Re: [PATCH v7 08/11] arm64: dts: mediatek: add ethernet support for mt8365-evk

[AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>]

Il 11/05/23 18:29, Alexandre Mergnat ha scritto:
> - Enable "vibr" and "vsim2" regulators to power the ethernet chip.
> 
> Tested-by: Kevin Hilman <khilman@baylibre.com>
> Signed-off-by: Alexandre Mergnat <amergnat@baylibre.com>
> ---
>   arch/arm64/boot/dts/mediatek/mt8365-evk.dts | 57 +++++++++++++++++++++++++++++
>   1 file changed, 57 insertions(+)
> 
> diff --git a/arch/arm64/boot/dts/mediatek/mt8365-evk.dts b/arch/arm64/boot/dts/mediatek/mt8365-evk.dts
> index 3a472f620ac0..cf81dace466a 100644
> --- a/arch/arm64/boot/dts/mediatek/mt8365-evk.dts
> +++ b/arch/arm64/boot/dts/mediatek/mt8365-evk.dts
> @@ -88,6 +88,28 @@ optee_reserved: optee@43200000 {
>   	};
>   };
>   
> +&ethernet {
> +	pinctrl-0 = <&ethernet_pins>;
> +	pinctrl-names = "default";
> +	phy-handle = <&eth_phy>;
> +	phy-mode = "rmii";
> +	/*
> +	 * Ethernet and HDMI (DSI0) are sharing pins.
> +	 * Only one can be enabled at a time and require the physical switch
> +	 * SW2101 to be set on LAN position
> +	 */
> +	status = "disabled";
> +
> +	mdio {
> +		#address-cells = <1>;
> +		#size-cells = <0>;
> +
> +		eth_phy: ethernet-phy@0 {
> +			reg = <0>;
> +		};
> +	};
> +};
> +
>   &i2c0 {
>   	clock-frequency = <100000>;
>   	pinctrl-0 = <&i2c0_pins>;
> @@ -137,12 +159,47 @@ &mt6357_pmic {
>   	#interrupt-cells = <2>;
>   };
>   
> +/* Needed by analog switch (multiplexer), HDMI and ethernet */

What part of the ethernet HW needs this regulator?

> +&mt6357_vibr_reg {
> +	regulator-always-on;
> +};
> +
>   /* Needed by MSDC IP */
>   &mt6357_vmc_reg {
>   	regulator-always-on;
>   };
>   
> +/* Needed by ethernet */

Same question for this one. If a device needs us to turn on a regulator in
order for it to be powered (read: if the supply is not fixed-on), setting
that supply as always-on is not beneficial for anyone, as eventually in a
power-off sleep/idle/whatever-pm state, this device (whole chip or IP) *will*
leak some amount of power.

If hardware engineers decided to connect a device to a supply that *can be*
shut down entirely there must be a reason, right? :-)

Regards,
Angelo