vkCreateDevice fails with SEGV_ACCERR (ioctl → close → crash) on RTX 4090 + Radeon 610M – reproduces on both 615.78.08 and 610.57.04 (open modules)

Open
  • Is something wrong?
#1,434 1 comment 1 reaction 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
5/5
Estimated time
Over a week
Newbie friendliness
8/100
Issue type
Bug
Clarity
Needs clarification
Activity status
Active
Tech stack
c, linux
Domain
operating-systems

Research direction

The crash is a SEGV_ACCERR after an ioctl and close on a DMA-BUF file descriptor during vkCreateDevice, and the payload names no source file, function or test in the open kernel module tree. Reproduce with vulkaninfo --summary and capture the strace to confirm the ioctl command; the issue does not yet establish which component handles it. Done would mean a root cause identified in the kernel module or userspace path, which needs deep driver knowledge and is not a newcomer task.

Written by the indexing model from the issue text.

Description

bug
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

Nyarch Linux x86_64 (KDE Plasma Wayland)

Kernel Release

Linux BowsetteThroneRoom-Linux 6.18.55-1-lts #1 SMP PREEMPT_DYNAMIC Sat, 03 Oct 2026 15:08:01 +0000 x86_64 GNU/Linux

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 4090 (UUID: GPU-b11a195f-8e27-1a7a-667e-0475e37d773c)

Describe the bug

vkCreateDevice fails with a segmentation fault (SEGV_ACCERR) on an RTX 4090 + Radeon 610M system when using the NVIDIA open kernel modules. The failure is 100% reproducible with vulkaninfo --summary and any Vulkan application. The crash occurs after an ioctl + close on a DMA-BUF related file descriptor. The exact same strace pattern and crash happens on both driver versions tested:
nvidia-open-dkms 615.78.08
nvidia-open-dkms 610.57.04 (clean install after full purge)

nvidia-smi works normally and there are no Xid errors in dmesg at the time of the crash. The issue started after recent system updates; Vulkan previously worked.

To Reproduce
  1. Boot into a Wayland session (KDE Plasma / KWin) on a system with RTX 4090 + Radeon 610M.
  2. Ensure nvidia-open-dkms is loaded (tested on both 615.78.08 and clean 610.57.04).
  3. Run: vulkaninfo --summary
  4. Observe: textERROR: [Loader Message] Code 0 : vkCreateDevice: Failed to create device chain.
    Segmentation fault (core dumped)

The crash is immediate and 100% reproducible.

Additional isolation performed:

GSP firmware disabled (NVreg_EnableGpuFirmware=0)
Hybrid and discrete modes (via envycontrol)
Forced NVIDIA ICD only
All NVIDIA layers disabled
Full purge + clean reinstall of 610.57.04
Tested on both kernel 7.2.9 and 6.18.55-1-lts

Bug Incidence

Always

nvidia-bug-report.log.gz

nvidia-bug-report.log.gz

More Info

This started happening very recently and it occurs on the bleeding edge (615.78.08), and most recent stable (615.71.09) drivers too. And I was getting this issue on both current 6.18.55-1-lts kernel and 7.2.9 kernel.

01:00.0 VGA compatible controller [0300]: NVIDIA Corporation AD102 [GeForce RTX 4090] [10de:2684] (rev a1)
Subsystem: ZOTAC International (MCO) Ltd. Device [19da:4675]
Kernel driver in use: nvidia
Kernel modules: nouveau, nvidia_drm, nvidia

0d:00.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Granite Ridge [Radeon 610M] [1002:13c0] (rev cb)
Subsystem: ASUSTeK Computer Inc. Device [1043:8877]
Kernel driver in use: amdgpu
Kernel modules: amdgpu

Strace signature (identical on 615.78.08 and 610.57.04):
textioctl(25, _IOC(_IOC_READ|_IOC_WRITE, 0x46, 0xcf, 0x10), ...) = 0
close(25) = 0
--- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_ACCERR, si_addr=0x...} ---
Environment:

Distro: Nyarch Linux (Arch-based)
Desktop: KDE Plasma Wayland
Only NVIDIA ICD present (/usr/share/vulkan/icd.d/nvidia_icd.json)
nvidia-smi works, GPU is healthy, no Xid errors during the crash
Issue appeared after recent package updates (previously functional)

Current NVIDIA packages:
local/cuda 13.4.2-1
local/egl-gbm 1.1.4-1
local/egl-wayland 4:1.1.24-1
local/egl-wayland2 1.0.2-1
local/egl-x11 1.0.6-1
local/envycontrol 3.5.2-2
local/lib32-nvidia-utils 610.57.04-1
local/libvdpau 1.5-4
local/linux-firmware-nvidia 20260916-1
local/nvidia-open-dkms 610.57.04-1
local/nvidia-utils 610.57.04-1

Happy to provide nvidia-bug-report.log.gz, full strace, or any additional logs.

Expected Behavior
Vulkan working system wide. Games/Apps on Vulkan/Dx9-10-11-12 able to detect capable card (RTX 4090) and open up

Dominant language
C
Stars
17.5k
Forks
1.9k
No merged PRs in 30d

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from NVIDIA/open-gpu-kernel-modules

All issues in NVIDIA/open-gpu-kernel-modules

Similar issues

More C issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.