Re: [PATCH] arm64: dts: rockchip: Fix vcc_sdio regulator max voltage on Pinebook Pro

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] The patch leaves `vcc3v0_sd` incorrectly defined as 3.0V, causing Linux to misconfigure `vcc_sdio` and expose the RK3399 I/O pins to overvoltage from the SD card.
- [High] `vcc_sdio` remains powered during suspend while the SD card VDD is removed, causing potentially damaging back-powering.
--

commit 2b1796d52c7328502f3af71aefe96d0d149cf297
Author: Hugo Osvaldo Barrera <hugo@whynothugo.nl>

arm64: dts: rockchip: Fix vcc_sdio regulator max voltage on Pinebook Pro

This commit updates the vcc_sdio regulator to support a maximum of 3.4V,
matching the hardware capability. This fixes downstream u-boot issues where
it falls back to 1.8V, preventing SD card boot.

> diff --git a/arch/arm64/boot/dts/rockchip/rk3399-pinebook-pro.dts b/arch/arm64/boot/dts/rockchip/rk3399-pinebook-pro.dts
> index b5c05928142c0..f21fcdc1b49a5 100644
> --- a/arch/arm64/boot/dts/rockchip/rk3399-pinebook-pro.dts
> +++ b/arch/arm64/boot/dts/rockchip/rk3399-pinebook-pro.dts
> @@ -526,7 +526,7 @@ vcc_sdio: LDO_REG4 {
>  				regulator-always-on;
>  				regulator-boot-on;
>  				regulator-min-microvolt = <1800000>;
> -				regulator-max-microvolt = <3000000>;
> +				regulator-max-microvolt = <3400000>;

This isn't a bug introduced by this patch, but should the vcc3v0_sd regulator
also be updated to 3.3V?

The vcc3v0_sd regulator is defined as 3.0V in the DT, but it is a fixed load
switch powered by vcc3v3_sys (3.3V), so the SD card physically receives 3.3V.

The MMC core relies on vcc3v0_sd to determine the maximum SD card voltage,
limiting the I/O voltage (vqmmc, mapped to vcc_sdio) to 3.0V.

With the SD card operating at 3.3V and the RK3399 I/O pins powered at 3.0V,
will the SD card drive 3.3V data lines into the 3.0V I/O pins? This exceeds
the RK3399 absolute maximum rating for I/O pins (VCCIO + 0.3V), which could
risk hardware degradation and latch-up.

>  
>  				regulator-state-mem {
>  					regulator-on-in-suspend;

This isn't a bug introduced by this patch, but does keeping vcc_sdio on during
suspend cause back-powering issues?

The vcc_sdio regulator is configured with regulator-on-in-suspend and
regulator-suspend-microvolt = <3000000>, keeping the RK3399 sdmmc I/O domain
powered at 3.0V during suspend.

However, the SD card's main VDD supply (vcc3v0_sd) has regulator-off-in-suspend,
dropping to 0V.

With the I/O lines pulled up to 3.0V by the RK3399 and the SD card unpowered,
can current leak through the SD card's internal ESD protection diodes, draining
the battery during suspend and potentially causing latch-up or damage?

Since vcc_sdio only powers the sdmmc-supply IO domain on this board, should
its state memory be changed to regulator-off-in-suspend to match vcc3v0_sd?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260519094439.7918-1-hugo@whynothugo.nl?part=1