Re: [PATCH v3 1/3] dt-bindings: clock: qcom: Move x1e80100 TCSR to own binding

[Qiang Yu <qiang.yu@oss.qualcomm.com>]

On Thu, Aug 13, 2026 at 09:38:36AM +0200, Krzysztof Kozlowski wrote:
> On 13/08/2026 08:55, Qiang Yu wrote:
> >>> +required:
> >>> +  - compatible
> >>> +  - clocks
> >>> +  - '#clock-cells'
> >>> +  - '#reset-cells'
> >>
> >> This is unexplained ABI change. When you are asked, you need to address
> >> feedback from LLM, not blindly implement it without checking its
> >> validity.
> >>
> > After adding these two properties to the required list, I ran a dtb
> > check and saw no related errors or warnings. But I overlooked one
> > thing: my dtb check was against the current tree only. This change
> > could make dtb checks fail for historical TCSR DT nodes that did not
> > have these two properties. So to be safe, I don't think they should be
> > added to the required list.
> 
> Did you change the driver code to fail when DTS is not having these?

No, the driver logic is unchanged.

For #clock-cells: I did not add any failure path, but a missing
#clock-cells already makes consumers fail to get their clocks --
of_parse_phandle_with_args() returns -EINVAL when the provider lacks
the cells property, so of_clk_get() fails. This is enforced by the OF
core, not by this driver.

For #reset-cells: the TCSR does not provide any reset and no consumer
gets a reset from it, so there is nothing that would fail.

- Qiang Yu