Re: [PATCH 1/4] arm64: dts: qcom: sc8280xp: rename i2c5 to i2c21

[Konrad Dybcio]

On 13.12.2022 16:17, Johan Hovold wrote:
> On Tue, Dec 13, 2022 at 03:54:05PM +0100, Johan Hovold wrote:
>> On Mon, Dec 12, 2022 at 01:23:11PM -0500, Brian Masney wrote:
>>> According to the downstream 5.4 kernel sources for the sa8540p,
>>> i2c@894000 is labeled i2c bus 21, not 5. The interrupts and clocks
>>> also match. Let's go ahead and correct the name that's used in the
>>> three files where this is listed.
>>>
>>> Signed-off-by: Brian Masney <bmasney@xxxxxxxxxx>
>>> Fixes: 152d1faf1e2f3 ("arm64: dts: qcom: add SC8280XP platform")
>>> Fixes: ccd3517faf183 ("arm64: dts: qcom: sc8280xp: Add reference device")
>>> Fixes: 32c231385ed43 ("arm64: dts: qcom: sc8280xp: add Lenovo Thinkpad X13s devicetree")
>>
>>> diff --git a/arch/arm64/boot/dts/qcom/sc8280xp.dtsi b/arch/arm64/boot/dts/qcom/sc8280xp.dtsi
>>> index 109c9d2b684d..875cc91324ce 100644
>>> --- a/arch/arm64/boot/dts/qcom/sc8280xp.dtsi
>>> +++ b/arch/arm64/boot/dts/qcom/sc8280xp.dtsi
>>> @@ -827,7 +827,7 @@ qup2_uart17: serial@884000 {
>>>  				status = "disabled";
>>>  			};
>>>  
>>> -			qup2_i2c5: i2c@894000 {
>>> +			qup2_i2c21: i2c@894000 {
>>
>> Note that the node is labelled qup2_i2c5 and not qup_i2c5.
>>
>> That is, the QUP nodes are labelled using two indices, and specifically
>>
>> 	qup2_i2c5
>>
>> would be another name for
>>
>> 	qup_i2c21
>>
>> if we'd been using such a flat naming scheme (there are 8 engines per
>> QUP).
>>
>> So there's nothing wrong with how these nodes are currently named, but
>> mixing the two scheme as you are suggesting would not be correct.
> 
> It appears sc8280xp is the only qcom platform using a qup prefix (even
> if some older platform use a blsp equivalent), and we're not even using
> it consistently as we, for example, have both
> 
> 	qup2_uart17, and
> 	qup2_i2c5
> 
> (where the former should have been qup2_uart1).
> 
> So either we fix up the current labels or just drop the qup prefixes and
> use a flat naming scheme (e.g. uart17 and i2c21).
Oh, I didn't notice that! I suppose sticking with i2cN as we've been
doing ever since i2c-geni was introduced sounds like the best option..

Konrad
> 
> Either way, there's no need for any Fixes tags as this isn't a bug.
> 
> Johan