DisplayLink causing ethernet dropout

[Unorthodox]

Environment
- Machine: ASUS ExpertBook B9450CEA, Intel i7-1165G7, Windows 11 (domain-joined)
- Dock: Lenovo ThinkPad Hybrid USB-C with USB-A Dock (VID_17E9, PID_6015)
- MI_00 / MI_01: DisplayLink display interfaces
- MI_02: Lenovo USB Audio
- DisplayLink version: 12.1.2591.0 (oem2.inf), previously 12.2.2204.0
- Ethernet adapter: Lenovo USB Ethernet via dlcdcncm.inf / oem49.inf

---

Problem

Intermittent WAN dropout on the dock's ethernet adapter. LAN connectivity and IP address remain intact, but default gateway / DNS resolution over WAN fails. Wi-Fi is unaffected. Disabling and re-enabling the Lenovo USB Ethernet adapter restores connectivity temporarily.

The dropout correlates exactly with a flood of **UserModePowerService Event ID 12** entries in the System event log, occurring every second:

> Process C:\Windows\System32\WUDFHost.exe (PID: XXXX) reset policy scheme from {381b4222-f694-41f0-9685-ff5bb260df2e} to {381b4222-f694-41f0-9685-ff5bb260df2e}

The offending WUDFHost process has an **empty DeviceGroupId** (`-DeviceGroupId:`) in its command line, meaning it has no device bound to it.

---

What we have observed

On every boot, several orphan WUDFHost processes spawn:
- 1 × `WudfDefaultDevicePool` — legitimate
- 1 × `DlDevices` — legitimate DisplayLink display driver (serves both MI_00 and MI_01)
- 3 × `(none)` / empty DeviceGroupId — **orphans**

All of them use `-HostGUID:{193a1820-d9ac-4997-8c55-be817523f6aa}`.

The WUDFDiagnosticInfo registry keys for both MI_00 and MI_01 correctly point to the `DlDevices` WUDFHost PID — confirming both display interfaces are properly served. The 3 orphans are not serving any device.

One of the 3 orphans (the one accumulating CPU time) is the process flooding the power scheme resets. Killing it causes it to immediately respawn via services.exe (SCM), and the resets continue on the new PID.

---

What we have tried

1. Disabled USB selective suspend and power management on the ethernet adapter
2. Ran DisplayLink Cleaner in Safe Mode — removed old 11.4 audio driver
3. Blocked Windows Update driver installs for MI_02 (USB\VID_17E9&PID_6015&MI_02) via domain GPO to prevent audio driver conflicts
4. Identified and deleted stale `dlidusb` WUDF service key — this fixed the ethernet dropout but broke monitors (CM_PROB_FAILED_ADD error 31 on MI_00/MI_01)
5. Attempted manual recreation of the `dlidusb` WUDF service key — monitors would not recover until a full reboot with the key absent, at which point Windows recreated it correctly
6. Ran MSI repair (`/fa`) — completed without error but did not recreate the dlidusb key
7. Tested both DisplayLink 12.2.2204.0 and 12.1.2591.0 — orphan processes and power resets occur on both versions
8. Confirmed only 2 device instances exist for VID_17E9 (MI_00 and MI_01 at instance `7&394E5318&2&xxxx`) — no stale phantom device nodes
9. MI_02 has no WUDF subkey in its Device Parameters

---

Current state

DisplayLink 12.1.2591.0 installed. Monitors working. Ethernet connected but will drop intermittently due to the orphan WUDFHost power resets.

What is spawning 3 WUDFHost processes with empty DeviceGroupId on every boot when only 2 DisplayLink display devices (MI_00, MI_01) are present — and why does one of them flood UserModePowerService Event ID 12 power resets?

Any guidance would be appreciated. Happy to provide additional logs or registry exports.