Re: [PATCH v3 0/5] platform/x86: asus-wmi: TUF keyboard shutdown naming and RGB readback

[Ilpo Järvinen]

On Sat, 10 Oct 2026, Bartu Alev wrote:

> This series fixes the name of the fourth TUF keyboard power state and
> makes the TUF keyboard RGB attributes readable.
> 
> The fourth flag of kbd_rgb_state was labelled "keyboard" and treated as
> a no-op since it was introduced.  In hardware it gates the backlight
> during the ACPI S5 power-off sequence: with the bit set the keyboard
> lights up white while the machine powers off, with the bit clear the
> machine powers off dark.  Patch 2 renames the field and carries a Fixes
> tag for stable.
> 
> The remaining patches prepare and add readback.  DSTS device 0x0010005B
> evaluates the DSDT method EC0.KBLS(), which returns a 16-byte snapshot
> of the EC's keyboard state: lighting mode, red, green and blue levels,
> animation speed, and the four power flags.  Both kbd_rgb_mode and
> kbd_rgb_state become readable and report the write format with the
> command word fixed at 1.
> 
> Responses to the v2 review:
> 
> Ilpo:
> - Documentation: patch 5 adds the missing ABI entries.
> - Errno: the helper already separates the cases.
>   asus_wmi_evaluate_method_buf() returns -EIO for a failed evaluation
>   and -ENODEV for an unsupported DEVID.  The readback keeps the
>   err < 0 ? err : -ENODEV normalization for the positive ACPI integers
>   that leak through, and the AML signature check returns -ENODEV as
>   well.
> - Named defines: patch 3 names the three speed codes and the four state
>   bits, and the decoding comments are gone.
> - Struct: patch 4 parses the status buffer through a packed struct.
>
> Denis:
> - DEVID naming: the write DEVID is now ASUS_WMI_DEVID_TUF_RGB_CMD
>   (patch 1) and the read DEVID is ASUS_WMI_DEVID_TUF_RGB_READ_STATUS,
>   as you suggested.
> - -ENODATA: asus_wmi_evaluate_method_buf() already returns -ENODATA for
>   empty buffers, so no extra handling was added.
> 
> Provenance: the KBLS buffer layout and the speed-code semantics were
> recovered by static analysis of the EC firmware (ITE IT5570VG),
> assisted by an LLM-driven toolchain; the readback patch carries an
> Assisted-by tag.  Tested on a TUF Gaming A15 (FA507NV): the v2
> readback on kernel 7.2.6, and the S5 behaviour of the shutdown flag by
> A/B testing both values.
> 
> Changes in v3, relative to v2:
> - split into five patches, one logical change each
> - renamed the write DEVID to ASUS_WMI_DEVID_TUF_RGB_CMD and the new one
>   to ASUS_WMI_DEVID_TUF_RGB_READ_STATUS, per Denis Benato
> - named the speed codes and the power state bits and dropped the
>   decoding comments
> - parse the status buffer through a packed struct
> - added the ABI documentation, with the attributes documented at their
>   real location under /sys/class/leds/asus::kbd_backlight/ and worded
>   for all laptops that expose the zonal RGB devices, not only TUF
>   models
> - added linux-api to Cc, which the earlier versions missed
> 
> v2: https://lore.kernel.org/all/20260926005625.171560-1-bartualev@xxxxxxxxx/
> v1: https://lore.kernel.org/all/20260925200744.129714-1-bartualev@xxxxxxxxx/
> 
> Bartu Alev (5):
>   platform/x86: asus-wmi: rename TUF RGB state DEVID to _CMD
>   platform/x86: asus-wmi: fix TUF keyboard power state shutdown naming
>   platform/x86: asus-wmi: name the TUF keyboard RGB state and speed
>     codes

Thanks.

I took the first 3 patches into the review-ilpo-next branch now as there 
is another series that conflicts/overlaps with them.

>   platform/x86: asus-wmi: add TUF keyboard RGB readback support

v2 thread has my thoughts about this one.

>   Documentation/ABI: asus-wmi: document TUF keyboard RGB attributes
> 
>  .../ABI/testing/sysfs-platform-asus-wmi       |  42 ++++++
>  drivers/platform/x86/asus-wmi.c               | 123 +++++++++++++++---
>  include/linux/platform_data/x86/asus-wmi.h    |   5 +-
>  3 files changed, 153 insertions(+), 17 deletions(-)

-- 
 i.