On 09/23/2013 04:01 PM, Rhyland Klein wrote:
> On 9/23/2013 5:53 PM, Stephen Warren wrote:
>> On 09/19/2013 10:18 AM, Rhyland Klein wrote:
>>> Adding driver support for bq24735 charger chipset.
>>
>>> diff --git a/Documentation/devicetree/bindings/power_supply/ti,bq24735.txt b/Documentation/devicetree/bindings/power_supply/ti,bq24735.txt

>>> + - ti,charge-current : Used to control and set the charging current. This value
>>> +   must follow the below guidelines:
>>> +	bit 0 - 5:	Not used
>>> +	bit 6:		1 = Adds 64mA of charger current
>>> +	bit 7:		1 = Adds 128mA of charger current
>>> +	bit 8:		1 = Adds 256mA of charger current
>>> +	bit 9:		1 = Adds 512mA of charger current
>>> +	bit 10:		1 = Adds 1024mA of charger current
>>> +	bit 11:		1 = Adds 2048mA of charger current
>>> +	bit 12:		1 = Adds 4096mA of charger current
>>> +	bit 13 - 15:	Not used
>>
>> That's a little odd. Why not just put the number of mA directly into the
>> property unshifted?
> 
> This is how the hw register is defined, its the literal number of mA.
> This is cleaned up in the upcoming revision.

OK. If you still want to use the raw register encoding, which seems
reasonable, why not just say:

ti,charge-current: Value for charge current register as described in the
HW documentation.

>>> +   Setting the value to < 128mA or > 8.128A terminates charging.
>>
>> "terminates charging" is a driver action, not a description of HW. It's
>> fine to say that what min/max value should be specified in the property
>> for it to be valid, but not what action SW should take in response to that.
> 
> This isn't sw. This is defined in the HW documentation as to what
> happens if an invalid value is used.

OK, if the HW documentation already says that, I don't see the need for
the DT binding to also say it; I would suggest just removing that text.