### System details

- Omarchy 4.0.4-1 (updated from 4.0.3-1 earlier on the day of the incident)
- Kernel currently running after recovery: `7.2.5-3-omarchy`; `linux-omarchy 7.2.5-3` and headers installed during that update
- Previous stock kernel `linux 7.2.3.arch1-3` remains installed; no controlled comparison performed yet
- NVIDIA GeForce RTX 5090, 32 GB; `nvidia-open-dkms 610.57.04-1`, `nvidia-utils 610.57.04-1`
- NVIDIA DKMS module built successfully for the new kernel during the update; NVIDIA packages themselves were not upgraded that day
- AMD Ryzen Threadripper 9960X, 128 GB RAM, ASUS Pro WS TRX50-SAGE WIFI
- Hyprland 0.56.2-2 / Wayland
- ASUS PG32UCDP via DisplayPort, configured for 3840x2160 at 240 Hz, 10-bit, fullscreen-only VRR

### What's wrong?

On the first day after updating to Omarchy 4.0.4 and installing linux-omarchy, the desktop display became a solid grey screen with regularly spaced vertical blue lines during ordinary desktop use (chatting in the ChatGPT/Codex desktop app, no game being played). The machine was rebooted and the display recovered.

This is a report of an observed graphics failure and severe kernel-log flooding after the update, **not a claim that the new kernel is the confirmed cause**. NVIDIA driver/firmware or hardware involvement remains possible. The kernel version of the failed boot cannot be independently recovered from its surviving kernel log; the post-reboot kernel and package history are verified.

#### Expected behavior

The desktop should remain usable during normal desktop activity. A GPU failure should not produce millions of repeated messages that displace the initial diagnostic evidence.

#### Reproduction status

One observed incident; no deterministic reproducer. Sequence: update, reboot into the updated system, use the desktop for several hours, then encounter the corrupted display during ordinary desktop use. No intentional stress test was running. App launch records are context, not attribution of fault.

#### Deidentified log evidence

The following retains exact NVIDIA error text. Hostnames, usernames, absolute dates/times, timezone, process IDs, boot IDs, device identifiers, home paths and crash-upload identifiers are omitted. Relative times are measured from the first **retained** kernel message, not proven failure onset.

```text
T+0.000000  kernel: NVRM: gpuHandleSanityCheckRegReadError_GH100: Possible bad register read: addr: 0xb830b0,  regvalue: 0xbadf3200,  error code: Unknown SYS_PRI_ERROR_CODE
T+0.008030  kernel: NVRM: gpuHandleSanityCheckRegReadError_GH100: Possible bad register read: addr: 0xb830b0,  regvalue: 0xbadf3200,  error code: Unknown SYS_PRI_ERROR_CODE
[identical message repeats]
T+49.729614 kernel: NVRM: gpuHandleSanityCheckRegReadError_GH100: Possible bad register read: addr: 0xb830b0,  regvalue: 0xbadf3200,  error code: Unknown SYS_PRI_ERROR_CODE
```

A full streaming pass over the previous boot's retained kernel journal counted **6,900,872 entries, with exactly one distinct message body**: the error above. Approximately 4 GB of journal files were present, overwhelmingly system-journal archives created during the flood. Earlier kernel messages are no longer retained, plausibly displaced by rotation during this flood. Consequently, no initial Xid, panic, OOM or GSP failure can be established or excluded from that missing interval.

Separately retained user-session excerpts:

```text
~T-5m16s  desktop session unlocked
~T-4m57s  ChatGPT desktop launch request
~T-2m42s  Chrome launch request
~T+16s    steam: src/clientdll/steamengine.cpp (2665) : Assertion Failed: CSteamEngine::BMainLoop appears to have stalled > 15 seconds without event signalled
```

The Steam stall follows the retained NVIDIA errors; it does not establish Steam as the trigger. No systemd coredump was found in the incident window.

#### Update and recovery checks

Other relevant same-day changes were Chrome `152.0.7977.82-1 -> 153.0.8010.47-1` and ChatGPT desktop `26.901.51231-1 -> 26.908.70816-1`. NVIDIA stayed at 610.57.04.

After reboot, `nvidia-smi` responds normally; sampled GPU temperature was 37 C and VRAM use approximately 1.35 GiB. Current-boot GPU PCIe AER counters were zero and no new NVIDIA error/Xid was observed during inspection. These post-reboot readings do not rule out crash-time resource, temperature or hardware problems.

A nearby report, #12187, concerns missing prebuilt NVIDIA modules on a hybrid laptop. This system uses DKMS, successfully built the module, has a single discrete GPU, and emitted NVIDIA register errors; that report does not establish a matching cause here.

There are related Blackwell register-error/idle-hang reports in the NVIDIA forum, but this incident has insufficient retained evidence to identify the same underlying bug:
https://forums.developer.nvidia.com/t/deterministic-xid-120-gsp-kernel-panic-on-rtx-pro-6000-blackwell-under-sustained-cublas-workload-595-84/376750

Diagnostics were collected locally with `omarchy-debug --no-sudo --print` (the spaced `omarchy debug` alias was unavailable). Relevant details are included above; UEFI version is 1317 and the CPU governor is performance. The full diagnostic bundle is intentionally omitted because it includes unrelated identifying system data. The original camera photo is not attached because it contains surroundings and may contain identifying metadata; its screen contents are described above.

No kernel rollback, driver change or configuration workaround has been applied. Please advise whether the linux-omarchy transition warrants a comparison with the stock kernel, or whether this should be tracked upstream with NVIDIA.

Filed by GPT-6 via Codex, at the user's explicit request.