Re: [PATCH v6 2/2] arm64: dts: ti: k3-am62l3-evm: Add HDMI support

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
- [Medium] arm64: dts: ti: missing reset-gpios for sii9022 HDMI bridge
- [Medium] dts: missing gpio-hog for VOUT0_FET_SEL0 hardware mux
- [Medium] dts: ineffective bootph-all properties on HDMI endpoints and pinmux
--

commit 67599eb578ca3fc87e07bad393a95787a31afca5
Author: Devarsh Thakkar <devarsht@ti.com>

arm64: dts: ti: k3-am62l3-evm: Add HDMI support

This commit adds HDMI bridge and connector nodes to the TI AM62L3 EVM device
tree. It configures the DPI pinmux connecting VOUT0 to the SII9022 bridge, and
hooks up the DSS output to the bridge input.

> diff --git a/arch/arm64/boot/dts/ti/k3-am62l3-evm.dts b/arch/arm64/boot/dts/ti/k3-am62l3-evm.dts
> index 84a0cfaaf196a..7733a0d265ab4 100644
> --- a/arch/arm64/boot/dts/ti/k3-am62l3-evm.dts
> +++ b/arch/arm64/boot/dts/ti/k3-am62l3-evm.dts
> @@ -44,6 +44,19 @@ usr: button-usr {
>  		};
>  	};
>  
> +	hdmi0: connector-hdmi {
> +		compatible = "hdmi-connector";
> +		label = "hdmi";
> +		type = "a";
> +
> +		port {
> +			hdmi_connector_in: endpoint {
> +				remote-endpoint = <&sii9022_out>;
> +				bootph-all;
> +			};
> +		};
> +	};
> +

[Severity: Medium]
The hdmi_connector_in endpoint has a bootph-all property, but the parent
device node hdmi0 lacks it.

U-Boot's fdtgrep tool extracts subnodes structurally if a child has bootph-all,
but strips the properties (such as compatible and reg) of the parent node
unless the parent itself also has bootph-all.

Will this cause the connector driver to not probe in early boot, making the
bootph-all property ineffective and dead code?

[ ... ]

> @@ -192,6 +205,37 @@ exp2: gpio@23 {
>  		bootph-all;
>  	};
>  
> +	sii9022: bridge-hdmi@3b {
> +		compatible = "sil,sii9022";
> +		reg = <0x3b>;
> +		interrupt-parent = <&exp1>;
> +		interrupts = <20 IRQ_TYPE_EDGE_FALLING>;
> +		#sound-dai-cells = <0>;
> +		sil,i2s-data-lanes = < 0 >;

[Severity: Medium]
The sii9022 node is missing the reset-gpios property. Does this allow the
driver to probe silently without a reset line?

The AM62L3 EVM hardware has a dedicated reset line (GPIO_HDMI_RSTn) on the
exp2 expander:

	exp2: gpio@23 {
		...
		gpio-line-names = "BT_EN_SOC", "VOUT0_FET_SEL0",
		...
				  "GPIO_QSPI_NAND_RSTn", "GPIO_HDMI_RSTn",

By omitting reset-gpios, the kernel fails to actively manage the reset state,
relying on the bootloader to leave the pin de-asserted.

If the bootloader leaves it floating or asserted, or if the hardware loses
state during suspend, will the HDMI bridge silently fail to operate?

[Severity: Medium]
Additionally, the exp2 GPIO expander defines a VOUT0_FET_SEL0 line, indicating
a hardware FET switch multiplexes the DPI/VOUT0 signals.

Without a gpio-hog to actively select the HDMI output path, does the expander
default to inputs (high impedance), leaving the FET switch floating or resting
in an unverified default hardware state?

Could this cause display signals to not reach the HDMI bridge, rendering the
HDMI output non-functional?

> +
> +		ports {
> +			#address-cells = <1>;
> +			#size-cells = <0>;
> +
> +			port@0 {
> +				reg = <0>;
> +
> +				sii9022_in: endpoint {
> +					remote-endpoint = <&dpi_out>;
> +					bootph-all;
> +				};
> +			};
> +
> +			port@1 {
> +				reg = <1>;
> +
> +				sii9022_out: endpoint {
> +					remote-endpoint = <&hdmi_connector_in>;
> +					bootph-all;
> +				};
> +			};
> +		};
> +	};

[Severity: Medium]
Similar to the hdmi0 node, the sii9022_in and sii9022_out endpoints have
bootph-all but the parent sii9022 node lacks it.

Will the missing bootph-all on the parent prevent the driver from probing in
early boot, rendering the endpoints' bootph-all properties dead code?

[ ... ]

> @@ -321,6 +400,24 @@ AM62LX_IOPAD(0x0004, PIN_OUTPUT, 0) /* (AA23) WKUP_UART0_TXD */
>  	};
>  };
>  
> +&dss {
> +	status = "okay";
> +	pinctrl-names = "default";
> +	pinctrl-0 = <&dpi_pins_default>;
> +};
> +
> +&dss_ports {
> +	/* VP1: DPI Output */
> +	port@0 {
> +		reg = <0>;
> +
> +		dpi_out: endpoint {
> +			remote-endpoint = <&sii9022_in>;
> +			bootph-all;
> +		};
> +	};
> +};

[Severity: Medium]
The dpi_out endpoint has bootph-all, but the parent dss and dss_ports nodes
lack it.

Will the missing bootph-all on the parents prevent the display subsystem from
probing in early boot, making the endpoint bootph-all properties ineffective?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261007100707.1065564-1-s-jain1@ti.com?part=2