### System details

Lenovo Slim Pro 7 14ARP8, AMD Ryzen (Rembrandt APU, amdgpu YELLOW_CARP) integrated GPU driving the panel, NVIDIA GeForce RTX 3050 as an offload-only discrete GPU (no display output — hybrid/Optimus-style config). Omarchy 4.0.4-1, Hyprland 0.56.2, Quickshell 0.3.1-1, Qt 6.11.2, NVIDIA driver 610.57.04, Linux 7.2.5-3-omarchy. Suspend mode is s2idle (Low-power S0 idle).

### What's wrong?

After closing and reopening the laptop lid, the entire screen displays rainbow/color corruption. The lock screen is visible just enough through the corruption to type the password and unlock, but the corruption persists into the desktop afterward — the only reliable recovery is a full reboot.

A photo of the corruption is attached. It's not random static — it's a structured pattern of concentric colored contour lines/moiré rings warping the screen, with the underlying UI (a notification pill) faintly visible through it. That specific ring pattern looks like a GPU tiling/stride misinterpretation (a compressed or tiled framebuffer being scanned out as if it had a different layout), not simple random noise.

### Steps to reproduce

1. Close the laptop lid to suspend (s2idle).
2. Reopen the lid to resume.
3. Screen shows rainbow/color corruption instead of a normal lock screen.
4. Password entry works (barely visible), but corruption remains after unlocking into the desktop.

### What's been ruled out

- **Not a kernel/driver-level resume failure.** `journalctl -b -1` for the exact suspend/resume window shows a clean resume: `amdgpu` SMU resumes successfully, GART re-enabled, all render rings re-established, no DRM errors; `nvidia-resume.service` also completes with no errors.
- **Not fixed by NVIDIA suspend/resume tuning** — enabled `NVreg_PreserveVideoMemoryAllocations=1` plus `nvidia-suspend.service`/`nvidia-resume.service`/`nvidia-hibernate.service`, rebuilt the UKI. No change to the corruption (expected in hindsight, since NVIDIA isn't driving the panel in this hybrid config — `nvidia 0000:01:00.0: [drm] No compatible format found` / `Cannot find any crtc or sizes` confirms it has no display output here).
- **Not fixed by restarting the Omarchy shell.** Ran `omarchy-restart-shell` after unlocking through the corrupted lock screen — Quickshell visibly restarted (brief blank), but the rainbow corruption remained afterward. This suggests the corruption isn't in Quickshell's own render buffers/EGL context.
- **Not fixed by resetting the display output.** Ran `hyprctl keyword monitor eDP-1,disable` followed by `hyprctl reload` (forcing amdgpu/Hyprland to fully redo mode-setting on the panel). No change.

Given the above, the corruption looks like it's either in Hyprland's own compositor renderer state or the amdgpu DRM/KMS scanout buffer itself, surviving both a shell-level and an output-level reset, and only cleared by a full reboot.

### Expected behavior

The panel renders correctly after lid-resume, or at minimum a lighter-weight recovery than a full reboot (e.g. `omarchy restart hyprland`, untested so far) resolves it.

### Additional notes

Searched existing issues for a matching report (resume/lock/AMD/color-corruption keywords) and found several other resume-related bugs but nothing matching this specific whole-screen rainbow corruption that survives both a shell restart and a display reset.


## omarchy-debug
 https://logs.omarchy.org/JfTHOm

<img width="4000" height="2252" alt="Image" src="https://github.com/user-attachments/assets/20e08f97-4fa2-481f-9836-2ebdd6a29acd" />