[PATCH V2 4/4] mfd: mediatek: add MT6323 support to MT6397 driver

[John Crispin]

On 26/01/2016 09:34, Lee Jones wrote:
> On Tue, 26 Jan 2016, John Crispin wrote:
>> On 26/01/2016 04:07, Henry Chen wrote:
>>> On Mon, 2016-01-25 at 19:59 +0100, John Crispin wrote:
>>>>
>>>> On 25/01/2016 19:44, Matthias Brugger wrote:
>>>>> On Monday 25 Jan 2016 16:36:40 John Crispin wrote:
>>>>>> Hi,
>>>>>>
>>>>>> On 25/01/2016 13:41, Lee Jones wrote:
>>>>>>> Please honour the subject format of the subsystem you are contributing
>>>>>>> to.
>>>>>>>
>>>>>>> `git log --oneline -- $subsystem` gives you this.
>>>>>>>
>>>>>>> On Mon, 25 Jan 2016, John Crispin wrote:
>>>>>>>> Signed-off-by: John Crispin <blogic at openwrt.org>
>>>>>>>> ---
>>>>>>
>>>>>> [...]
>>>>>>
>>>>>>>> @@ -261,6 +271,15 @@ static int mt6397_probe(struct platform_device
>>>>>>>> *pdev)
>>>>>>>>
>>>>>>>>  	}
>>>>>>>>  	
>>>>>>>>  	switch (id & 0xff) {
>>>>>>>>
>>>>>>>> +	case MT6323_CID_CODE:
>>>>>>>> +		mt6397->int_con[0] = MT6323_INT_CON0;
>>>>>>>
>>>>>>> This is confusing.  You're still using memory allocated for a mt6397
>>>>>>> device.
>>>>>>
>>>>>> the variable is currently defined as struct mt6397_chip *mt6397;
>>>>>> shall i only change the name or also create a patch to rename the struct ?
>>>>>>
>>>>>
>>>>> I think we should rename the struct and the file as well.
>>>>>
>>>>> Cheers,
>>>>> Matthias
>>>>
>>>> Hi,
>>>>
>>>> that would have been my next question. renaming the struct would imply
>>>> renaming the driver and the whole namespace contained within. We would
>>>> then also need to change the Kconfig and Makefile. I am happy to do this
>>>> but want to be sure that is is actually wanted.
>>>>
>>>> 	John
>>> Hi,
>>>
>>> Since mt6323 was similar with mt6397, I think we can reuse the
>>> mt6397_chip without duplicate code.
>>>
>>> Maybe we can rename the local variable name to avoid confusing.
>>>
>>> struct mt6397_chip *mt_pmic;
>>> ...
>>> ...
>>> switch (id & 0xff) {
>>> 	case MT6323_CID_CODE:
>>> 		mt_pmic->int_con[0] = MT6323_INT_CON0;
>>> 		mt_pmic->int_con[1] = MT6323_INT_CON1;
>>> ...
>>> ...
>>>
>>> Henry
>>
>> Hi,
>>
>> IMHO we should either rename the namespace or not. renaming some
>> variables seems weird as that will just move the confusion/inconsistency
>> to another place in the code. I am however rather indifferent on this
>> matter.
> 
> It's common to name a driver after the device which was enabled first,
> so no need to rename the files or CONFIGs; however, it does seem
> prudent to generify the struct (both parts).
> 

Hi,

renaming struct mt6397_chip is not as straightforward as it seems as
other drivers such as pinctrl and rtc share the data structure with the
mfd driver.

i have kept the structs name as is for now and only renamed the instance.

If you want i can create a separate series that addresses only the
renaming of the structure. The "rename struct mt6397" patch would need
to do a change to drivers from 3 subsystems in 1 single commit as not to
break compile and merge order.

	John