[PATCH] ARM: dts: mx5: Pass the memory unit-address

Shawn Guo shawnguo at kernel.org
Wed Feb 28 16:41:41 PST 2018


On Wed, Feb 28, 2018 at 09:00:50AM -0300, Fabio Estevam wrote:
> Hi Shawn,
> 
> On Wed, Feb 28, 2018 at 12:03 AM, Shawn Guo <shawnguo at kernel.org> wrote:
> 
> > One thing I'm unsure is which memory node will be fixed up with correct
> > memory configuration, considering we will have two memory nodes.  One is
> > the following that defined in imx51.dtsi.
> >
> >  memory { device_type = "memory"; };
> >
> > And the other is memory at 90000000.
> 
> On my tests the correct memory@ is filled with the real memory size.
> 
> One further cleanup I can make to avoid this double memory node is:
> 
> --- a/arch/arm/boot/dts/imx6qdl-sabresd.dtsi
> +++ b/arch/arm/boot/dts/imx6qdl-sabresd.dtsi
> @@ -20,6 +20,7 @@
>         };
> 
>         memory at 10000000 {
> +               device_type = "memory";
>                 reg = <0x10000000 0x40000000>;
>         };
> 
> diff --git a/arch/arm/boot/dts/imx6qdl.dtsi b/arch/arm/boot/dts/imx6qdl.dtsi
> index c003e62..f3376f0 100644
> --- a/arch/arm/boot/dts/imx6qdl.dtsi
> +++ b/arch/arm/boot/dts/imx6qdl.dtsi
> @@ -23,7 +23,6 @@
>          * Also for U-Boot there must be a pre-existing /memory node.
>          */
>         chosen {};
> -       memory { device_type = "memory"; };
> 
>         aliases {
>                 ethernet0 = &fec;
> 
> If you agree I can do this for all i.MX dts.
> 
> What do you think?

I agree.  In the end, we will need to have every single board DTS
contain a memory at unit-address node.  Although the size field can be
dummy and fixed up by bootloader.

Shawn



More information about the linux-arm-kernel mailing list