Re: [PATCH 1/3] dt-bindings: spi: add SpacemiT K1 SPI support

From: Alex Elder <elder@riscstar.com>
To: Yixun Lan <dlan@gentoo.org>
Cc: broonie@kernel.org, robh@kernel.org, krzk+dt@kernel.org,
	conor+dt@kernel.org, linux-spi@vger.kernel.org,
	devicetree@vger.kernel.org, paul.walmsley@sifive.com,
	palmer@dabbelt.com, aou@eecs.berkeley.edu, alex@ghiti.fr,
	p.zabel@pengutronix.de, spacemit@lists.linux.dev,
	linux-riscv@lists.infradead.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 1/3] dt-bindings: spi: add SpacemiT K1 SPI support
Date: Wed, 17 Sep 2025 21:59:35 -0500	[thread overview]
Message-ID: <f7e12797-ea17-4e92-bd25-cc562c66d2a7@riscstar.com> (raw)
In-Reply-To: <20250918001632-GYA1270371@gentoo.org>

On 9/17/25 7:16 PM, Yixun Lan wrote:
> Hi Alex,
> 
> On 18:40 Wed 17 Sep     , Alex Elder wrote:
>> On 9/17/25 6:15 PM, Yixun Lan wrote:
>>> Hi Alex,
>>>
>>> On 17:07 Wed 17 Sep     , Alex Elder wrote:
>>>> Add support for the SPI controller implemented by the SpacemiT K1 SoC.
>>>>
>>>> Signed-off-by: Alex Elder <elder@riscstar.com>
>>>> ---
>>>>    .../bindings/spi/spacemit,k1-spi.yaml         | 94 +++++++++++++++++++
>>>>    1 file changed, 94 insertions(+)
>>>>    create mode 100644 Documentation/devicetree/bindings/spi/spacemit,k1-spi.yaml
>>>>
>>>> diff --git a/Documentation/devicetree/bindings/spi/spacemit,k1-spi.yaml b/Documentation/devicetree/bindings/spi/spacemit,k1-spi.yaml
>>>> new file mode 100644
>>>> index 0000000000000..5abd4fe268da9
>>>> --- /dev/null
>>>> +++ b/Documentation/devicetree/bindings/spi/spacemit,k1-spi.yaml
> ..
>>>> +
>>>> +  spacemit,k1-ssp-id:
>>>> +    description: SPI controller number
>>>> +    $ref: /schemas/types.yaml#/definitions/uint32
>>> could you explain a little bit why this vendor specific property should
>>> be introduced? I took a look at the code, and didn't get the reason
>>> behind.. or what's the problem of simply using "pdev->id"?
>>
>> This property was carried over from the vendor code.  It is
> inherit from vendor code isn't a valid reason
> 
>> optional, and if it isn't specified, the platform device ID (-1)
>> will be used.  It's just intended to provide a well-defined ID
>> for a particular SPI controller.
>>
> while looking at the code, it seems you can use alias to map specific id
> to the spi controller, it even can do non-linear map, something like
> 	spi0 = &spi3;
I've never used this before, but yes, it looks like it's exactly
what I want.  I'll just get rid of the "spacemit,k1-ssp-id" DT
property entirely.  Easy.

> plese check of_alias_get_id()
> 
> note, I haven't actually verified on board, just look through the code
> 
>>> we should really be careful to introduce vendor specific property..
>>
>> If there were a standard way of doing this I'd love to use it.

Looks like you have told me the standard way of doing this.

Thank you.

					-Alex

>>
>> And if it isn't necessary, please just explain to me why.  I
>> have no problem removing it.
>>
> on the opposite, please have explicit good reason to introduce vendor
> speifici property, and if there is generic way, then we shouldn't do it
> 

_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv