Hello,
I am experiencing a severe issue when waking my PC from suspend on Fedora Linux 43 (KDE Plasma 6, Wayland). Upon waking up, the desktop is frozen, the screen flickers heavily, and even the monitor’s built-in hardware OSD starts flickering along with it. Re-plugging the DisplayPort cable results in a permanent “No Signal” state.
System Specs:
OS: Fedora Linux 43 (Kernel 6.18.12)
DE: KDE Plasma 6 (Wayland)
GPU: NVIDIA GeForce RTX 4060 8GB (Driver: 580.119.02)
RAM: 16 GB
Display: LG UltraGear 24GS65F-B (connected via DisplayPort)
Kernel parameter nvidia.NVreg_PreserveVideoMemoryAllocations=1 is explicitly enabled.
Video Evidence:
I have recorded this behavior. In the video, you can clearly see the monitor’s own OSD flickering alongside the frozen desktop, indicating a physical sync/link collapse, not just a frozen image buffer.
The Issue & Logs:
I managed to switch to TTY during the flickering state and pulled the logs. dmesg clearly shows that the NVIDIA module runs out of memory while trying to re-allocate/restore VRAM during the resume process:
[ 7632.479447] NVRM: nvCheckOkFailedNoLog: Check failed: Out of memory [NV_ERR_NO_MEMORY] (0x00000051) returned from _memdescAllocInternal(pMemDesc) @ mem_desc.c:1359 [ 7632.479454] NVRM: nvCheckOkFailedNoLog: Check failed: Out of memory [NV_ERR_NO_MEMORY] (0x00000051) returned from rmStatus @ system_mem.c:356 [ 7632.479461] NVRM: nvAssertOkFailedNoLog: Assertion failed: Out of memory [NV_ERR_NO_MEMORY] (0x00000051) returned from pRmApi->Alloc(pRmApi, device->session->handle, isSystemMemory ? device->handle : device->subhandle, &physHandle, isSystemMemory ? NV01_MEMORY_SYSTEM : NV01_MEMORY_LOCAL_USER, &memAllocParams, sizeof(memAllocParams)) @ nv_gpu_ops.c:4968
It appears that because I have 16GB of system RAM and 8GB of VRAM, the driver runs out of memory (likely due to /tmp being mounted as tmpfs in RAM by default on Fedora).
While setting NVreg_TemporaryFilePath=/var/tmp on a physical drive serves as a workaround, the driver failing to allocate memory in tmpfs shouldn’t cause the hardware DisplayPort link to completely collapse and corrupt the monitor’s OSD sync. The driver should handle the OOM exception more gracefully.
Attached is the nvidia-bug-report.log.gz and kernel_log.txt generated directly from the TTY during the frozen/flickering state.
Disclosure: The logs were successfully extracted manually via TTY, but the root cause analysis (identifying the OOM error during VRAM restoration) and the formatting of this bug report were prepared with the assistance of an AI.
nvidia-bug-report.log (1.3 MB)
kernel_log.txt (97.4 KB)