Re: [PATCH v2 1/5] dt-bindings: phy: cadence-torrent: Add optional input reference clock for PLL1

[Roger Quadros]

On 14/12/2023 09:22, Krzysztof Kozlowski wrote:
> On 14/12/2023 08:02, Swapnil Kashinath Jakhade wrote:
>> Hi Krzysztof,
>>
>>> -----Original Message-----
>>> From: Krzysztof Kozlowski <krzysztof.kozlowski@xxxxxxxxxx>
>>> Sent: Wednesday, December 13, 2023 12:19 PM
>>> To: Swapnil Kashinath Jakhade <sjakhade@xxxxxxxxxxx>; vkoul@xxxxxxxxxx;
>>> kishon@xxxxxxxxxx; robh+dt@xxxxxxxxxx; krzysztof.kozlowski+dt@xxxxxxxxxx;
>>> conor+dt@xxxxxxxxxx; linux-phy@xxxxxxxxxxxxxxxxxxx; linux-
>>> kernel@xxxxxxxxxxxxxxx; devicetree@xxxxxxxxxxxxxxx
>>> Cc: Milind Parab <mparab@xxxxxxxxxxx>; rogerq@xxxxxxxxxx; s-
>>> vadapalli@xxxxxx
>>> Subject: Re: [PATCH v2 1/5] dt-bindings: phy: cadence-torrent: Add optional
>>> input reference clock for PLL1
>>>
>>> EXTERNAL MAIL
>>>
>>>
>>> On 12/12/2023 12:48, Swapnil Jakhade wrote:
>>>> Torrent PHY can have two input reference clocks. Update bindings
>>>
>>> It already supports two.
>>>
>>
>> Thanks for your comments.
>> refclk and pll1_refclk are the two input reference clocks for the PLLs.
>> phy_en_refclk is used to enable output reference clock in some cases.
> 
> Why input clock is used to enable output reference clock?

Looking at the driver code, "phy_en_refclk" is used at 2 places only to
set the parent of 2 clocks apparently called "received reference clock"
and "derived reference clock", either of which can be used to enable an
optional reference clock output feature.

My understanding is that it is a separate refclk input than 'refclk' (pll0)
or 'pll1_refclk' which can be used to enable a reference clock output feature.

So this PHY can support a total of 3 input reference clocks:
"refclk" is PLL0 reference clock and is required on all platforms
"pll1_refclk" is PLL1 reference clock and is required on some platforms
"phy_en_refclk" is reference for output reference clock generator and is optional on all platforms

Swapnil, can you please confirm my understanding? Thanks!

> 
>>
>>>> to support dual reference clock multilink configurations.
>>>>
>>>> Signed-off-by: Swapnil Jakhade <sjakhade@xxxxxxxxxxx>
>>>> ---
>>>>  .../devicetree/bindings/phy/phy-cadence-torrent.yaml        | 6 +++---
>>>>  1 file changed, 3 insertions(+), 3 deletions(-)
>>>>
>>>> diff --git a/Documentation/devicetree/bindings/phy/phy-cadence-
>>> torrent.yaml b/Documentation/devicetree/bindings/phy/phy-cadence-
>>> torrent.yaml
>>>> index dfb31314face..98946f549895 100644
>>>> --- a/Documentation/devicetree/bindings/phy/phy-cadence-torrent.yaml
>>>> +++ b/Documentation/devicetree/bindings/phy/phy-cadence-torrent.yaml
>>>> @@ -35,14 +35,14 @@ properties:
>>>>      minItems: 1
>>>>      maxItems: 2
>>>>      description:
>>>> -      PHY reference clock for 1 item. Must contain an entry in clock-names.
>>>> -      Optional Parent to enable output reference clock.
>>>> +      PHY input reference clocks - refclk & pll1_refclk (optional).
>>>> +      Optional Parent to enable output reference clock (phy_en_refclk).
>>>
>>> So third clock? But you allow only two? Confusing.
>>>
>>
>> Yes, if both refclk and pll1_refclk are present, phy_en_refclk can't be used.
>>
>>>>
>>>>    clock-names:
>>>>      minItems: 1
>>>>      items:
>>>>        - const: refclk
>>>> -      - const: phy_en_refclk
>>>> +      - enum: [ pll1_refclk, phy_en_refclk ]
>>>
>>> This does not match your commit msg. You already had two clocks there.
>>>
>> Yes, but refclk was the single input reference clock used for PLLs earlier.
>> As stated in commit message, a new input reference clock (pll1_refclk) is added here.
> 
> existing phy_en_refclk is also input reference clock, isn't it?
> 
>>
> 
> Best regards,
> Krzysztof
> 

-- 
cheers,
-roger