Re: [PATCH v3 35/42] ARM: dts: ep93xx: add ts7250 board

[Krzysztof Kozlowski]

On 24/07/2023 15:41, Nikita Shubin wrote:
>>> diff --git a/arch/arm/boot/dts/cirrus/ep93xx-ts7250.dts
>>> b/arch/arm/boot/dts/cirrus/ep93xx-ts7250.dts
>>> new file mode 100644
>>> index 000000000000..625202f8cd25
>>> --- /dev/null
>>> +++ b/arch/arm/boot/dts/cirrus/ep93xx-ts7250.dts
>>> @@ -0,0 +1,145 @@
>>> +// SPDX-License-Identifier: GPL-2.0
>>> +/*
>>> + * Device Tree file for Technologic Systems ts7250 board based on
>>> Cirrus EP9302 SoC
>>> + */
>>> +/dts-v1/;
>>> +#include "ep93xx.dtsi"
>>> +#include <dt-bindings/dma/cirrus,ep93xx-dma.h>
>>> +
>>> +/ {
>>> +       compatible = "technologic,ts7250", "cirrus,ep9301";
>>> +       model = "TS-7250 SBC";
>>> +       #address-cells = <1>;
>>> +       #size-cells = <1>;
>>> +
>>> +       chosen {
>>> +       };
>>> +
>>> +       memory@0 {
>>> +               device_type = "memory";
>>> +               /* should be set from ATAGS */
>>> +               reg = <0x00000000 0x02000000>,
>>> +                     <0x000530c0 0x01fdd000>;
>>> +       };
>>> +
>>> +       nand-controller@60000000 {
>>
>> Where is this address? It does not work like that. If this is part of
>> SoC, then should be in DTSI and part of soc node. If not, then it is
>> some other bus which needs some description. Top-level is not a bus.
>>
> 
> It's some kind of EBI, but it doesn't need a driver it is transparent 

I did not mention any drivers. It's not really important here.


> on ts7250, the logic is controlled through installed CPLD.
> 
> The EBI it self is a part of the SoC through:

So should be in soc.

> 
> https://elixir.bootlin.com/linux/v6.5-rc3/source/arch/arm/mach-ep93xx/soc.h#L35
> 
> EP93XX_CS0_PHYS_BASE_ASYNC to EP93XX_CS0_PHYS_BASE_SYNC.
> 
> So for ts7250 this includes:
> 
> - NAND
> - m48t86
> - watchdog
> 
> I don't even know how to represent it correctly, would "simple-bus"
> with "ranges" defined suit here, so it will represent hierarchy but
> won't do anything ?

You said it is part of soc, so why shouldn't it be in the soc?

> 
>> You should see errors when testing dtbs with W=1.
> 
> Strangely - i don't see any, but anyway the above will change.



Best regards,
Krzysztof