Re: [PATCH 6/7] arm64: dts: renesas: r8a779md: Add Renesas R-Car R8A779MD M3Le DTs

[Marek Vasut <marek.vasut@mailbox.org>]

On 4/29/26 2:16 PM, Geert Uytterhoeven wrote:

Hello Geert,

[...]

>> +/delete-node/ &pciec1;
>> +/delete-node/ &sata;
>> +/delete-node/ &sdhi3;
> 
> I am not sure disabling sdhi3 is correct.
> The R-Car M3Le documentation says:
> 
>     "M3Le does not support MMC1.
>      M3Le supports one MMC channel (MMC0) only.
>      SD3 is used only for SDHI and is not used for MMC.
>      M3Le does not support MMC HS400 class transfer rate."
> 
> The main R-Car Gen3 documentation says:
> 
>     "This LSI has four SD card interfaces (SDHI0 to SDHI3), two of
>      which can also be used as MMC interfaces (SDHI2 and SDHI3)."
> 
> So while MMC1 is indeed sdhi3, sdhi3 can still be used in SD mode?
> Do we need a new compatible value, as R-Car M3Le does not support HS400?

I think we do need new compatible string, yes, I'll include that in V2.

We can inhibit MMC support in sdhi3 with 'no-mmc' DT property, which I 
will also include in V2.

[...]

>> +/delete-node/ &vin7csi20;
>> +/delete-node/ &xhci0;
> 
> Shouldn't the following be disabled, too?
>    - ohci1, ehci1, and usb2_phy1,

Yes, they should.

>    - cmm1 and vspd1 (cfr. DU below).

Please see my response to Laurent in here in
[PATCH 6/7] arm64: dts: renesas: r8a779md: Add Renesas R-Car R8A779MD 
M3Le DTs
as well as my response to
[PATCH 1/7] dt-bindings: display: renesas,du: Document Renesas R-Car 
R8A779MD M3Le

The CMM1 and VSP1 are actually present, so I think the DT should 
describe them, even if they are not used. If only to allow the 
driver(s)to bind to those IPs and bring them to low power state.

>> +
>> +&du {
>> +       compatible = "renesas,du-r8a779md";
> 
> Assuming the DU needs a new compatible value...
> 
>> +       renesas,cmms = <&cmm0>, <&cmm3>;
>> +       renesas,vsps = <&vspd0 0>, <&vspd1 0>;
> 
> Shouldn't this be
> 
>      renesas,vsps = <&vspd0 0>, <&vspd0 1>;
> 
> ? DU1 is unused, but DU3 is used.
It should, and will be fixed in V2, thanks !

[...]