[RFC PATCH 2/2] lib: reset: thead: Correct the naming convention of dts

[Jessica Clarke]

On 12 Jun 2023, at 01:57, Guo Ren <guoren at kernel.org> wrote:
> 
> Hi Conor,
> 
> Jisheng Zhang would update the Linux yaml patch, here is the final dts
> format of the reset controller:
> 
>               reset-controller at ffff019050 {
>                       compatible = "thead,cpu-reset";
>                       reg = <0xff 0xff019050 0x0 0x4>, <0xff
> 0xff015004 0x0 0x0>;
>                       reset-ctrl = <0x1c>;
>                       clone-csrs = <0x7f3 0x7c0 0x7c1 0x7c2 0x7c3

This is pretty horrible, there’s no indirect CSR read/write
instruction, so what are you expecting drivers to do with this? Have a
big switch statement for every possible CSR?

Jess

> 0x7c5 0x7cc 0x7ce>;
>                };
> 
> The reset-ctrl is used to control different parts of soc, generally, a
> bit indicates a reset signal (a core/a interconnect/a subsystem).
> 
> On Thu, May 25, 2023 at 2:06 PM Guo Ren <guoren at kernel.org> wrote:
>> 
>> On Thu, May 25, 2023 at 1:33 PM Conor Dooley <conor at kernel.org> wrote:
>>> 
>>> 
>>> 
>>> On 25 May 2023 04:15:36 IST, Guo Ren <guoren at kernel.org> wrote:
>>> 
>>>> So it could be "thead,cpu-reset", okay?
>>> 
>>> As a generic fallback compatible.
>>> 
>>>> Actually, our core could let SoC vendors define their own custom
>>>> CSRs/custom reset values of CSRs, so we don't know what would be added
>>>> in the future. Put a array in dts instead of hard-code table is much
>>>> more flexiblity.
>>> 
>>> If there's going to be 700 different variations depending on what people do with openc910, then allowing it to be passed sounds like a good idea.
>> Yes, that is what we want. th1520 is a example.
>> 
>> Thanks for the review and correction! Your help is greatly appreciated
>> in improving th1520 upstream.
>> 
>>> 
>>> Thanks,
>>> Conor.
>> 
>> 
>> 
>> --
>> Best Regards
>> Guo Ren
> 
> 
> 
> -- 
> Best Regards
> Guo Ren