Skip to content

[610.57.04 Open] RTX 3070 Laptop: GSP GFW_BOOT failure causes intermittent reboot hang requiring hard power-off #1294

Description

@SwiftExplorer567

NVIDIA Open GPU Kernel Modules Version

610.57.04

Please confirm this issue does not happen with the proprietary driver (of the same version). This issue tracker is only for bugs specific to the open kernel driver.

  • I confirm that this does not happen with the proprietary driver package.

Operating System and Version

CachyOS (Arch Linux-based rolling release), KDE Plasma 6.7.4, Wayland

Kernel Release

Linux 7.1.8-1-cachyos Also reproduced on: Linux 6.18.42-1-cachyos-lts

Please confirm you are running a stable release kernel (e.g. not a -rc). We do not accept bug reports for unreleased kernels.

  • I am running on a stable kernel release.

Hardware: GPU

NVIDIA GeForce RTX 3070 Laptop GPU (Ampere, PCI ID 10de:249d) Laptop: Acer Nitro AN515-57 CPU: Intel Core i7-11800H iGPU: Intel Tiger Lake-H UHD Graphics (i915) Hybrid Intel + NVIDIA graphics BIOS: V1.20 Secure Boot: disabled

Describe the bug

I am experiencing an intermittent reboot/shutdown hang on an Acer Nitro AN515-57 laptop with an NVIDIA GeForce RTX 3070 Laptop GPU when using the NVIDIA 610.57.04 Open Kernel Module.

When the problem occurs, I initiate a normal reboot from Linux. The graphical session begins shutting down and the display eventually goes black, but the laptop remains powered on indefinitely and never completes the reboot. A long press of the power button is required.

The issue reproduces on both:

  • linux-cachyos 7.1.8-1
  • linux-cachyos-lts 6.18.42-1

The system uses Intel + NVIDIA hybrid graphics with KDE Plasma on Wayland.

I completely reinstalled the NVIDIA driver using the default CachyOS laptop profile:

sudo chwd -f -i nvidia-open-dkms.prime

The resulting NVIDIA stack was:

nvidia-utils 610.57.04
nvidia-settings 610.57.04
linux-cachyos-nvidia-open 7.1.8-1
linux-cachyos-lts-nvidia-open 6.18.42-1

I then rebuilt the initramfs and Limine boot configuration.

After rebooting, the active kernel module was confirmed as:

NVIDIA UNIX Open Kernel Module for x86_64 610.57.04

The reboot hang reproduced again after this clean reinstall.

During affected boots, the NVIDIA driver logs a GSP firmware initialization failure:

NVRM: gpuWaitForGfwBootComplete_TU102: failed to wait for GFW_BOOT: (progress 0x9)
NVRM: kgspWaitForGfwBootOk_TU102: failed to wait for GFW boot complete: 0x55 VBIOS version 94.04.43.00.82
NVRM: kgspWaitForGfwBootOk_TU102: (the GPU may be in a bad state and may need to be reset)

During shutdown, nvidia-powerd then fails to terminate normally and eventually times out even after SIGKILL:

nvidia-powerd.service: State 'final-sigterm' timed out. Killing.
nvidia-powerd.service: Killing process (nvidia-powerd) with signal SIGKILL.
nvidia-powerd.service: Processes still around after final SIGKILL.
nvidia-powerd.service: Failed with result 'timeout'.
nvidia-powerd.service: Unit process (nvidia-powerd) remains running after unit stopped.

KWin also fails to terminate during the same shutdown:

plasma-kwin_wayland.service: State 'stop-sigterm' timed out. Killing.
plasma-kwin_wayland.service: Killing process (kwin_wayland) with signal SIGKILL.
plasma-kwin_wayland.service: Failed with result 'timeout'.

On another failed reboot with 610.57.04 Open, I also observed NVIDIA MMU/reset-related errors including:

mmuWalkUnmap: Failed to unmap VA Range
NV_ERR_INVALID_STATE
NV_ERR_RESET_REQUIRED

As an A/B test, I replaced the 610.57.04 Open Kernel Module with the NVIDIA proprietary 580.173.02 DKMS driver.

With 580.173.02 proprietary, reboot completed normally during testing and the GSP/GFW_BOOT/reset error pattern was not observed during the successful reboot.

I then switched back to 610.57.04 Open, performed a clean reinstall through CachyOS CHWD, rebuilt the initramfs, and reproduced the reboot hang again.

Current test matrix:

Kernel NVIDIA driver Module type Result
7.1.8-1-cachyos 610.57.04 Open Reboot hang reproduced
6.18.42-1-cachyos-lts 610.57.04 Open Reboot hang reproduced
7.1.8-1-cachyos 580.173.02 Proprietary Reboot completed normally during testing
7.1.8-1-cachyos 610.57.04 clean reinstall Open Reboot hang reproduced again

Important caveat:

I have not yet tested the proprietary kernel module at the same 610.57.04 version.

Therefore, I cannot currently conclude whether this is specifically an Open Kernel Module bug, a regression in the 610 branch, or an interaction between 610 and the GSP firmware/power-management path.

However, the GSP/GFW_BOOT failure occurring before the failed reboot, followed by nvidia-powerd and kwin_wayland becoming unable to terminate, appears strongly correlated with the reboot hang.

To Reproduce

Prerequisites:

  • Acer Nitro AN515-57 or similar Intel + NVIDIA hybrid graphics laptop
  • NVIDIA GeForce RTX 3070 Laptop GPU
  • NVIDIA Open Kernel Module 610.57.04
  • KDE Plasma / Wayland
  • CachyOS

Steps:

  1. Install the NVIDIA 610.57.04 Open Kernel Module.

    On CachyOS:

    sudo chwd -f -i nvidia-open-dkms.prime
    
  2. Confirm the Open Kernel Module is loaded:

    cat /proc/driver/nvidia/version
    

    Expected output includes:

    NVIDIA UNIX Open Kernel Module for x86_64 610.57.04
    
  3. Log into a KDE Plasma Wayland session.

  4. Use the system normally for a short period.

  5. Reboot normally:

    systemctl reboot
    
  6. Intermittently, the graphical session shuts down and the screen goes black, but the laptop remains powered on indefinitely.

  7. Force power-off by holding the power button.

  8. Boot again and inspect the previous boot:

    journalctl -b -1 -k
    

The failed boot contains GSP firmware errors such as:

gpuWaitForGfwBootComplete_TU102: failed to wait for GFW_BOOT
kgspWaitForGfwBootOk_TU102: failed to wait for GFW boot complete
the GPU may be in a bad state and may need to be reset

The shutdown journal also shows nvidia-powerd and kwin_wayland timing out while stopping.

Bug Incidence

Sometimes

nvidia-bug-report.log.gz

nvidia-bug-report.log.gz

More Info

Kernel command line during the clean 610.57.04 Open reproduction:

quiet nowatchdog splash rw rootflags=subvol=/@ nvidia_drm.modeset=1 nvidia_drm.fbdev=1 acpi=force i915.enable_psr=0

A previously present:

reboot=pci

parameter was removed during troubleshooting.

The reboot hang still reproduced without it.

The issue was also tested with the CachyOS LTS kernel, so it does not appear to be specific to the 7.1 kernel.

Possibly related issues:

I did not find an existing issue specifically covering an RTX 3070 Laptop GPU with NVIDIA 610.57.04 Open where the GSP/GFW_BOOT failure is followed by an ordinary reboot/shutdown hanging indefinitely.

nvidia-610-failed-reboot-kernel-sanitized.log

nvidia-610-failed-reboot-relevant-sanitized.log

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions