Re: [PATCH v2 1/4] soc: qcom: ice: Remove platform_driver support and expose as a pure library

[Konrad Dybcio]

On 2/10/26 1:19 PM, Manivannan Sadhasivam wrote:
> On Tue, Feb 10, 2026 at 10:39:54AM +0100, Konrad Dybcio wrote:
>> On 2/10/26 7:56 AM, Manivannan Sadhasivam via B4 Relay wrote:
>>> From: Manivannan Sadhasivam <manivannan.sadhasivam@xxxxxxxxxxxxxxxx>
>>>
>>> The current platform driver design causes probe ordering races with
>>> consumers (UFS, eMMC) due to ICE's dependency on SCM firmware calls. If ICE
>>> probe fails (missing ICE SCM or DT registers), devm_of_qcom_ice_get() loops
>>> with -EPROBE_DEFER, leaving consumers non-functional even when ICE should
>>> be gracefully disabled. devm_of_qcom_ice_get() cannot know if the ICE
>>> driver probe has failed due to above reasons or it is waiting for the SCM
>>> driver.
>>
>> [...]
>>
>>> -static void qcom_ice_put(const struct qcom_ice *ice)
>>> +static void qcom_ice_put(struct kref *kref)
>>>  {
>>> -	struct platform_device *pdev = to_platform_device(ice->dev);
>>> -
>>> -	if (!platform_get_resource_byname(pdev, IORESOURCE_MEM, "ice"))
>>> -		platform_device_put(pdev);
>>> +	platform_device_put(to_platform_device(ice_handle->dev));
>>> +	ice_handle = NULL;
>>>  }
>>>  
>>>  static void devm_of_qcom_ice_put(struct device *dev, void *res)
>>>  {
>>> -	qcom_ice_put(*(struct qcom_ice **)res);
>>> +	const struct qcom_ice *ice = *(struct qcom_ice **)res;
>>> +	struct platform_device *pdev = to_platform_device(ice->dev);
>>> +
>>> +	if (!platform_get_resource_byname(pdev, IORESOURCE_MEM, "ice"))
>>> +		kref_put(&ice_handle->refcount, qcom_ice_put);
>>
>> IIUC this makes the refcount go down only in the legacy DT case - why?
>>
> 
> It is the other way around, no? Absence of 'ice' reg range in the consumer node
> means it is using *new* binding.

Yeah obviously you're right

I suppose in the legacy case we don't need any refcounting, since there's only
a single reference, from the storage controller device..

This assumption would break if someone had a funny idea to specify the same
"ice" range for sdcc and ufs though (via the legacy binding), but let's hope
no one does that..

Konrad