[PATCH 1/2] dt-bindings: mfd: syscon: Add ti,dss-oldi-io-ctrl compatible

[Andrew Davis]

On 8/9/23 4:31 PM, Conor Dooley wrote:
> On Thu, Aug 10, 2023 at 12:30:03AM +0530, Aradhya Bhatia wrote:
>>
>>
>> On 09-Aug-23 23:29, Andrew Davis wrote:
>>> On 8/9/23 12:39 PM, Aradhya Bhatia wrote:
>>>> Hi Andrew,
>>>>
>>>> Thank you for quickly whipping up these patches! =)
>>>>
>>>> On 09-Aug-23 22:27, Andrew Davis wrote:
>>>>> Add TI DSS OLDI-IO control registers compatible. This is a region of 5
>>>>> 32bit registers found in the TI AM65 CTRL_MMR0 register space[0]. They
>>>>> are used to control the characteristics of the OLDI DATA/CLK IO as
>>>>> needed
>>>>> by the DSS display controller node.
>>>>
>>>> As long as the driver takes care of it, we can reuse the same compatible
>>>> even when OLDI IO Ctrl registers change from SoC to SoC, (in this case,
>>>> AM65 to AM62), right?
>>>>
>>>
>>> That depends, is the register space still "compatible" with the AM65
>>> version of this space? If not then we would want to qualify these
>>> with their SoC versions.
> 
> Even if they are compatible, having soc-specific compatibles with a
> fallback to the common oldi compatible string would be ideal.
> 
>> It is certainly not compatible. More on this below.
> 
> But as they're not compatible, that's kinda moot anyway.
> 
>>> A quick check of the documentation shows the register space is still
>>> 5 registers, 4 DATA and 1 CLK. The contents are different though, but
>>> since this compatible string is not used to match with a driver that
>>> would care (that is handled by the DSS node which does have different
>>> compatibles for each device), I'm actually not sure.
> 
>>> Guess we can leave
>>> that decision to the DT binding maintainers..
> 
> I'm not 100% sure what this decision actually is. Could you elaborate?
> 

I was not exactly aware of the extent of the incompatibilities.
If they are not compatible, then there is really no decision needed,
they must have different strings.

> 
>> Exactly. The DSS driver in our, as you like to call, evil-vendor-tree
>> uses the compatible information to decide which register offsets to
>> write to, and what to write.
>>
>>
>> On the register compatibility situation...
>>
>> AM62 OLDI IO Ctrl has, in total, 12 registers (as opposed to 5 in AM65).
>>
>> There are 4 Data and 1 Clk registers (per OLDI). And there are 2 OLDI
>> transmitters in AM62. However, their contents are different as you
>> noted.
>>
>> There are 2 more registers in AM62 (unlike AM65), that control
>> power-down and loop-back. And it is the power-down register, that will
>> be primarily used by the driver.
> 
>>>>> [0] https://www.ti.com/lit/pdf/spruid7
>>>>>
>>>>> Signed-off-by: Andrew Davis <afd at ti.com>
>>>>> ---
>>>>>    Documentation/devicetree/bindings/mfd/syscon.yaml | 1 +
>>>>>    1 file changed, 1 insertion(+)
>>>>>
>>>>> diff --git a/Documentation/devicetree/bindings/mfd/syscon.yaml
>>>>> b/Documentation/devicetree/bindings/mfd/syscon.yaml
>>>>> index 8103154bbb529..5029abd6d6411 100644
>>>>> --- a/Documentation/devicetree/bindings/mfd/syscon.yaml
>>>>> +++ b/Documentation/devicetree/bindings/mfd/syscon.yaml
>>>>> @@ -69,6 +69,7 @@ properties:
>>>>>                  - rockchip,rk3588-qos
>>>>>                  - rockchip,rv1126-qos
>>>>>                  - starfive,jh7100-sysmain
>>>>> +              - ti,dss-oldi-io-ctrl
> 
> So it sounds like this compatible, that appears to be generic, should
> instead be soc-specific as the register layout is different between
> SoCs?
> 

Sounds like it, I'll go and make this soc-specific.

Thanks,
Andrew

> Apologies if I have misunderstood.
> 
> Thanks,
> Conor.
> 
>>>>>                - const: syscon
>>>>>    
>>