[PATCH 2/4] ARM: STi: Supply I2C configuration to STiH416 SoC

[Maxime COQUELIN]

On 09/19/2013 03:01 PM, Srinivas KANDAGATLA wrote:
> On 19/09/13 08:16, Maxime COQUELIN wrote:
>> Hi Srini,
>>
>> On 09/18/2013 03:17 PM, Srinivas KANDAGATLA wrote:
>>> On 18/09/13 13:46, Maxime COQUELIN wrote:
>>>> On 09/18/2013 02:03 PM, Lee Jones wrote:
>>>>>>>> This patch supplies I2C configuration to STiH416 SoC.
>>>>>>>>
>>>>>>>> Cc: Srinivas Kandagatla <srinivas.kandagatla at st.com>
>>>>>>>> Signed-off-by: Maxime Coquelin <maxime.coquelin at st.com>
>>>>>>>> ---
>>>>>>>>  arch/arm/boot/dts/stih416-pinctrl.dtsi |   35 ++++++++++++++++++++
>>>>>>>>  arch/arm/boot/dts/stih416.dtsi         |   57 ++++++++++++++++++++++++++++++++
>>>>>>>>  2 files changed, 92 insertions(+)
>>>>>>>>
>>>>>>>> diff --git a/arch/arm/boot/dts/stih416-pinctrl.dtsi b/arch/arm/boot/dts/stih416-pinctrl.dtsi
>>>>>> I genuinely don't know the answer to this question, but are these
>>>>>> nodes identical to the ones you've just put in the stih415 DTSI file?
>>>>>> If so, I think it will be worth creating a stih41x DTSI rather than
>>>>>> duplicating lots of stuff unnecessarily.
>>>> There are close to be identical indeed.
>>>> For the clocks and pinctrl, the references names are the same, but they are
>>>> pointing on different nodes, as STiH415 and STiH416 have their own
>>>> clocks and pinctrl dtsi files.
>>>>
>>>> Srini, what is opinion about this?
>>> There is already a stih41x.dtsi file, but I don't think it is the right
>>> place for the pinctrl nodes there.
>>>
>>> Am not OK with the idea of common pinctrl nodes for STiH415 and STiH416
>>> for two reasons.
>>>
>>> 1> If we common up the pinctrl nodes, it will be very difficult to
>>> accommodate new pinctrls layout which is not guaranteed to be in same
>>> layout in future SOCs.
>>>
>>> 2> The retiming values in the pinctrl nodes tend to change as per SOC,
>>> so it will be difficult to manage it if we common it up.
>>>
>>> Am sure we can come up with a dt layout which can reduce duplication,
>>> but we have to be careful here not to lose the flexiblity to accommodate
>>> new picntrl layouts, new retimings values based on SOC.
>> Ok. What do you think of declaring the i2c nodes inside the stih41x.dtsi
>> file,
>> and overload them with the pinctrl and clock properties in the stih416
>> and stih415 dtsi files?
> Am not very comfortable with this idea.
>
> As there is no guarantee that the interrupt number/memory map and the
> i2c numbering will be same in future SOCs or other IPs.
>
> You might be already aware that the number of i2cs on each SOC are
> different as example on STiH415 we have 10 SSCs and on STiH416 we have
> 11 SSCs. So, At what point you decide that which devices/IPs should be
> in stih41x and which should in stih415/Stih416?
Yes, I know there is one more SSC on STiH416.

On one hand, this could add some confusion. But on the other hand,
someone who will need to activate a SSP will know which one he has
to activate.

>
> Each i2c node will save around 5 lines if we common it up, but if the
> interrupt number or map changes this difference will be negligible.
>
> Common up at this level and mixing un-common ones in stih415.dtsi or
> stih416.dtsi will add confusion to readers as the information is split
> at multiple places.
I agree it will be messy if one part of the node declared at one place,
and the rest at another place.
>
> IMO the common up idea sounds good but reduces the readability and has
> no effect on final dtb size.

Fair enough. Lee, are you ok with keeping it as is?

Thanks,
Maxime
>
> Thanks,
> srini
>
>
>> Regards,
>> Maxime
>>>
>>> thanks,
>>> srini
>>>
>>>
>>
>>
>