ACPI: video: backlight non-functional on HP OMEN 16-am0015ns - _BCM/_BCL/_BQC under NVIDIA subtree, native path inert

From: Luis David García Waló

Date: Thu Nov 19 2026 - 16:14:34 EST


Hi,

I am reporting a backlight control failure on an HP OMEN 16-am0015ns
(Intel Arrow Lake-P iGPU + NVIDIA RTX 5060 Max-Q, hybrid/Optimus). I
filed it as Bugzilla #221648 in June; it has had no maintainer
response in five months, so I am bringing it to the list.

Hardware / firmware
-------------------
- HP OMEN Gaming Laptop 16-am0015ns, SKU BP4Q7EA#ABE
- BIOS F.14 (latest published by HP), DSDT OEM ID HPQOEM 8D2F, OEM
revision 01072009
- Kernel 7.3.4-300.fc45.x86_64 (Fedora 45)

Symptom
-------
Panel backlight cannot be adjusted. Brightness keys, GUI sliders and
direct sysfs writes all have no visible effect on the panel.

Firmware layout
---------------
The _BCM, _BCL and _BQC methods are located under
_SB.PCI0.PEGP.PEGP.DEDP, i.e. the NVIDIA subtree, rather than under
GFX0 (the Intel iGPU node).

_BCM does write to the Intel IGD opregion fields BCLP and ASLC, so the
firmware path is internally coherent. i915 never invokes it, because
it only searches for these methods under its own ACPI node.

The WMI GUID 603E9613-EF25-4338-A3D0-C46177516DB7 required by
nvidia_wmi_ec_backlight is also absent from this firmware, so that
driver cannot bind.

Full disassembled DSDT is attached to Bugzilla #221648.

How this differs from the known quirks
--------------------------------------
This looks close to two cases already handled in video_detect.c, and I
want to be explicit about where mine diverges, so it is not mistaken
for a duplicate:

- "ACPI: video: force native backlight on HP OMEN 16 (8A44)" (April
2026): same family, same reported symptom of writes updating the
software value without changing physical brightness. There, forcing
native was sufficient.

- The Acer Nitro AN515-58 / AN515-46 quirk (September 2026): firmware
advertising the NVIDIA WMI EC backlight GUID on a hybrid system with
the internal panel wired to the Intel iGPU, where
acpi_backlight=native let intel_backlight register and restored
control.

On this machine, forcing native is NOT sufficient.
/sys/class/backlight/intel_backlight is registered, accepts writes and
reports the new value back, but the panel does not change. So the
usual quirk would register an interface that is already present and
already inert.

Also tested and ruled out: all acpi_backlight= variants (vendor,
video, native, none), dpcd_backlight, and NVreg parameters.

Questions
---------
1. Given that the native interface registers but is inert, and _BCM is
reachable but under the wrong node, is there an existing mechanism to
point the ACPI video layer at _BCM methods located outside GFX0?

2. If not, what would be the preferred shape of a fix here? I am
willing to prepare and test a patch if someone can tell me which
direction is acceptable.

3. Is there any diagnostic you would want first - I can provide full
dmesg, acpidump, or instrumented runs on request. The hardware is
permanently available to me.

Thanks,
David