Regression: Resume failure after long s2idle on Dell Precision 7780 (RTX 3500 Ada, 610.43.03)
Summary
I’m experiencing intermittent resume failures after the laptop has been suspended for several hours. The system resumes to a black screen and becomes completely unresponsive, requiring a hard power-off.
Initially the kernel backtrace pointed to nvEvbDisableVblankSemControl() inside nvidia_modeset.
Following a recommendation from the NVIDIA Linux Developer Forum, I tested:
options nvidia_modeset vblank_sem_control=0
The issue still occurs. After applying the workaround, the kernel call trace changed to nv_set_system_power_state.part.0 → nv_pm_notifier → notifier_call_chain, suggesting the workaround changes the driver’s execution path but does not resolve the underlying resume failure.
Here is the link to the bug report too: Log in | NVIDIA Developer
Hardware
Dell Precision 7780
BIOS: 1.28.2
CPU: Intel Core i9-13950HX
Integrated GPU: Intel UHD Graphics (Raptor Lake)
Discrete GPU: NVIDIA RTX 3500 Ada Laptop GPU (AD104GLM)
Bluetooth/Wi-Fi: Intel AX211
Graphics mode: PRIME Render Offload (Intel drives displays)
Software
Fedora Linux 44 KDE Plasma
Kernel: 7.1.4-200.fc44.x86_64
KDE Plasma 6.7.3 (Wayland)
NVIDIA Driver: 610.43.03 (RPM Fusion akmod-nvidia)
nvidia_drm.modeset=1
NVIDIA suspend/resume services enabled
System supports only s2idle (no deep/S3).
Reproduction
1. Boot normally.
2. Use the machine normally.
3. Suspend.
4. Leave suspended for several hours (typically overnight).
5. Resume.
The issue is intermittent but much more likely after long suspend durations.
Observed behaviour
- Black screen after resume.
- Complete system lockup.
- Bluetooth keyboard and mouse reconnect messages appear only after the failure.
- Magic SysRq does not recover the machine.
- Hard power-off required.
Troubleshooting performed
• Reproduced with and without Dell dock.
• Reproduced without Bluetooth peripherals.
• Confirmed latest Dell BIOS (1.28.2).
• Confirmed latest Fedora updates.
• Confirmed clean RPM Fusion NVIDIA installation.
• Verified matching kernel (7.1.4) and NVIDIA module (610.43.03).
• Ruled out Bluetooth, Intel AX211, Keychron devices and IRQ conflicts.
• Tested NVIDIA workaround (vblank_sem_control=0); issue persists.
Expected behaviour
The system should resume normally from s2idle without hanging.
Request
Is this a known regression affecting Ada mobile GPUs and driver 610.43.03?
I’m happy to test patches, debug kernels or additional logging if requested.
Attachments
1. journalctl -k -b -1 from failed resume
2. Original photograph of first kernel backtrace (nvEvbDisableVblankSemControl)
3. inxi -Fzxx output
inix-output.txt (7.6 KB)
resume-after-workaround.log (199.5 KB)
