Re: [PATCH 05/17] arm64: dts: imx93-11x11-evk: Add pinmux and mark conflicting devices for MQS1

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [Medium] Undocumented Functional Change: The patch silently modifies the &sai1 node by adding clocks and clock-names (including pll8k).
- [Medium] Incomplete Conflict Handling: The sound-micfil audio card is not labeled as a conflicting device, leading to infinite probe deferrals when &micfil is disabled.
--

--- Patch [5]: [PATCH 05/17] arm64: dts: imx93-11x11-evk: Add pinmux and mark conflicting devices for MQS1 ---
commit cd0bb14467afd18f66a4e88a88eba385a4784c70
Author: Chancel Liu <[email protected]>

arm64: dts: imx93-11x11-evk: Add pinmux and mark conflicting devices for MQS1

This commit adds the MQS1 pinctrl group and labels conflicting devices so
that an upcoming overlay can disable them. It also updates the clocks and
clock-names for the SAI1 interface.

> diff --git a/arch/arm64/boot/dts/freescale/imx93-11x11-evk.dts b/arch/arm64/boot/dts/freescale/imx93-11x11-evk.dts
> --- a/arch/arm64/boot/dts/freescale/imx93-11x11-evk.dts
> +++ b/arch/arm64/boot/dts/freescale/imx93-11x11-evk.dts
> @@ -105,7 +105,7 @@ cpu {
>  	};
>  };
>  
> -&micfil {
> +mqs1_conflict_dev: &micfil {

[Severity: Medium]
When the future overlay disables this device, what happens to the sound-micfil
audio card?

The sound-micfil node depends on the micfil device for its CPU DAI:

sound-micfil {
	...
	pri-dai-link {
		...
		cpu {
			sound-dai = <&micfil>;
		};
	};
};

If sound-micfil isn't also labeled as a conflict and disabled by the overlay
alongside this node, will the ALSA SoC core continuously fail to find the CPU
DAI, leading to endless probe deferrals and kernel log spam?

>  	pinctrl-names = "default", "sleep";
>  	pinctrl-0 = <&pinctrl_pdm>, <&pdm_fun>;

[ ... ]

> @@ -127,6 +132,10 @@ &sai1 {
>  	pinctrl-names = "default", "sleep";
>  	pinctrl-0 = <&pinctrl_sai1>;
>  	pinctrl-1 = <&pinctrl_sai1_sleep>;
> +	clocks = <&clk IMX93_CLK_SAI1_IPG>, <&clk IMX93_CLK_DUMMY>,
> +		 <&clk IMX93_CLK_SAI1_GATE>, <&clk IMX93_CLK_DUMMY>,
> +		 <&clk IMX93_CLK_DUMMY>, <&clk IMX93_CLK_AUDIO_PLL>;
> +	clock-names = "bus", "mclk0", "mclk1", "mclk2", "mclk3", "pll8k";

[Severity: Medium]
Could the commit message be updated to mention these changes to the SAI1 node?

The addition of the "pll8k" clock alters how the fsl_sai ALSA driver computes
audio rates at runtime, as the driver explicitly checks for this clock to
reparent clocks.

Since this changes the functional behavior of the existing SAI1 configuration,
should it be documented in the commit message or split into a separate patch?

>  	assigned-clocks = <&clk IMX93_CLK_SAI1>;
>  	assigned-clock-parents = <&clk IMX93_CLK_AUDIO_PLL>;

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=5