[RFC] device.h: add device_set_platdata routine

[viresh kumar]

On 03/01/2011 01:29 PM, Uwe Kleine-König wrote:
> [added gregkh and lkml to Cc:]
> 

thanks.

> On Tue, Mar 01, 2011 at 10:03:20AM +0530, Viresh Kumar wrote:
>> device.h supports device_get_platdata but doesn't support device_set_platdata.
>> This routine is required by platforms in which device structure is declared
>> in a machine specific file and platform data comes from board specific file.
>>
>> This will be used by SPEAr patches sent in separate patch series.
>>
>> Signed-off-by: Viresh Kumar <viresh.kumar at st.com>
>> ---
>>  include/linux/device.h |    5 +++++
>>  1 files changed, 5 insertions(+), 0 deletions(-)
>>
>> diff --git a/include/linux/device.h b/include/linux/device.h
>> index 1bf5cf0..6ce0f20 100644
>> --- a/include/linux/device.h
>> +++ b/include/linux/device.h
>> @@ -576,6 +576,11 @@ static inline void *dev_get_platdata(const struct device *dev)
>>  	return dev->platform_data;
>>  }
>>  
>> +static inline void dev_set_platdata(struct device *dev, void *platdata)
>> +{
>> +	dev->platform_data = platdata;
>> +}
>> +
> Note that dev->platform_data was designed to hold dynamically allocated
> memory, at least it's kfreed in platform_device_release.  And note there
> is platform_device_add_data that kmemdups its argument into
> pdev->dev.platform_data.
> 

Ok. So we should use platform_device_add_data instead and mark our platform
data's struct as __init. So that it doesn't consume any memory after
this routine is done??

> Compared to your dev_set_platdata platform_device_add_data only works
> for platform_devices, don't know if it's worth to change that.
> 

Currently i need this for platform devs only. So its good enough for me.

-- 
viresh