Re: [PATCH v2 ath-current 2/2] dt-bindings: net: wireless: ath11k-pci: remove obsolete firmware-name property

[Krzysztof Kozlowski <krzk@kernel.org>]

On 19/01/2026 02:34, Miaoqing Pan wrote:
> 
> 
> On 1/13/2026 3:28 PM, Krzysztof Kozlowski wrote:
>> On 14/12/2025 03:52, Miaoqing Pan wrote:
>>> The firmware-name property was introduced to allow end-users and
>>> integrators to select usecase specific firmware for the WCN6855.
>>> However, specifying firmware for an M.2 WLAN module in the Device
>>> Tree is not appropriate. Instead, this functionality will be handled
>>> within the ath11k driver.
>>>
>>> The driver has removed all support for firmware-name, and no upstream
>>> Device Tree files reference this property. Therefore, this patch
>>> removes the property from the binding and marks it as obsolete.
>>
>> No, it does not mark it obsolete. Point me to the place.
>>
>>>
>>> This is a DT ABI-breaking change, but safe since there are no in-tree
>>> users.
>>
>> It's not safe. What about my board using this WiFi? Or Mr. foo's board?
>>
>> Still NAK, you did not improve it.
>>
>>
> 
> I think it’s necessary to clarify the background here. As you can see 
> from the git log, all changes related to ath11k firmware-name were 
> submitted by me, and the intention was to allow the lemans-evk, 
> monaco-evk, and hamoa-iot-evk boards to specify dedicated firmware for 
> the WCN6855 Wi-Fi chip. However, the Wi-Fi‑related DTS nodes for these 
> boards have never been submitted upstream, because adding a 
> firmware-name property for an M.2 device is not appropriate and would be 
> difficult for the community to accept. Therefore, the original approach 
> was abandoned.

You added new ABI which can be used by anyone and your commit did not
help me to understand the impact on other users of this ABI.

> 
> The alternative solution is to propose a static lookup table that maps 
> device compatibles to firmware names. As a result, we have not submitted 
> any DTS patches adding firmware-name for those boards to date. This is 
> why I believe that removing firmware-name from the bindings is safe.
> 
> If this explanation is still not sufficient, please let me know what 
> additional steps are required for accepting these two patches. Thank you.

You need to deprecate the property and keep the ABI backwards compatible.

Best regards,
Krzysztof