Xid 120 GSP crash (RTX 3050 Ti Laptop) — reproducible, matches recently closed forum thread

Hi,

I’m reporting a reproducible Xid 120 → Xid 154 GPU reset crash. I found a closed forum thread describing what looks like the same root cause (GSP crash tied to EDID/DP dongle reads on Ampere laptop GPUs with external displays), but couldn’t reply there since it’s locked, so I’m sending this report directly.

Hardware: Acer Nitro 5, NVIDIA GeForce RTX 3050 Ti Laptop GPU (GA107, Ampere), Intel Tiger Lake iGPU (Optimus hybrid)
Driver: 595.84 (open kernel module, nvidia-open-dkms)
OS: Nobara Linux (Fedora 44 based), KDE Plasma 6 (Wayland)
Kernel: 7.1.4-200.nobara.fc44.x86_64

NVRM: Xid (PCI:0000:01:00): 120, GSP kernel exception: store access page fault (cause:0xf) @ pc:0xffffffff93006a9c, partition:2#0
NVRM: Xid (PCI:0000:01:00): 154, GPU recovery action changed from 0x0 (None) to 0x1 (GPU Reset Required)

This has occurred 3 separate times across different boots/days, always at the exact same pc address (0xffffffff93006a9c) — fully deterministic, not random corruption. In one occurrence it was preceded by two Xid 109 (CTX SWITCH TIMEOUT) events a few minutes earlier.

Trigger pattern: fullscreen mode transitions and external monitor handling. I’ve also had a related crash where the external monitor lost signal entirely during gameplay.

nvidia-bug-report.log.gz is attached, generated right after the most recent occurrence.

Thanks

nvidia-bug-report.log.gz (1.2 MB)