[REGRESSION] Kernel panic during suspend on Pascal MX150 with NVIDIA 580.142 and Linux 7.0.0-15

Graphics / Linux Linux
, , ,
1

Environment:

  • Distribution: Ubuntu 26.04 LTS

  • Kernel: 7.0.0-15-generic

  • GPU: NVIDIA GeForce MX150 (Pascal)

  • NVIDIA Driver: 580.142

  • Session: X11

  • Laptop: Xiaomi TM1703 (Kaby Lake platform)

Problem Description:
System consistently crashes during suspend.

The crash occurs immediately after entering suspend state and results in a black screen with kernel trace output.

Relevant kernel trace:
kernel BUG at arch/x86/kernel/jump_label.c:73
nvkms_kthread_q_callback [nvidia_modeset]

Observed Behavior:

  • suspend triggers kernel panic

  • system becomes unusable

  • requires hard reboot

  • issue is fully reproducible

Reproduction:

  1. Boot Linux 7.0.0-15

  2. Load NVIDIA 580.142 driver

  3. Execute:
    systemctl suspend

  4. Kernel panic occurs

Additional Testing:
The issue reproduces under:

  • PRIME nvidia mode

  • PRIME on-demand mode

  • s2idle suspend

  • deep suspend

  • nvidia-drm modeset=0

Workarounds Tested:
Successful workaround:

  • booting Linux 6.17.0-29-generic

  • same NVIDIA driver stack

  • suspend no longer causes kernel panic

Additional note:
On this hardware, deep suspend (mem_sleep_default=deep) is required for proper sleep behavior. With s2idle, the system does not enter real suspend correctly.

Regression Information:
Suspend worked correctly on previous Ubuntu LTS installation before upgrading to Ubuntu 26.04 / kernel 7.0 stack.

Suspected Cause:
Possible regression in suspend/resume interaction between:

  • Linux kernel 7.0

  • NVIDIA 580.x proprietary driver

  • Pascal mobile GPUs

Relevant modules involved:

  • nvidia

  • nvidia_modeset

  • nvidia_drm

2

I get a similar kernel panic (from nvkms_kthread_q_callback [nvidia_modeset]) on OpenSuse Tumbleweed with kernel 6.19.12-1 and nvidia drivers 580.159.03. My GPU is GTX 1650 Mobile (Turing).