Re: [PATCH v4] platform/x86: asus-wmi: fix camera key LED on Zenbook S14

From: skyedryv

Date: Sun Nov 15 2026 - 06:56:09 EST


Hi Rong, Zane, José,

On Wed, 2026-10-21 at 20:34 +0800, Rong Zhang wrote:
> Still, could you confirm whether pressing the camera key during s2idle
> toggles the camera state without emitting a WMI event or not? If so, a
> resync should be performed on resume.

Data from another model: Zenbook S14 UX5406AA (Panther Lake), BIOS
UX5406AA.310. I applied v4 to the asus-wmi in CachyOS's 7.3.4 kernel
(cachyos-7.3.4-1) and loaded it in place of the stock module.

On this machine, pressing the camera key during s2idle does toggle the
camera, and the WMI event isn't lost, so v4's event-driven resync covers
it. I tried it twice:

- Stock asus-wmi, with a userspace logger on KEY_CAMERA: the press did
not wake the machine. On resume, KEY_CAMERA (WMI 0x85) arrived
within about a second of "PM: suspend exit", and the camera was
disabled.
- v4: the camera was enabled and the LED off before suspend. The press
during s2idle lit the LED straight away (it seems to have woken the
machine this time), and after resume asus::camera read 1, matching
the camera state. No userspace resync was involved: a CAMERA_LED_NEG
read a second after resume returned the "LED on" value, so the LED
was already correct by then.

With v4, the rest also behaves as described:

- asus::camera is registered (max_brightness 1). CAMERA_LED
(0x00060079) is absent here, as on the UX5406SA
- the LED matches the camera state at load
- on camera key presses, the LED, brightness and brightness_hw_changed
follow the camera (1 = disabled), and video goes black and comes
back
- KEY_CAMERA still reaches userspace (WMI code 0x85)

Cold boot: on this model the firmware re-enables the camera across a
full power-off. With the camera disabled (LED lit) before shutting
down, it came back enabled (UVC privacy 0, video working), with the
camera and mic-mute LEDs both dark, which is correct. So the "dark LED
with the camera disabled after cold boot" case doesn't occur here. Not
tested: anything on a model that has both device IDs.

Tested-by: Sky Dy <skyedryv@xxxxxxxxx>

I worked through this with help from an AI assistant (Claude); the
observations and logs are from my own laptop.

Thanks,
Sky Dy