OM is freezing at Loading Initial Ramdisk, even after fresh reinstall

Hello,

Requirements:

I have searched the forum for my issue and found nothing related or helpful

I did find this old post which seems slightly related, but different:
title: boot-hangs-after-initiating-ramdisk-when-installed
topic: #6599

It is different because my situation happens even without attempting to install nvidia drivers.

I have checked the Resources category
I have reviewed the wiki for relevant information
I have read the the Release Notes and Errata

OpenMandriva Lx version:

OpenMandriva Lx release 25.06 (ROME) Rolling for x86_64

Desktop environment (KDE, LXQT…):
KDE

Description of the issue (screenshots if relevant):
My OM computer was starting to show lots of issues, since first installing in June of 2025. The biggest unsolvable issue was that sometimes the system would just randomly cut to a black screen. Sometimes I could get to TTY from that blackscreen and use A.I. to troubleshoot and reboot. Other times, I had to reinstall OM from scratch.
With the help of A.I., we narrowed down the issue to the official Nvidia driver crashing.

At the time this was happening, the forum had that email sign-up glitch, so I couldn’t get an account and post here.

With the help of A.I., I got the nouveau driver installed and my nvidia driver has otherwise not been able to do anything other than display graphics.

The problem went away from awhile, then the nouveau driver started acting funny.

When I would open up a Dolphin window, the whole computer would hang for about a minute. I could wait it out and the system would kind of recover, but if I mouse hovered over any file or tried to open a folder, the system would hang for another minute.

With the help of Claude, we determined that the nouveau driver was the issue.

There was a new official Nvidia 580 driver release on Jan 13, 2026 that is supposed to support my 980 Ti gpu.

I attempted to install, both via the Welcome Center and via terminal.

Both paths yielded a bizarre bug. When rebooting, the screen would hang at “Loading Initial Ramdisk.”

After troubleshooting for a long time with the help of Claude, I decided to do a complete reinstall, and to switch to ROME (previously, I was using ROCK).

This is where things became very weird. I have asked Claude AI to summarize. I hope you don’t mind the A.I. content. It seems thorough and gets to the point. A lot of this area of computing is outside my area of expertise.


OpenMandriva Rome 25.06 - Critical dracut/initrd Bug Report

Summary

Fresh installations of OpenMandriva Rome 25.06 create initrd images with ZERO kernel modules due to a path format mismatch in modules.dep. The system boots anyway by loading modules after mounting the root filesystem, but this causes complete boot failure when NVIDIA proprietary drivers are installed (system freezes at “Loading initial ramdisk”).

The Bug

The modules.dep file in /lib/modules/$(uname -r)/modules.dep contains relative paths:

kernel/drivers/gpu/drm/nouveau/nouveau.ko.zst
kernel/arch/x86/events/amd/power.ko.zst
kernel/arch/x86/crypto/twofish-x86_64.ko.zst: kernel/crypto/twofish_common.ko.zst

These are relative paths starting with kernel/ rather than absolute paths like:

/lib/modules/6.14.2-desktop-3omv2590/kernel/drivers/gpu/drm/nouveau/nouveau.ko.zst

When dracut builds the initrd, it appears unable to find modules using these relative paths, resulting in an initrd containing no kernel modules whatsoever.

Why The System Still Boots (For Now)

Despite having an empty initrd, the system boots successfully with nouveau (open source) drivers because:

  1. Initrd boots using basic framebuffer/VESA mode
  2. Kernel successfully mounts root filesystem from /dev/sdc2
  3. systemd/udev loads modules directly from /lib/modules/ after boot
  4. Graphics initialize late but work fine

Evidence - modules are loaded and working:

$ lsmod | grep nouveau
nouveau              3203072  62
drm_gpuvm              45056  1 nouveau
[... many more DRM modules ...]

$ lspci -k | grep -A 3 VGA
01:00.0 VGA compatible controller: NVIDIA Corporation GM204 [GeForce GTX 980]
        Kernel driver in use: nouveau

Why NVIDIA Driver Installation Causes Boot Failure

When NVIDIA proprietary drivers are installed:

  1. Installation blacklists nouveau driver
  2. dracut rebuilds initrd (still with 0 modules due to modules.dep bug)
  3. System reboots
  4. Initrd tries to initialize NVIDIA driver early in boot
  5. NVIDIA modules aren’t in the initrd (because of the bug)
  6. System freezes at “Loading initial ramdisk”
  7. Complete reinstallation required

How to Verify You’re Affected

Check modules.dep format:

head -20 /lib/modules/$(uname -r)/modules.dep

If paths start with kernel/ (relative) instead of /lib/modules/... (absolute), you’re affected.

Check if your initrd contains ANY kernel modules:

sudo lsinitrd /boot/initrd-$(uname -r).img | grep "\.ko" | wc -l

If this returns 0, your initrd has no kernel modules.

Check what’s actually loaded in your running system:

lsmod | wc -l

If this shows many modules (should be 100+), they’re being loaded post-boot, not from initrd.

Reproduction Steps

  1. Fresh install OpenMandriva Rome 25.06
  2. Verify modules.dep has relative paths and initrd has 0 modules
  3. System boots fine with nouveau
  4. Install NVIDIA drivers: sudo dnf install dkms-nvidia
  5. Reboot
  6. System freezes at “Loading initial ramdisk”

System Info

  • Distribution: OpenMandriva Lx 25.06 (ROME)
  • Kernel: 6.14.2-desktop-3omv2590
  • GPU: NVIDIA GTX 980 Ti
  • Installation: Fresh install from 6.14 ROCK live USB, then switched to Rome
  • Filesystem: ext4 on /dev/sdc2 (root), /dev/sdc3 (home)

Questions for Developers

  1. Which package generates modules.dep with relative paths? (kmod? depmod?)
  2. Should dracut be patched to handle relative paths, or should modules.dep generation be fixed to use absolute paths?
  3. Is this affecting all Rome 25.06 installations?
  4. Why does the current boot process work without modules in initrd, and is this intentional?
  5. Are there other scenarios besides NVIDIA driver installation that could trigger boot failure?

Impact

  • All fresh Rome 25.06 installations appear to have empty initrd images
  • System works with open source drivers (nouveau, radeon, etc.) because modules load post-boot
  • Installing proprietary NVIDIA drivers causes unrecoverable boot failure
  • Potentially affects other scenarios requiring early module loading

Relevant informations (hardware involved, software version, logs or output…):

$ inxi -Fxz
System:
  Kernel: 6.14.2-desktop-3omv2590 arch: x86_64 bits: 64 compiler: clang
    v: 19.1.7
  Desktop: KDE Plasma v: 6.3.4 Distro: OpenMandriva Lx 25.06 ROME
Machine:
  Type: Desktop Mobo: ASUSTeK model: H170 PRO GAMING v: Rev X.0x
    serial: <superuser required> UEFI: American Megatrends v: 3805
    date: 05/16/2018
CPU:
  Info: quad core model: Intel Core i5-6600 bits: 64 type: MCP arch: Skylake-S
    rev: 3 cache: L1: 256 KiB L2: 1024 KiB L3: 6 MiB
  Speed (MHz): avg: 3692 min/max: 800/3900 cores: 1: 3692 2: 3692 3: 3692
    4: 3692 bogomips: 26399
  Flags: avx avx2 ht lm nx pae sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx
Graphics:
  Device-1: NVIDIA GM204 [GeForce GTX 980] vendor: eVga.com. driver: nouveau
    v: kernel arch: Maxwell bus-ID: 01:00.0 temp: 47.0 C
  Display: x11 server: X.org v: 1.21.1.18 driver: X: loaded: modesetting
    unloaded: fbdev,vesa dri: nouveau gpu: nouveau resolution: 1920x1080~60Hz
  API: EGL v: 1.5 drivers: nouveau,swrast platforms:
    active: gbm,x11,surfaceless,device inactive: wayland
  API: OpenGL v: 4.5 compat-v: 4.3 vendor: mesa v: 25.0.7 glx-v: 1.4
    direct-render: yes renderer: NV124
  API: Vulkan v: 1.4.312 drivers: N/A surfaces: xcb,xlib devices: 1
  Info: Tools: api: clinfo, eglinfo, glxinfo, vulkaninfo de: kscreen-console
    x11: xdpyinfo, xprop, xrandr
Audio:
  Device-1: Intel 100 Series/C230 Series Family HD Audio vendor: ASUSTeK
    driver: snd_hda_intel v: kernel bus-ID: 00:1f.3
  Device-2: NVIDIA GM204 High Definition Audio vendor: eVga.com.
    driver: snd_hda_intel v: kernel bus-ID: 01:00.1
  API: ALSA v: k6.14.2-desktop-3omv2590 status: kernel-api
  Server-1: PipeWire v: 1.4.4 status: active
Network:
  Device-1: Intel Ethernet I219-V vendor: ASUSTeK driver: e1000e v: kernel
    port: N/A bus-ID: 00:1f.6
  IF: enp0s31f6 state: up speed: 1000 Mbps duplex: full mac: <filter>
  Device-2: Broadcom BCM4360 802.11ac Dual Band Wireless Network Adapter
    vendor: ASUSTeK driver: bcma-pci-bridge v: N/A bus-ID: 02:00.0
Drives:
  Local Storage: total: 5.92 TiB used: 63.5 GiB (1.0%)
  ID-1: /dev/nvme0n1 vendor: Crucial model: CT4000P3PSSD8 size: 3.64 TiB
    temp: 34.9 C
  ID-2: /dev/sda vendor: Samsung model: SSD 850 EVO 1TB size: 931.51 GiB
  ID-3: /dev/sdb vendor: Samsung model: SSD 840 EVO 1TB size: 931.51 GiB
  ID-4: /dev/sdc vendor: Samsung model: SSD 850 EVO 500GB size: 465.76 GiB
  ID-5: /dev/sdd vendor: Kingston model: DataTraveler 3.0 size: 7.2 GiB
    type: USB
Partition:
  ID-1: / size: 287.31 GiB used: 6.16 GiB (2.1%) fs: ext4 dev: /dev/sdc2
  ID-2: /boot/efi size: 299.4 MiB used: 2 MiB (0.7%) fs: vfat dev: /dev/sdc1
  ID-3: /home size: 161.04 GiB used: 57.33 GiB (35.6%) fs: ext4
    dev: /dev/sdc3
Swap:
  ID-1: swap-1 type: zram size: 7.79 GiB used: 123.3 MiB (1.5%)
    dev: /dev/zram0
  ID-2: swap-2 type: partition size: 7.81 GiB used: 0 KiB (0.0%)
    dev: /dev/sdc4
Sensors:
  System Temperatures: cpu: 31.0 C mobo: N/A gpu: nouveau temp: 47.0 C
  Fan Speeds (rpm): N/A gpu: nouveau fan: 0
Info:
  Memory: total: 16 GiB available: 15.57 GiB used: 3.43 GiB (22.0%)
  Processes: 242 Uptime: 1h 9m Init: systemd target: graphical (5)
  Packages: N/A note: see --rpm Compilers: N/A Shell: Bash v: 5.2.37
    inxi: 3.3.37

you cant use anymore dkms-nvidia because thes version 590 is only for 1650 or better nvidia card

try to replace dkms-nvidia-580xx if possible or older one

I’ve tried a few times now to install 580.

I downloaded the NVIDIA-Linux-x86_64-580.126.09.run file and tried to run it using tty.

I can’t get the nouveau driver to stop running properly so that Nvidia can install.

I do try to stop the session using sudo systemctl stop sddm etc. and I have tried (and not tried) adding in blacklists in the grub settings.

One of the first errors is that the gcc lib doesn’t point to cc. That was easy enough to fix with a soft link.

However, any attempts to blacklist or block nouveau, including the Nvidia installer methods or using dracut, create a situation in which I am unable to recover the system normally.

I’ve used timeshift to restore my system to the point where nouveau worked.

Here’s the AI summary of the troubleshooting session.

Technical Summary: Nvidia Installation Failure & Initrd Deadlock on ROME 25

Environment:

  • OS: OpenMandriva ROME (Rolling)
  • Version: 25

The Issue: Attempting to install Nvidia drivers via the official manual .run script led to an unrecoverable system break characterized by a “Driver Vacuum” and an initrd boot hang.

Sequence of Events:

  1. Dependency Issues: Initial attempts to run the Nvidia installer failed due to a missing /usr/bin/cc symlink and Qt6 library conflicts during the distro-sync phase.
  2. Manual Blacklisting: Nouveau was blacklisted via /etc/modprobe.d/ and dracut --force was executed to regenerate the initramfs.
  3. Boot Hang: Upon reboot, the system hung indefinitely at Loading initial ramdisk. This was likely caused by the removal of the Nouveau driver from the initrd before the Nvidia kernel modules were successfully compiled and injected.
  4. Login Loop/TTY Failure: Attempts to bypass via nomodeset allowed access to the Display Manager (SDDM), but resulted in an immediate login loop or a total black screen when attempting to switch to TTY (Ctrl+Alt+F2).
  5. Recovery: System was only recoverable by booting a Live USB, installing Timeshift in the live environment, and restoring a “Raw” snapshot.

Key Observations for Developers:

  • The manual .run installer from Nvidia appears to conflict with the Clang/LLVM toolchain used in ROME.
  • Standard dracut --force procedures for blacklisting Nouveau resulted in a kernel panic/hang at the ramdisk stage, suggesting a dependency on Nouveau for basic console output that wasn’t being handed off correctly to the VESA/Basic frame buffer.

Thank you, by the way, for the time you are taking to attempt any assistance.

Unfortunately, we cannot support proprietary nvidia drivers. We also cannot provide the Legacy versions of their proprietary drivers as that falls even further out of the support scope. Those drivers are (and always have been) provided as a convenience. Because of how they version their release, we need to package the newest and consistently available version both because of the kernel version, and the architectures we build for.

If you have a card older than the 1650 and you need the proprietary driver, it should continue to work in Rock until the next release. Otherwise, you will need to get the binary from nvidia directly and install it according to their guidance. You will probably need dkms installed for it to work properly. Any issues you have with the driver will need to be discussed in the nvidia forums.

Okay, thank you for the help.

1 Like

This topic was automatically closed 15 days after the last reply. New replies are no longer allowed.