## System details

- Omarchy 4.0.0-1
- Hyprland 0.56.2 (`efb50993780079460b0cbed1363e2166a2de1d9f`)
- NVIDIA GeForce RTX 4090 (driver 610.57.04 open) + AMD Raphael iGPU (`amdgpu`)
- Two monitors on **different GPUs**:
  - DP-1 / `card1` (NVIDIA): Samsung Odyssey G8, 3840x2160@240, scale 1.6
  - DP-4 / `card0` (AMD iGPU): Corsair XENEON EDGE, 2560x720@60, scale 1.6, physically below the G8

## What's wrong?

Idle lock works. Typing the password on the G8 authenticates and the G8 shows the desktop. The XENEON EDGE stays on the Quickshell lock field (`Enter Password`) indefinitely.

The lock plugin itself thinks the session is unlocked:

```
omarchy-shell lock status
{"locked":false,"requested":false,"pending":false,"sessionLocked":false,"secure":false,"realScreens":2,...,"lastEvent":"unlocked"}
```

`loginctl` `LockedHint=no`. `omarchy-hyprland-session-locked` exits 1 (unlocked). `hyprctl layers` on DP-4 shows only `omarchy-background` and `omarchy-bar` — no lock overlay. Session-lock surfaces are not layer-shell, so that last point does not prove the leftover is gone, but the IPC/logind state does.

`grim -o DP-4` hangs until killed. `grim -o DP-1` returns immediately. Capturing a region that sits on the Edge layout (`-g "400,1350 20x20"`) also hangs. So the AMD output is in a bad copy/scanout state after unlock, not just a stale wallpaper.

## Journal (idle lock, then unlock on the G8)

```
15:29:04 lock-requested
15:29:04 lock-pending: screen-stabilizing
15:29:09 secure=true
15:31:34 idle-monitor: idle          # idle cycle starts *while already locked*
15:34:04 lock-system: lock-timeout   # second omarchy-system-lock while already locked
16:05:16 idle-monitor: active        # mouse wake, ~36 min later
16:05:29 PAM omarchy-lock-password authenticated successfully
16:05:29 secure=false
16:05:29 session-locked=false
16:05:29 unlocked
```

There is no `session-locked=true` line between `lock-requested` and `secure=true`.

`omarchy-system-wake` then no-ops: `omarchy-brightness-display on` skips the DPMS enable when every monitor already reports `dpmsStatus: true`. After the mouse wake, both outputs report on, so unlock never modesets the AMD panel.

## Steps to reproduce

1. Mixed-GPU desktop: one monitor on NVIDIA, one on the AMD iGPU. The iGPU panel here is a 2560x720 XENEON EDGE.
2. Wait for idle lock (`idle.lock` 300s). Displays blank ~5s later (`idleBlankTimer`).
3. Move the mouse / type the password on the NVIDIA monitor.
4. NVIDIA monitor unlocks. AMD monitor keeps showing `Enter Password`.

## Related

- #2449 was the hyprlock-era dual-monitor leftover. This is Quattro / `WlSessionLock`.
- #7399 is the DPMS wake skip-guard on a slow-sync panel. Same `omarchy-brightness-display on` guard, different symptom.

## Local workaround that recovered the panel

Recreating only DP-4:

```lua
hl.monitor({ output = "DP-4", disabled = true })
-- then
hl.monitor({ output = "DP-4", mode = "2560x720@60", position = "400x1350", scale = 1.6, transform = 0, disabled = false })
```

plus `omarchy restart shell`. I also started a Hyprland `unlock>>` listener that repeats that DP-4 toggle so the next idle unlock does not stick again.

## Suggestions

- Do not start a new idle lock cycle while `omarchy-shell lock isLocked` is already true (the 15:31/15:34 double lock above).
- `omarchy-brightness-display on` should not treat `dpmsStatus: true` as "this output actually has a fresh frame", at least on a secondary GPU.
- After `finishUnlock()`, force a modeset (or `hl.dsp.force_renderer_reload()`) on outputs that did not render the PAM field, or skip DPMS-off on mixed-GPU outputs.