Message-ID: <1f8dd436-1f01-48f5-b2f5-2ba9d998be3b@kernel.org>
Date: Mon, 17 Jun 2024 14:11:20 +0200
From: Krzysztof Kozlowski <krzk@...nel.org>
To: Aisheng Dong <aisheng.dong@....com>, Vabhav Sharma
 <vabhav.sharma@....com>, Rob Herring <robh@...nel.org>,
 Krzysztof Kozlowski <krzk+dt@...nel.org>, Conor Dooley
 <conor+dt@...nel.org>, Franck Lenormand <franck.lenormand@....com>,
 Shawn Guo <shawnguo@...nel.org>, Sascha Hauer <s.hauer@...gutronix.de>,
 Pengutronix Kernel Team <kernel@...gutronix.de>,
 Fabio Estevam <festevam@...il.com>, Peng Fan <peng.fan@....com>
Cc: "devicetree@...r.kernel.org" <devicetree@...r.kernel.org>,
 "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
 "imx@...ts.linux.dev" <imx@...ts.linux.dev>,
 "linux-arm-kernel@...ts.infradead.org"
 <linux-arm-kernel@...ts.infradead.org>, Varun Sethi <V.Sethi@....com>,
 Silvano Di Ninno <silvano.dininno@....com>,
 Pankaj Gupta <pankaj.gupta@....com>, Frank Li <frank.li@....com>,
 Daniel Baluta <daniel.baluta@....com>
Subject: Re: [PATCH 1/4] dt-bindings: firmware: secvio: Add device tree
 bindings

On 17/06/2024 12:11, Aisheng Dong wrote:
> Hi Krzystof,
> 
>> From: Krzysztof Kozlowski <krzk@...nel.org>
>> Sent: 2024年6月16日 15:34
>> On 13/06/2024 10:48, Aisheng Dong wrote:
>>> Hi Krzysztof,
>>>
>>>> From: Krzysztof Kozlowski <krzk@...nel.org>
>>>> Sent: 2024年6月13日 14:14
>>>>
>>>> On 12/06/2024 09:20, Aisheng Dong wrote:
>>>>> Hi Krzysztof
>>>>>
>>>>>> From: Krzysztof Kozlowski <krzk@...nel.org>
>>>>>> Sent: 2024年6月7日 15:08
>>>>>>
>>>>>> On 07/06/2024 06:58, Vabhav Sharma wrote:
>>>>>>>>
>>>>>>>> Missing SoC compatibles.
>>>>>>> Ok, I will use fsl,imx8dxl-sc-secvio
>>>>>>>>
>>>>>>>> So no, that's just abuse of DT to instantiate driver.
>>>>>>>>
>>>>>>>> NAK. Drop the binding.
>>>>>>> I will detail the dt binding to describe the real hardware
>>>>>>
>>>>>> Still looks like way just to instantiate driver. Why it cannot be
>>>>>> part of existing firmware SCU node?
>>>>>>
>>>>>
>>>>> Technically yes. But SCU case is a little bit complicated as
>>>>> there're many functions and all of them are already added as sub
>>>>> nodes in SCU node for consistency and handling platform difference.
>>>>>
>>>>> I guess some of them, e.g. rtc, could be part of SCU node (reuse)
>>>>> while some couldn't. e.g. pinctrl Do you want us to only make secvio
>>>>> reuse existing SCU node?
>>>>
>>>> Yes
>>>>
>>>
>>> Digging a bit more on the implementation. It seems there will be a
>>> 'parent depends on child' issue when reusing the parent SCU node for secvio.
>>> Considering the defer probe support and ocotop could be modules, I'm
>>> still thinking If any solution. Do you have a good suggestion?
>>
>> I don't see any problem there. You will have even worse if making it children
>> and using populate - unpredictable order.
> 
> Sorry I didn't find a good solution without making OCOTP node to be party of SCU
> node too which we can't due to extra required properties of nvmem properties. 
> E.g. #addr/size-cells.
> 
> I think the key problem is OCOTP is already a child node and there's a mismatch issue
> If only making secvio, which depends on OCOTP, re-use their parent SCU node.
> Maybe you have a good idea. Would you mind share a bit more?
> 
> BTW, I don't understand the order problem if making secvio a children as all
> child nodes are Individual function devices and supports defer probe well.

You keep talking about drivers but what does it have to do with bindings
at all?

You can order your probe however you wish regardless of bindings. You
can handle or not handle deferred probe. You can create circular
dependencies or break them. All regardless of the bindings.

Best regards,
Krzysztof

Powered by blists - more mailing lists