Re: Re: [PATCH v2 1/3] dt-bindings: ata: eswin: Document for EIC7700 SoC ahci

From: Niklas Cassel <cassel@kernel.org>
To: luyulin@eswincomputing.com
Cc: Rob Herring <robh@kernel.org>,
	dlemoal@kernel.org, krzk+dt@kernel.org, conor+dt@kernel.org,
	linux-ide@vger.kernel.org, devicetree@vger.kernel.org,
	linux-kernel@vger.kernel.org, vkoul@kernel.org,
	kishon@kernel.org, linux-phy@lists.infradead.org,
	ningyu@eswincomputing.com, zhengyu@eswincomputing.com,
	linmin@eswincomputing.com, huangyifeng@eswincomputing.com,
	fenglin@eswincomputing.com, lianghujun@eswincomputing.com
Subject: Re: Re: [PATCH v2 1/3] dt-bindings: ata: eswin: Document for EIC7700 SoC ahci
Date: Thu, 28 Aug 2025 15:05:15 +0200	[thread overview]
Message-ID: <aLBUC116MdJqDGIJ@flawful.org> (raw)
In-Reply-To: <692e11ca.843.198f0337528.Coremail.luyulin@eswincomputing.com>

On Thu, Aug 28, 2025 at 06:22:40PM +0800, luyulin@eswincomputing.com wrote:
> 
> Do you mean that ports-implemented should be removed from the dts,
> and the corresponding register should be configured by the firmware
> (which is U-Boot on the HiFive Premier P550 board)? Is this understanding correct?
> If so, when the driver is removed, a reset will be triggered,
> causing the configuration of this register to be lost,
> which will result in an error when insmod the driver again.

My 50 cents,

if the ports implemented register gets reset from the reset_control_reset()
in ahci_platform_assert_rsts(), then it seems like having ports-implemented
in device tree is acceptable.

There are a bunch of device trees that have this already:
arch/arm/boot/dts/qcom/qcom-apq8064.dtsi:                       ports-implemented = <0x1>;
arch/arm/boot/dts/qcom/qcom-ipq8064-v1.0.dtsi:                  ports-implemented = <0x1>;
arch/arm/boot/dts/qcom/qcom-ipq8064-v2.0.dtsi:  ports-implemented = <0x1>;
arch/arm/boot/dts/samsung/exynos5250.dtsi:                      ports-implemented = <0x1>;
arch/arm/boot/dts/socionext/uniphier-pro4.dtsi:                 ports-implemented = <1>;
arch/arm/boot/dts/socionext/uniphier-pro4.dtsi:                 ports-implemented = <1>;
arch/arm/boot/dts/socionext/uniphier-pxs2.dtsi:                 ports-implemented = <1>;
arch/arm/boot/dts/st/stih407-family.dtsi:                       ports-implemented = <0x1>;
arch/arm/boot/dts/st/stih407-family.dtsi:                       ports-implemented = <0x1>;
arch/arm/boot/dts/ti/omap/dra7-l4.dtsi:                         ports-implemented = <0x1>;
arch/arm/boot/dts/ti/omap/omap5-l4.dtsi:                                ports-implemented = <0x1>;
arch/arm64/boot/dts/mediatek/mt7622.dtsi:               ports-implemented = <0x1>;
arch/arm64/boot/dts/rockchip/rk3568.dtsi:               ports-implemented = <0x1>;
arch/arm64/boot/dts/rockchip/rk356x-base.dtsi:          ports-implemented = <0x1>;
arch/arm64/boot/dts/rockchip/rk356x-base.dtsi:          ports-implemented = <0x1>;
arch/arm64/boot/dts/rockchip/rk3576.dtsi:                       ports-implemented = <0x1>;
arch/arm64/boot/dts/rockchip/rk3576.dtsi:                       ports-implemented = <0x1>;
arch/arm64/boot/dts/rockchip/rk3588-base.dtsi:          ports-implemented = <0x1>;
arch/arm64/boot/dts/rockchip/rk3588-base.dtsi:          ports-implemented = <0x1>;
arch/arm64/boot/dts/rockchip/rk3588-extra.dtsi:         ports-implemented = <0x1>;
arch/arm64/boot/dts/socionext/uniphier-pxs3.dtsi:                       ports-implemented = <1>;
arch/arm64/boot/dts/socionext/uniphier-pxs3.dtsi:                       ports-implemented = <1>;

Sure, if the ports implemented register was sticky (kept its value after a
reset), then I think Rob's suggestion would make sense.

Kind regards,
Niklas