On 22/03/2019 04:37, Sameer Pujar wrote:
>
> On 3/21/2019 11:03 PM, Jon Hunter wrote:
>> On 21/03/2019 17:23, Sameer Pujar wrote:
>>
>> ...
>>
>> >>> data = of_device_get_match_data(&pdev->dev);
>>>>> if (!data) {
>>>>> dev_err(&pdev->dev, "no device match found\n");
>>>>> return -ENODEV;
>>>>> }
>>>>> + chip_pm->clk_data = data;
>>>>> + platform_set_drvdata(pdev, chip_pm);
>>>> So why has this been moved? There is a comment in gic_runtime_resume()
>>>> on why this was set after pm_runtime_get_sync() in probe.
>>> This is moved up to make sure clk_data in chip_pm is populated by the
>>> time
>>> gic_get_clocks() is called. The comment in runtime_resume() would still
>>> hold
>>> good, gic would be NULL till gic_of_init_child() is called. May be the
>>> comment
>>> needs to be slightly updated, because chip_pm is the new driver data
>>> now.
>> I see, but I am not sure it is necessary as you are directly passing
>> chip_pm to gic_get_clocks().
> Ah yes, chip_pm need not be passed for gic_get_clocks(). But driver data
> needs
> to be set before runtime_resume() happens.
OK yes indeed. Then that should be fine.
Thanks
Jon
--
nvpublic
