### System details

Omarchy 4.0.1-1, Quickshell 0.3.1-1, Qt 6.11.2 / qt6-wayland 6.11.2-1, Hyprland 0.56.2-1, nvidia-open-dkms 610.57.04 (GB203 / RTX 5070 Ti, Wayland direct, GSP), linux 7.1.9-arch1-2. Triple monitors: DP-1 LG ultrawide + DP-2/DP-3 dual ASUS VY279HF via DisplayPort. uwsm session.

Related: #7380 (same fatal-error signature on Intel), #8453 (same trigger, EGL abort variant), #9262, #8863.

### What's wrong?

Since upgrading to 4.0.1 on Aug 30 (omarchy 4.0.0→4.0.1, quickshell 0.3.0.r20→0.3.1, kernel 7.1.8→7.1.9), every *unattended* idle→screensaver→lock transition kills omarchy-shell with a fatal Wayland protocol error, and on my NVIDIA triple-DP setup the aftermath is much worse than the flash described in #7380: after the shell relaunches and re-locks, the kernel logs `nvidia-modeset: WARNING: GPU:0: Unable to read EDID` for the ASUS VY279HF panels, the outputs come back with no valid mode, and the desktop is unusable until reboot. Hyprland itself survives; the shell + monitor link do not.

This did not happen on 4.0.0. First occurrence ever in the journal: Aug 30 14:46, exactly the day of the 4.0.1 upgrade. 8 occurrences total, each pairing shell-fatal + kernel EDID failure within ~1s.

### Steps to reproduce

1. Triple-monitor NVIDIA (GB203) + DP, `idle.screensaver=150`, `idle.lock=300`.
2. Let it sit idle through screensaver → lock.
3. Kernel logs EDID unreadable; shell relaunches with stranded-lock recovery; monitor link dead until reboot.

Not reproducible manually: explicit `dpms off/on` and manual `omarchy-shell lock lock` both complete cleanly — only the unattended idle transition with the shell mid-screensaver handling dies.

### Representative timeline (Aug 31 20:54, journalctl)

```text
20:54:30 omarchy-shell[51314]: shell restarting (previous instance mid-lock cycle)
20:54:33 uwsm_hyprland.desktop[1566]: error in client communication (pid 51314)
20:54:33 omarchy-shell[51314]:   WARN: The Wayland connection experienced a fatal error: Invalid argument
20:54:33 omarchy-shell[63787]: Omarchy shell exited with status 255; relaunching.
20:54:36 omarchy-shell[64036]: DEBUG qml: omarchy lock ... lock-stranded: recovering
20:54:36 omarchy-shell[64036]: DEBUG qml: omarchy lock ... lock-pending: screen-stabilizing
20:54:36 omarchy-shell[64036]: DEBUG qml: omarchy lock ... secure=true
20:54:37 kernel: nvidia-modeset: WARNING: GPU:0: Unable to read EDID for display device Asustek Computer Inc VY279HF (DP-5)
```

Second variant of the same death, same boot (Aug 31 21:23 — note the EGL errors matching #8453's family):

```text
21:23:04 omarchy-shell[64036]:   WARN qt.qpa.wayland: Could not create EGL surface (EGL error 0x3003)
21:23:04 omarchy-shell[64036]:   WARN qt.qpa.wayland: eglSwapBuffers failed with 0x300d, surface: 0x0
21:23:04 omarchy-shell[64036]:   WARN: The Wayland connection experienced a fatal error: Invalid argument
21:23:05 omarchy-shell[97768]: Omarchy shell exited with status 255; relaunching.
21:23:08 kernel: nvidia-modeset: WARNING: GPU:0: Unable to read EDID for display device Asustek Computer Inc VY279HF (DP-5)
```

8 shell-fatal + EDID-loss pairs total across Aug 30-31; zero before Aug 30 14:46. Unlike #7380 there is no FALLBACK monitor event in my journal — the outputs themselves lose EDID.

### Ruled out

- Kernel: no panic/oops/Xid in any boot; all reboots were clean shutdowns.
- NVIDIA suspend plumbing: nvidia-suspend/-resume/-hibernate/-powerd disabled with `NVreg_PreserveVideoMemoryAllocations=1` (self-managed, per driver docs).
- Plain DPMS cycle: clean, no kernel warnings, monitors recover.
- Session never actually suspends — this is display-DPMS + shell churn only.

Omarchy's own `omarchy-hyprland-monitor-watch` documents this recovery mode: "Powering the monitor on fires no DRM hotplug... only a reload re-reads the EDID... a compositor that stays silent has gone, taking the session and this loop's reason with it."

### System info

- Omarchy 4.0.1-1, quickshell 0.3.1-1, Hyprland 0.56.2-1 (same commit as #7380)
- nvidia-open-dkms / nvidia-utils 610.57.04-1, Qt 6.11.2, kernel 7.1.9-arch1-2
- AMD Ryzen 9 3900X, ASUS RTX 5070 Ti (GB203)

### Workaround in place

`idle.lock` raised in `~/.config/omarchy/shell.json` to skip unattended lock until a fix lands (manual lock + manual DPMS are unaffected).