On 13/02/2025 18:37, matthew.gerlach@xxxxxxxxxxxxxxx wrote:
>
>
> On Wed, 12 Feb 2025, Krzysztof Kozlowski wrote:
>
>> On 11/02/2025 16:17, Matthew Gerlach wrote:
>>> Add required clock-frequency property to fixed-clock nodes
>>> to fix schema check warnings.
>>>
>>> Signed-off-by: Matthew Gerlach <matthew.gerlach@xxxxxxxxxxxxxxx>
>>> ---
>>> v6:
>>> - New patch to series.
>>> ---
>>> arch/arm64/boot/dts/intel/socfpga_agilex.dtsi | 4 ++++
>>> 1 file changed, 4 insertions(+)
>>>
>>> diff --git a/arch/arm64/boot/dts/intel/socfpga_agilex.dtsi b/arch/arm64/boot/dts/intel/socfpga_agilex.dtsi
>>> index 1235ba5a9865..42cb24cfa6da 100644
>>> --- a/arch/arm64/boot/dts/intel/socfpga_agilex.dtsi
>>> +++ b/arch/arm64/boot/dts/intel/socfpga_agilex.dtsi
>>> @@ -114,21 +114,25 @@ clocks {
>>> cb_intosc_hs_div2_clk: cb-intosc-hs-div2-clk {
>>> #clock-cells = <0>;
>>> compatible = "fixed-clock";
>>> + clock-frequency = <0>;
>>
>> That's not a correct frequency. You silence some error by introducing
>> incorrect properties. That's wrong.
>
> A clock-frequency of 0 seems valid for a clock that is disabled or not
> used on a particular board. I chose this approach because it already has
> widespread usage in the kernel:
>
> grep 'clock-frequency = <0>' arch/arm64/boot/dts/*/*.dtsi | wc -l
> 198
If the clock is not there, it should be removed or disabled. Otherwise
what is the point of setting here clock frequency=0? To silence some
warning. Why is there warning in the first place?
>
>>
>> Don't fix the warnings just to silence them, while keeping actual errors
>> still in the code.
>
> I actually want to fix the existing warnings, but it seems appropriate to
> only address the existing warnings that are related to this patch set of
> adding PCIe Root Port support to the Agilex family of chips. This patch
> set requires touching the file, socfpga_agilex.dtsi; so I fixed the
> warnings I thought were in this file. I believe the other warnings need to
> be fixed by converting text binding descriptions to yaml or by touching
> files unrelated to this patch set.
>
> Setting the value of the status property to "disabled" also silences the
> particular fixed-clock, but I didn't see any other usage by a fixed-clock.
> What do suggest is the best way to handle this warning?
DTSI and DTS represent actual hardware, so common DTSI should not have
clocks which do not exist. Such clocks should be in DTS or some other
common DTSI file, depending on how the hardware is organized.
Instead of fixing the warning, look at the cause - what is wrong in DTSI
in this hardware description.
Best regards,
Krzysztof
