Re: [PATCH v13 1/2] cpufreq: Add Kryo CPU scaling driver

[Sudeep Holla]

On 25/05/18 05:05, Viresh Kumar wrote:
> On 25-05-18, 07:00, Ilia Lin wrote:
>>
>>
>> On May 25, 2018 6:54:12 AM GMT+03:00, Viresh Kumar <viresh.kumar@xxxxxxxxxx> wrote:
>>> On 24-05-18, 18:03, Ilia Lin wrote:
>>>> +static int __init qcom_cpufreq_kryo_init(void)
>>>> +{
>>>> +	struct device_node *np;
>>>> +	struct device *cpu_dev;
>>>> +	int ret;
>>>> +
>>>> +	cpu_dev = get_cpu_device(0);
>>>> +	if (NULL == cpu_dev)
>>>> +		ret = -ENODEV;
>>>> +
>>>> +	np = dev_pm_opp_of_get_opp_desc_node(cpu_dev);
>>>> +	if (IS_ERR(np))
>>>> +		return PTR_ERR(np);
>>>> +
>>>> +	ret = of_device_is_compatible(np, "operating-points-v2-kryo-cpu");
>>>> +	of_node_put(np);
>>>> +	if (!ret)
>>>> +		return -ENOENT;
>>>> +
>>>
>>> I hate the fact that it is taking so long to get done with this. But
>>> can't you
>>> just check machine compatibility instead of this complicated setup to
>>> check OPP
>>> node ? Like:
>>>
>>>        if (!of_device_is_compatible("qcom,apq8096") &&
>>>            !of_device_is_compatible("qcom,msm8996"))
>>>                return;
>>
>> I have to check the "operating-points-v2-kryo-cpu" anyway, so I moved it from probe to the init.
> 
> Okay, leave it as is then. Don't send anything yet and wait for Sudeep to
> respond.
> 

I have already mentioned that I can't suggest since I don't have much
knowledge on QCOM PM. I see lots of patches flying around these days
with multiple people doing same things in different ways. That's why I
suggested Ilia to check with Taniya or Saravana or SBoyd to see what's
the best thing to do in this context. I don't prefer using OPP
compatible and would go with machine compatibility as you suggested.
But I am fine with OPP, just wanted to make sure it won't clash with
anything else.
-- 
Regards,
Sudeep