## What happened

Device idled normally (screensaver at 120s, lock at 150s per `~/.config/omarchy/shell.json`). After ~15 minutes idle, pressing a key to wake did not show the lockscreen — monitors stayed black. Two blind password attempts were typed (unlock worked previously with same setup). System then fully froze — no further journal entries at all, required a hard powercycle.

## Expected

Pressing a key after idle should wake the display and show the hyprlock prompt.

## Log evidence (journalctl, previous boot)

Idle/lock cycle started normally:
```
15:22:44 idle-monitor: idle
15:22:44 idle-cycle-start: screensaver=120 lock=150
15:23:14 lock-armed: lock-timeout
```

On wake attempt, the lock surface failed to attach to a real output:
```
15:28:22 idle-monitor: active
15:28:22 lock-system: activity-after-grace
15:28:22 process-start: lock omarchy-system-lock
15:28:24 omarchy-shell: WARN: Layershell screen does not correspond to a real screen. Letting the compositor pick.
15:28:24 omarchy-shell: DEBUG: Not creating lock surface for screen QScreen(0x7c4e77e08ca0, name="") as it is not backed by a valid wayland output.
15:28:25 omarchy-shell: WARN quickshell.hyprland.ipc: Got removal for monitor "FALLBACK" which was not previously tracked.
```
(repeats several times over 15:28:24–15:28:26)

Two blind unlock attempts against the black screen:
```
15:28:31 quickshell.service.pam.subprocess: Starting pam session ... 
15:28:31 unix_chkpwd: password check failed for user (matthew)
15:28:34 quickshell.service.pam.subprocess: Failed to authenticate.
15:29:27 quickshell.service.pam.subprocess: Starting pam session ...
15:29:27 quickshell.service.pam.subprocess: Relaying pam message: "Password: " ...
```
Journal for that boot ends mid this second attempt — no shutdown/reboot entry follows. This is a hard freeze, not a clean shutdown. Kernel log (i915) shows no GPU reset/hang error for this period, so the freeze appears to originate in the compositor/lock-surface path rather than the GPU driver.

## Root cause (best guess from logs)

On wake from idle-DPMS-off, Hyprland appears to drop/re-enumerate the monitor as `FALLBACK` right as `omarchy-system-lock` (hyprlock via quickshell) tries to attach its lock surface. The lock surface gets no real wayland output, so nothing renders (black screen), while the session is still considered locked/waiting on auth — leaving no way to recover except a hard powercycle.

## System details

- Omarchy: 4.0.4-1
- Hyprland: 0.56.2 (tag v0.56.2, commit efb5099)
- quickshell: 0.3.1-1
- Kernel: 7.2.5-3-omarchy
- CPU: Intel Core i5-8500T
- GPU: Intel UHD Graphics 630 (CoffeeLake-S GT2, iGPU only)
- idle config: `~/.config/omarchy/shell.json` → `idle.screensaver: 120`, `idle.lock: 150` (defaults)

## Steps to reproduce

1. Leave device idle ~15+ minutes so screensaver (120s) then lock (150s) both trigger, monitors go to DPMS off.
2. Press a key to wake.
3. Intermittently: monitors stay black instead of showing hyprlock, and the session eventually hangs completely, requiring a hard power cycle.

Note: `omarchy debug --no-sudo --print` does not exist as a command on this Omarchy version (4.0.4-1) — `omarchy commands --all` has no `debug` entry, so no diagnostic bundle/URL is attached. Happy to provide fuller diagnostics if pointed at the right command for this version.