs needed to interact with the open kernel module.

nouveau is the open source project housed by the freedesktop org, and it makes use of proprietary firmware of nvidia cards to provide at least basic graphics support (and some 3D support to some cards).

[alangea]

As I stated in a support request of mine, on another thread, I suspect this is likely the issue I experienced when launching the new version of ROME by USB.

In my case, however, I’m not able to get around by removing the nvidia drivers, simply because everthing stick up at the first background image, and there’s no way to access the boot script.

[zeroability]

Please don’t hijack someone else’s support topic. It makes it difficult for someone to search for an answer buried in an unrelated topic.

[alangea]

Ok.

I apologia.