Title: Alienware 16 Area-51 AA16250: Thunderbolt-side USB host controller crashes (Code 24) whenever a display is connected through a Thunderbolt port at runtime (regression after recent updates)

 

SUMMARY

Since a recent round of BIOS, driver, and Windows updates, connecting a monitor through either Thunderbolt port while Windows is running immediately takes down one of the two "Intel USB 3.20 eXtensible Host Controller" devices. It shows Code 24 in Device Manager, all USB devices on both Thunderbolt ports disappear, and only a reboot recovers it. Displays and Wi-Fi keep working. The same hardware setup worked without problems before the updates. I've isolated the trigger to runtime DisplayPort tunnel creation over USB4/Thunderbolt (details below).

 

SYSTEM

- Alienware 16 Area-51 AA16250

- Intel Core Ultra 200HX, HM870 chipset

- NVIDIA GeForce RTX 5070Ti, Hybrid Graphics/Advanced Optimus enabled (default)

- Windows 11 25H2

- BIOS 2.7.0

- Intel graphics driver: 32.0.101.8860, A00

- NVIDIA driver: 32.0.15.9202, A01

 

PERIPHERALS

- Anker Nano 8-in-1 USB-C hub (A210B), dual HDMI via DP Alt Mode, no Ethernet

- Anker USB-C to Gigabit Ethernet adapter (A8313)

- Two HDMI monitors

- Also tested with an earlier Anker hub that includes Ethernet; same behavior

 

SYMPTOM

- The USB-C Ethernet adapter disappears, even when it is on a different Thunderbolt port from the display.

- One of the two "Intel USB 3.20 eXtensible Host Controller" entries in Device Manager shows Code 24: "This device is not present, is not working properly, or does not have all its drivers installed."

- Unplug/replug, Disable/Enable, and Uninstall + Scan for hardware changes do not restore it. Only a restart does.

- Both external monitors, the internal display, and Wi-Fi keep working.

- The chipset USB-A ports (the other xHCI controller) are never affected.

 

ISOLATION TESTING

1. All devices and monitors connected at boot: stable indefinitely.

2. Monitors power-cycled (or HDMI switch toggled) while connected through the hub: controller crashes.

3. One monitor on the laptop's built-in HDMI port, one on the hub; monitors power-cycled: controller crashes. (MST is not required.)

4. Hub unplugged; one monitor on built-in HDMI only; monitor hotplugged repeatedly: no issue.

5. Hub plugged into a Thunderbolt port at runtime with no monitor attached: no issue.

6. Monitor then plugged into the already-connected hub: controller crashes immediately.

7. Ethernet adapter moved to a separate Thunderbolt port from the hub: still crashes (rules out bandwidth sharing on one port).

Conclusion: the trigger is creating (or tearing down and re-creating) a DisplayPort tunnel over USB4/Thunderbolt while Windows is running. Connecting a USB-C device alone is fine, and HDMI hotplug on the built-in port is fine.

 

OTHER OBSERVATIONS

- Setting BIOS to disable shared graphics left the internal screen dead and external monitors not driven. I had to revert to Optimus. That mode should work on this model, so this may be a second symptom of the same update.

- Rolling back the Intel display driver slightly improved behavior but did not resolve the crash.

- Event Viewer shows WHEA-Logger Event ID 17 (corrected PCIe error) on root port 0:6:1, PCI\VEN_8086&DEV_7ECA&SUBSYS_0CCC1028&REV_10, which appears to be the NVMe root port. It also shows ACPI Event ID 13 (embedded controller returned data when none was requested). I'm not sure either is related, but I'm including them for completeness.

 

CURRENT WORKAROUND

- Ethernet adapter on a USB-A port (chipset controller, unaffected)

- All displays connected before boot

- Monitor deep-sleep/auto-off disabled and Windows display timeout set to Never, to avoid hotplug events

 

QUESTIONS

1. Is this a known regression in the current BIOS or the Thunderbolt/USB4/retimer firmware bundled with it, or in the Windows USB4 connection manager?

2. Is a fix planned, or is there a known-good combination of BIOS, Intel graphics, chipset, and Thunderbolt drivers?

4. Is anyone else with an AA16250 (or another Core Ultra 200HX system) able to reproduce this by connecting a monitor through a Thunderbolt port at runtime?