Hard GPU lock (Xid 8) when launching gamescope while OBS is actively streaming on RTX 5060 Ti (driver 580.142)

[KeyofBlueS]

NVIDIA Open GPU Kernel Modules Version

580.142

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

Debian GNU/Linux forky/sid

Kernel Release

Linux debian 6.19.8+deb14-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.19.8-1 (2026-03-15) 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

GPU 0: NVIDIA GeForce RTX 5060 Ti (UUID: GPU-b36a9ed8-95ab-a0e0-bf49-5d3bccb1cf56)

Describe the bug

Other System Information

• Display Server: X11
• Desktop Environment: XFCE 4.20
• OBS Version: 32.1.0 (flatpak)
• gamescope Version: 3.16.20+ds-1

The system hard-freezes when launching gamescope while OBS Studio is actively streaming.

The issue occurs regardless of the selected encoder (NVENC or x264). The freeze can be triggered by:

• simply starting gamescope
• resizing the gamescope window
• maximizing the gamescope window

No interaction with the application inside gamescope is required (e.g., reproduces with vkcube).

The freeze is intermittent but occurs frequently under repeated tests.

During the freeze:

• the system becomes unresponsive
• the GPU reports Xid 8

Magic SysRq (REISUB) still works, indicating the kernel remains responsive.

A hard reboot (or SysRq reboot) is required to recover.

From journalctl -b -1:

mar 17 04:41:16 debian kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked!  Notify Timeout Seconds: 7
mar 17 04:41:16 debian kernel: NVRM: GPU at PCI:0000:01:00: GPU-b36a9ed8-95ab-a0e0-bf49-5d3bccb1cf56
mar 17 04:41:16 debian kernel: NVRM: GPU Board Serial Number: 0
mar 17 04:41:16 debian kernel: NVRM: Xid (PCI:0000:01:00): 8, pid=18099, name=obs, channel 0x0000003f
mar 17 04:41:24 debian kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked!  Notify Timeout Seconds: 7
mar 17 04:41:34 debian kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked!  Notify Timeout Seconds: 7
mar 17 04:41:43 debian kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked!  Notify Timeout Seconds: 7
mar 17 04:41:51 debian kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked!  Notify Timeout Seconds: 7
mar 17 04:42:00 debian kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked!  Notify Timeout Seconds: 7
mar 17 04:42:08 debian kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked!  Notify Timeout Seconds: 7
mar 17 04:42:16 debian kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked!  Notify Timeout Seconds: 7
mar 17 04:42:24 debian kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked!  Notify Timeout Seconds: 7
mar 17 04:42:33 debian kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked!  Notify Timeout Seconds: 7
mar 17 04:42:41 debian kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked!  Notify Timeout Seconds: 7
mar 17 04:42:49 debian kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked!  Notify Timeout Seconds: 7
mar 17 04:42:54 debian kernel: sysrq: HELP : loglevel(0-9) reboot(b) crash(c) terminate-all-tasks(e) memory-full-oom-kill(f) kill-all-tasks(i) thaw-filesystems(j) sak(k) show-backtrace-all-active-cpus(l) show-memory-usage(m) nice-all-RT->
mar 17 04:42:55 debian kernel: sysrq: Keyboard mode set to system default
mar 17 04:42:56 debian kernel: sysrq: This sysrq operation is disabled.
mar 17 04:42:57 debian kernel: sysrq: This sysrq operation is disabled.
mar 17 04:42:57 debian kernel: NVRM: krcWatchdog_IMPL: RC watchdog: GPU is probably locked!  Notify Timeout Seconds: 7
mar 17 04:42:59 debian kernel: sysrq: Emergency Sync
mar 17 04:42:59 debian kernel: Emergency Sync complete
mar 17 04:42:59 debian kernel: sysrq: Emergency Remount R/O
mar 17 04:42:59 debian kernel: EXT4-fs (nvme0n1p4): re-mounted 971bbf01-r2fc-4652-8a5f-6bf7e62f05f2 ro.

To Reproduce

1 - Start OBS Studio
2 - Begin streaming (tested with both NVENC and x264)
3 - Launch:

gamescope -W 1920 -H 1080 -r 60 -- vkcube

OR BETTER run this "stress" loop:

#!/bin/bash

while true; do
    gamescope -W 1920 -H 1080 -r 60 -- vkcube
    sleep 1
    killall -w gamescope-wl
done

Bug Incidence

Sometimes.
The freeze is intermittent but occurs the majority of the time.

nvidia-bug-report.log.gz

nvidia-bug-report.log.gz

More Info

Please let me know what additional information or logs you need to help track down and resolve this issue.

[KeyofBlueS]

The freeze has now also occurred when OBS was not running at all.

This indicates that OBS is not a requirement to trigger the issue, but when OBS is actively streaming, the probability of the freeze increases significantly.

The output of journalctl -b -1 is the same as previously reported, except for the process associated with the Xid 8 error:

• Previously: name=obs
• More recently: name=compiz

This suggests that the issue is not tied to OBS specifically, but instead affects whichever process owns the GPU channel at the time of the stall.
I attempted to reproduce the issue multiple times in this state, but without success.

I also tested the newer 595.45.04 beta driver and observed different behavior:

• The hard system freeze no longer occurs
• However, applications crash instead
• This happens even without gamescope and both with and without OBS running

After this test, I reverted back to driver version 580.142.

Please let me know if there are specific debug options, logs, or tests you would like me to run to help further isolate the issue.