Re: [PATCH v3 1/6] dt-binding: ARM: samsung: Add Samsung Exynos8855

On 06/07/2026 05:44, Alim Akhtar wrote:
> Hi Ivaylo,
> Thanks for your review. 
> 
>> -----Original Message-----
>> From: Ivaylo Ivanov <[email protected]>
>> Sent: Monday, July 6, 2026 4:11 AM
>> To: Alim Akhtar <[email protected]>; [email protected];
>> [email protected]; [email protected]; [email protected];
>> [email protected]
>> Cc: [email protected]; [email protected];
>> [email protected]; [email protected];
>> [email protected]
>> Subject: Re: [PATCH v3 1/6] dt-binding: ARM: samsung: Add Samsung
>> Exynos8855
>>
>> On 6/27/26 20:12, Alim Akhtar wrote:
>>> Add Samsung Exynos8855 smdk board to documentation
>>>
>>> Signed-off-by: Alim Akhtar <[email protected]>
>>> ---
>>>  .../devicetree/bindings/arm/samsung/samsung-boards.yaml     | 6 ++++++
>>>  1 file changed, 6 insertions(+)
>>>
>>> diff --git
>>> a/Documentation/devicetree/bindings/arm/samsung/samsung-boards.yaml
>>> b/Documentation/devicetree/bindings/arm/samsung/samsung-boards.yaml
>>> index 753b3ba1b607..273464400477 100644
>>> ---
>>> a/Documentation/devicetree/bindings/arm/samsung/samsung-boards.yaml
>>> +++ b/Documentation/devicetree/bindings/arm/samsung/samsung-
>> boards.yam
>>> +++ l
>>> @@ -235,6 +235,12 @@ properties:
>>>                - winlink,e850-96                 # WinLink E850-96
>>>            - const: samsung,exynos850
>>>
>>> +      - description: Exynos8855 based boards
>>> +        items:
>>> +          - enum:
>>> +              - samsung,exynos8855-smdk         # Samsung SMDK
>>> +          - const: samsung,exynos8855
>>
>> Is there any particular reason for using "exynos8855" rather than the
>> commercial name - exynos1580? We've already established using the latter
>> naming scheme as a convention (e.g. exynos3830 -> exynos850, exynos9830 ->
>> exynos990) rather than the development model numbers, so I don't think
>> breaking that will help anyone with the already painful model number confusion.
>>
> The reason for using Exynos8855 is that it comes from the chip-id register, which is RO register.
> And in my opinion it has to be chip-id, maintaining two names is confusing (as you also mentioned).
> Yes, there were couple of deviation but let's come back to the "original" convention (which is to use chip-id).
> 
> @ Krzysztof, Peter 
> Any input on this? 

Back in Exynos850 upstreaming, my preference was the chipid value, but I
agreed for marketing name. Marketing names are tricky, because:
1. They change,
2. They might multiply (two or three marketing names for the same die -
common case in Qualcomm),

That's why for Samsung SoCs, I rather prefer chipid value. For
Google/Axis/Tesla the case is different and we should use their naming -
some engineering name for example.



Best regards,
Krzysztof