Re: [PATCH V4 6/7] arm64: dts: imx8dxl/qm/qxp: Move power supply properties to Root Port node

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] Moving `vpcie3v3aux-supply` to the Root Port node delegates its lifecycle to the generic `pci-pwrctrl` framework, causing auxiliary power to be disabled during system suspend, breaking Wake-on-PCIe.
- [High] Removal of `vpcie-supply` and `vpcie3v3aux-supply` from the host node breaks Devicetree forward compatibility for older kernels.
--

commit 5e5016406a44aded0d05f19605180a501f38ea48
Author: Sherry Sun <sherry.sun@nxp.com>

arm64: dts: imx8dxl/qm/qxp: Move power supply properties to Root Port node
    
This commit moves the power supply properties from the PCIe controller nodes
to the Root Port child nodes to support the new PCI pwrctrl framework.

> diff --git a/arch/arm64/boot/dts/freescale/imx8dxl-evk.dts b/arch/arm64/boot/dts/freescale/imx8dxl-evk.dts
> index 78e8d41e67916..59d9fe687aaf0 100644
> --- a/arch/arm64/boot/dts/freescale/imx8dxl-evk.dts
> +++ b/arch/arm64/boot/dts/freescale/imx8dxl-evk.dts
> @@ -677,8 +677,6 @@ &pcie0 {
>  	pinctrl-names = "default";
>  	/* This property is deprecated, use reset-gpios from the Root Port node. */
>  	reset-gpio = <&lsio_gpio4 0 GPIO_ACTIVE_LOW>;
> -	vpcie-supply = <&reg_pcieb>;
> -	vpcie3v3aux-supply = <&reg_pcieb>;

[Severity: High]
Will removing these properties break Devicetree forward compatibility for
older kernels?

Older kernels lacking the new pci-pwrctrl framework rely on finding the
vpcie-supply property in the host controller node to power the PCIe physical
slots. Because these regulators are not marked regulator-always-on, older
kernels might fail to enable them, resulting in unpowered slots and complete
device enumeration failure.

Could these properties be kept in the host node as deprecated fallbacks,
similar to how reset-gpio is handled here? The kernel regulator framework
uses reference counting, which should safely handle multiple driver
consumers.

>  	status = "okay";
>  };
>  
> @@ -692,6 +690,8 @@ &pcie0_ep {
>  
>  &pcieb_port0 {
>  	reset-gpios = <&lsio_gpio4 0 GPIO_ACTIVE_LOW>;
> +	vpcie3v3-supply = <&reg_pcieb>;
> +	vpcie3v3aux-supply = <&reg_pcieb>;

[Severity: High]
Does moving vpcie3v3aux-supply to the Root Port node break Wake-on-PCIe
during system suspend?

By delegating the lifecycle to the generic pci-pwrctrl framework, the
regulator might be turned off during suspend. Looking at the pci-imx6.c
driver, during system suspend dw_pcie_suspend_noirq() delegates to the host
controller's deinit callback imx_pcie_host_exit().

This unconditionally calls pci_pwrctrl_power_off_devices(pci->dev) without
checking pci->pp.skip_pwrctrl_off, unlike other DWC drivers like pcie-qcom.

If the pci-pwrctrl framework takes ownership and turns the regulator off, it
would violate the PCIe hardware requirement to maintain Vaux for Wake-on-PCIe
(PME from D3cold). Can we ensure the auxiliary supply remains active during
suspend events?

[ ... ]

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260630060710.3294811-1-sherry.sun@oss.nxp.com?part=6