Re: [PATCH 1/2] ARM: Rockchip: Handle rk3288/rk3288w revision

[Mylene Josserand <mylene.josserand@collabora.com>]

Hi Geert,

On 3/26/20 4:31 PM, Geert Uytterhoeven wrote:
> Hi Mylene,
> 
> On Thu, Mar 26, 2020 at 2:50 PM Mylene Josserand
> <mylene.josserand@collabora.com> wrote:
>> On 3/6/20 11:45 AM, Geert Uytterhoeven wrote:
>>> On Wed, Mar 4, 2020 at 12:00 PM Heiko Stübner <heiko@sntech.de> wrote:
>>>> Am Montag, 2. März 2020, 16:57:02 CET schrieb Mylène Josserand:
>>>>> Determine which revision of rk3288 by checking the HDMI version.
>>>>> According to the Rockchip BSP kernel, on rk3288w, the HDMI
>>>>> revision equals 0x1A which is not the case for the rk3288 [1].
>>>>>
>>>>> As these SOC have some differences, the new function
>>>>> 'soc_is_rk3288w' will help us to know on which revision
>>>>> we are.
>>>>
>>>> what happened to just having a different compatible in the dts?
>>>> Aka doing a
>>>>
>>>> rk3288w.dtsi with
>>>>
>>>> #include "rk3288.dtsi"
>>>>
>>>> &cru {
>>>>           compatible = "rockchip,rk3288w-cru";
>>>> }
>>>>
>>>> I somehow don't expect boards to just switch between soc variants
>>>> on the fly.
>>>>
>>>> Also, doing things in mach-rockchip is not very future-proof:
>>>>
>>>> (1) having random soc-specific APIs spanning the kernel feels wrong,
>>>>       especially as at some point it might not be contained to our own special
>>>>       drivers like the cru. I cannot really see people being enthusiastic if
>>>>       something like this would be needed in say the core Analogix-DP bridge ;-)
>>>
>>> Indeed.  You're better of registering an soc_device_attribute using
>>> soc_device_register(), after which any driver can use soc_device_match()
>>> to differentiate based on the SoC revision.
>>
>> Thank you for this suggestion. The issue is that clocks are registered
>> at an early stage of the boot so using initcalls is too late for the
>> clock differentiation :(
> 
> IC, rk388 is still using CLK_OF_DECLARE().
> What about converting it to a platform driver, registered from e.g.
> subsys_initcall()?
> If you need some clocks early (e.g. for timers), you can do split
> registration, with the early part still using CLK_OF_DECLARE().
> That should work, assumed the timer clocks don't need differentiation.
> 
> Gr{oetje,eeting}s,
> 
>                          Geert
> 

oh, nice, thanks for the help!
I will try that and send a new version if it is working fine.

Best regards,
Mylène