Re: [PATCH v6 1/3] ASoC: dt-bindings: qcom,q6apm-lpass-dais: Document DAI subnode

[Mohammad Rafi Shaik]

On 7/21/2026 2:16 PM, Krzysztof Kozlowski wrote:
> On 21/07/2026 10:42, Mohammad Rafi Shaik wrote:
> > On 7/21/2026 1:50 PM, Krzysztof Kozlowski wrote:
> > > On 21/07/2026 10:14, Mohammad Rafi Shaik wrote:
> > > > On 7/21/2026 12:18 PM, Krzysztof Kozlowski wrote:
> > > > > On Mon, Jul 20, 2026 at 09:32:24PM +0530, Mohammad Rafi Shaik wrote:
> > > > > > +# Digital Audio Interfaces
> > > > > > +patternProperties:
> > > > > > +  '^dai@[0-9a-f]+$':
> > > > > > +    type: object
> > > > > > +    description:
> > > > > > +      Q6DSP Digital Audio Interfaces.
> > > > > > +
> > > > > > +    properties:
> > > > > > +      reg:
> > > > > > +        maxItems: 1
> > > > > > +        description:
> > > > > > +          Digital Audio Interface ID
> > > > > > +
> > > > > > +      clocks:
> > > > > > +        minItems: 1
> > > > > > +        items:
> > > > > > +          - description: MI2S bit clock
> > > > > > +          - description: MI2S external bit clock
> > > > > > +          - description: MI2S master clock
> > > > > > +
> > > > > > +      clock-names:
> > > > > > +        minItems: 1
> > > > > > +        items:
> > > > > > +          - enum: [bclk, eclk, mclk]
> > > > > > +          - const: mclk
> > > > >
> > > > > This does not match your clocks.
> > > >
> > > > will fix this.
> > > >
> > > > clocks descriptions are misleading because they imply an
> > > > ordered bclk + eclk configuration. I'll drop those descriptions.
> > > >
> > > > will keep only
> > > >
> > > > clocks:
> > > >       minItems: 1
> > > >       maxItems: 2
> > > >
> > > > > I gave you exact syntax to use, then in previous version you used
> > > > > something else. So I reminded. Again you used something else.
> > > > >
> > > > > If you decide not to use it, sure, your right but YOU CANNOT come with
> > > > > clearly invalid code as a substitute.
> > > > >
> > > > > NAK
> > > >
> > > > Suggested change:
> > > > https://lore.kernel.org/all/2b4448d0-81f8-410f-b849-0d957a2d2a17@xxxxxxxxxx/#t
> > > > minItems: 1
> > > > items:
> > > >     - enum: [ bclk, eclk ]
> > > >     - const: eclk
> > > >     - const: mclk
> > > >
> > > > I tested the suggested schema, but it results in DT binding check failures.
> > > >
> > > > DT binding check errors with the suggested change
> > > > 1. Only mclk enabled
> > > > bedais: dai@10:clock-names:0: 'mclk' is not one of ['bclk', 'eclk']
> > > >
> > > > 2. bclk and mclk enabled
> > > > bedais:dai@10:clock-names:1: 'eclk' was expected
> > > >
> > > >
> > > > With the current version of this patch, I do not see any DT binding
> > > > check issues. I tested all supported clock combinations, and
> > > > dt_binding_check passes successfully in each case.
> > >
> > > Your clocks told story that you have up to three clocks, so indeed if
> > > you do not have three clocks then my syntax is not accurate.
> > >
> > > Before we go further, where is upstream DTS using this?
> >
> > Currently, the Shikra DTS changes already present on the mailing list
> > include this patch as a dependency:
> > https://lore.kernel.org/all/20260616201315.2565115-7-mohammad.rafi.shaik@xxxxxxxxxxxxxxxx/
> >
> > +&q6apmbedai {
> > +	#address-cells = <1>;
> > +	#size-cells = <0>;
> > +	dai@40 {
> > +		reg = <SECONDARY_TDM_RX_0>;
> > +		clocks = <&q6prmcc LPASS_CLK_ID_AUD_INTF2_IBIT
> > +				 LPASS_CLK_ATTRIBUTE_COUPLE_NO>;
> > +		clock-names = "bclk";
> > +	};
> > +};
> >
> >
> > RubikPi3 DTS patch series uses both mclk and bclk:
> > https://lore.kernel.org/all/20260715-rubikpi-next-20260605-v2-6-67601d7a8581@xxxxxxxxxxxxxxx/
>
> Thanks, I did not see "eclk" used.

Yes, you're right. We have not used eclk in any upstream DTS so far 
since no such board using this.

include/dt-bindings/sound/qcom,q6dsp-lpass-ports.h#L156
#define LPASS_CLK_ID_PRI_MI2S_IBIT	1  (bclk)
#define LPASS_CLK_ID_PRI_MI2S_EBIT	2  (eclk)

My mistake—I understand it now. Both represent bit clocks; one is an 
internal bit clock (IBIT) and the other is an external bit clock (EBIT).

Since they are functionally the same clock type, we can use a single 
clock name (bclk) for both cases. At this point, only mclk and bclk are 
required, so there is no need to introduce a separate eclk clock name.

Sorry for the confusion earlier, i understood wrongly.

so can we keep like below syntex.

     clocks:
        minItems: 1
        maxItems: 2

      clock-names:
        minItems: 1
        items:
          - enum: [bclk, mclk]
          - const: mclk

This allows:
bclk
mclk
bclk, mclk

Thanks & regards,
Rafi.

> Best regards,
> Krzysztof