Re: [PATCH v2 1/5] dt-bindings: arm: google: Add dt bindings for frankel/blazer/mustang

Hi,

On Wed, Jul 22, 2026 at 2:56=E2=80=AFAM Peter Griffin <peter.griffin@linaro=
.org> wrote:
>
> -  # Bootloader requires empty ect node to be present
> +      # Google Tensor G5 AKA lga (laguna) SoC and boards
> +      - description:
> +          Google Pixel 10, 10 Pro, 10 Pro XL (Frankel, Blazer,
> +          Mustang).
> +        items:
> +          - enum:
> +              - google,lga-blazer
> +              - google,lga-frankel
> +              - google,lga-mustang
> +          - const: google,lga

I'm not sure I really want to make a big stink here since I'm
certainly interested in getting something landed. That being said, I
at least want to ask about the "google,lga" entry in the above
snippet. "google,lga" represents the SoC and I don't think that's
technically required in the top-level compatible string.

I had a previous email about this that essentially went nowhere [1],
but I'm hoping to continue the discussion (along with other similar
topics) and Plumbers. The essence of the question is: how much data do
we really want to jam into the top-level compatible string, and is the
top-level compatible string truly the best place to indicate which SoC
is present in the system? While many boards put the SoC here, I don't
think anything in the DT spec requires it, and I'm hoping that we can
stop doing it for new boards. When/if we need the SoC info in the
device tree, it seems like we could put it somewhere better...

[1] https://lore.kernel.org/r/CAD=3DFV=3DW+jE_L_LLgAhD8K_4+CtivSD9-9t7Xe63X=
[email protected]/