On Mon, Sep 25, 2023 at 01:35:42PM +0200, Alain Volmat wrote:
> On Fri, Sep 22, 2023 at 07:08:18PM +0300, Laurent Pinchart wrote:
> > On Fri, Sep 22, 2023 at 06:02:27PM +0200, Alain Volmat wrote:
> > > On Tue, Sep 05, 2023 at 12:02:58PM +0300, Laurent Pinchart wrote:
> > > > On Fri, Sep 01, 2023 at 05:57:23PM +0200, Alain Volmat wrote:
> > > > > From: Hugues Fruchet <hugues.fruchet@xxxxxxxxxxx>
> > > > >
> > > > > Add dcmipp support to STM32MP135.
> > > > >
> > > > > Signed-off-by: Hugues Fruchet <hugues.fruchet@xxxxxxxxxxx>
> > > > > Signed-off-by: Alain Volmat <alain.volmat@xxxxxxxxxxx>
> > > > > ---
> > > > > arch/arm/boot/dts/st/stm32mp135.dtsi | 8 ++++++++
> > > > > 1 file changed, 8 insertions(+)
> > > > >
> > > > > diff --git a/arch/arm/boot/dts/st/stm32mp135.dtsi b/arch/arm/boot/dts/st/stm32mp135.dtsi
> > > > > index abf2acd37b4e..beee9ec7ed0d 100644
> > > > > --- a/arch/arm/boot/dts/st/stm32mp135.dtsi
> > > > > +++ b/arch/arm/boot/dts/st/stm32mp135.dtsi
> > > > > @@ -8,5 +8,13 @@
> > > > >
> > > > > / {
> > > > > soc {
> > > > > + dcmipp: dcmipp@5a000000 {
> > > > > + compatible = "st,stm32mp13-dcmipp";
> > > > > + reg = <0x5a000000 0x400>;
> > > > > + interrupts = <GIC_SPI 79 IRQ_TYPE_LEVEL_HIGH>;
> > > > > + resets = <&rcc DCMIPP_R>;
> > > > > + clocks = <&rcc DCMIPP_K>;
> > > > > + status = "disabled";
> > > >
> > > > This needs a port, as it's marked as required in the bindings. You can
> > > > leave the endpoint out.
> > >
> > > I first agreed with your comment but, having done the check (make
> > > CHECK_DTBS=y ...) this doesn't seem to be required because the dcmipp
> > > node is kept disabled within our dtsi.
> >
> > Interesting.
> >
> > > (it is later on only enabled in dts file which as well have the port
> > > property).
> > > Indeed, to check this I changed it to okay and DTC_CHK complained about
> > > missing port property.
> > >
> > > Hence, I'd think that port doesn't have to be added in this dtsi file.
> > > Would you agree with that ?
> >
> > I still think the port belongs here, as it's an intrinsic property of
> > the dcmipp, not a property of the board. Does it cause any issue to add
> > a port in the .dtsi ?
>
> I agree that the port refers more to the SoC (hence dtsi) rather than
> the board (hence dts), however I am wondering if this is really
> something usually done. I had a look at other dtsi with node related
> to similar kind of devices and it seems to me that there is no such case
> of a dtsi with a port having nothing in it. Did I missed something ?
Look at the csi@32e4000 and csi@32e5000 nodes in
arch/arm64/boot/dts/freescale/imx8mp.dtsi for instance. There are quite
a few other examples.
> > > > With this fixed,
> > > >
> > > > Reviewed-by: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx>
> > > >
> > > > > + };
> > > > > };
> > > > > };
--
Regards,
Laurent Pinchart
