Re: [RFC PATCH] ARM: s3c: mark as deprecated and schedule removal after 2022

[Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>]

On 06/11/2021 21:13, Arnd Bergmann wrote:
> On Sat, Nov 6, 2021 at 8:33 PM Heiko Stübner <heiko@sntech.de> wrote:
>> Am Samstag, 6. November 2021, 19:20:05 CET schrieb Krzysztof Kozlowski:
>>> On 05/11/2021 18:49, Cedric Roux wrote:
>>>>
>>>> Anyway, there is at least one s3c2440 server on Earth powered
>>>> by the linux kernel. Now you know! :)
>>>
>>> Thanks for coming back! We will see how much of feedback we gather.
>>
>> I do believe all (or at least most) of s3c24xx could run with a devicetree
>> base with core peripherals enabled.
>>
>> So one possible alternative way could be to just deprecate (and then drop)
>> all the board-files + their platform-data parts in mach-s3c.
> 
> Yes, and this would be similar to what we've done for mach-imx and
> mach-omap2 in the past.
> 
> However, I think the Cragganmore/Speyside board that Mark mentioned
> is not in the category of easily converted machines, so leaving only the
> DT-enabled machines around would not help him.
> 
> If this is actually the case, one other possibility would be to remove
> all the board files except for that one, in the hope that the transformation
> (even a partial one) to DT becomes easier when there is no risk of
> breaking other machines.

Dropping board files would be a nice step but only a half-way, because
we could not remove the S3C support from the drivers.

The platform to stay should have either:
1. real users in the production who need updates
or
2. hackers who have hardware, like to play with it and will do the
testing/development.

In the second case (the hackers) having only the hardware but not
testing or running newest kernels does not revive the platform from the
almost-dead stage.


Best regards,
Krzysztof