Re: [PATCH v4 02/14] regulator: Add function to map modes to struct regulator_desc

From: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
To: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Cc: Mark Brown <broonie@kernel.org>,
	Kukjin Kim <kgene.kim@samsung.com>,
	Chanwoo Choi <cw00.choi@samsung.com>,
	Olof Johansson <olof@lixom.net>, Chris Zhong <zyw@rock-chips.com>,
	Abhilash Kesavan <kesavan.abhilash@gmail.com>,
	linux-samsung-soc@vger.kernel.org, linux-kernel@vger.kernel.org,
	devicetree@vger.kernel.org
Subject: Re: [PATCH v4 02/14] regulator: Add function to map modes to struct regulator_desc
Date: Tue, 04 Nov 2014 12:02:05 +0100	[thread overview]
Message-ID: <5458B22D.8000209@collabora.co.uk> (raw)
In-Reply-To: <1415097072.7941.5.camel@AMDC1943>

Hello Krzysztof,

Thanks a for your feedback.

On 11/04/2014 11:31 AM, Krzysztof Kozlowski wrote:
>> + *
>> + * @map_modes: Callback invoked to translate between hardware to standard modes.
> 
> Initially I thought it should map from standard to hardware. But then I
> looked at max77802 implementation and it maps from hardware to standard.
> Anyway I got confused (both are "modes" and both unsigned ints).
> 
> Could you describe which should be returned?
> 

Sure, maybe rewording to:

"Callback invoked to translate from hardware to standard modes." ?

But I'll add also document that the parameter should be a hardware
mode and the return value a standard mode.

 >>   */
>>  struct regulator_desc {
>>  	const char *name;
>> @@ -285,6 +287,8 @@ struct regulator_desc {
>>  	unsigned int enable_time;
>>  
>>  	unsigned int off_on_delay;
>> +
>> +	unsigned int (*map_modes)(unsigned int mode);
> 
> Shouldn't this be in regulator ops?
> 

regulator ops are for the operations that a regulator support
(enable, disable, set mode, etc). All the thse are actions but
how to translate between hardware and standard modes is not an
action but a non-varying configuration of the regulator.

So I believe that regulator desc was what fit the most. I don't
have a strong opinion though if people think that it should be
in regulator ops instead.

Best regards,
Javier