Re: [PATCH 2/2] regulator: s5m8767: Modify parsing method of the voltage table of buck2/3/4

[Chanwoo Choi <cw00.choi@samsung.com>]

On 10/10/2013 07:36 PM, Mark Rutland wrote:
> On Thu, Oct 10, 2013 at 02:41:36AM +0100, Chanwoo Choi wrote:
>> The s5m8767 regulator driver parse always the voltage table of buck2/3/4.
>> If gpio_dvs feature isn't used and dts haven't included the voltage table
>> of buck2/3/4, s5m8767 regulator driver return error and file probe state.
>>
>> This patch check only voltage table of buck on s5m8767_pmic_dt_parse_pdata()
>> if buck[2-4]_gpiodvs is true.
>>
>> Signed-off-by: Chanwoo Choi <cw00.choi@samsung.com>
>> Signed-off-by: YoungJun Cho <yj44.cho@samsung.com>
>> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
>> ---
>>  drivers/regulator/s5m8767.c | 54 +++++++++++++++++++++++----------------------
>>  1 file changed, 28 insertions(+), 26 deletions(-)
>>
>> diff --git a/drivers/regulator/s5m8767.c b/drivers/regulator/s5m8767.c
>> index cb6cdb3..29999c0 100644
>> --- a/drivers/regulator/s5m8767.c
>> +++ b/drivers/regulator/s5m8767.c
>> @@ -522,7 +522,7 @@ static int s5m8767_pmic_dt_parse_pdata(struct platform_device *pdev,
>>  	struct device_node *pmic_np, *regulators_np, *reg_np;
>>  	struct sec_regulator_data *rdata;
>>  	struct sec_opmode_data *rmode;
>> -	unsigned int i, dvs_voltage_nr = 1, ret;
>> +	unsigned int i, dvs_voltage_nr = 8, ret;
>>  
>>  	pmic_np = iodev->dev->of_node;
>>  	if (!pmic_np) {
>> @@ -586,15 +586,39 @@ static int s5m8767_pmic_dt_parse_pdata(struct platform_device *pdev,
>>  		rmode++;
>>  	}
>>  
>> -	if (of_get_property(pmic_np, "s5m8767,pmic-buck2-uses-gpio-dvs", NULL))
>> +	if (of_get_property(pmic_np, "s5m8767,pmic-buck2-uses-gpio-dvs", NULL)) {
>>  		pdata->buck2_gpiodvs = true;
> 
> As this line is being changed, please switch to of_property_read_bool.

OK.

> 
>>  
>> -	if (of_get_property(pmic_np, "s5m8767,pmic-buck3-uses-gpio-dvs", NULL))
>> +		if (of_property_read_u32_array(pmic_np,
>> +				"s5m8767,pmic-buck2-dvs-voltage",
>> +				pdata->buck2_voltage, dvs_voltage_nr)) {
> 
> Won't this break existing (not conforming to the binding) dts?
> 
> $(git grep s5m8767,pmic-buck2-dvs-voltage) shows me
> arch/arm/boot/dts/exynos5250-arndale.dts is only has one entry in its
> s5m8767,pmic-buck3-uses-gpio-dvs property's list.

I checked it but exynos5250-arndale.dts haven't included pmic-buck3-uses-gpio-dvs property.
What is reference code?

> 
>> +			dev_err(iodev->dev, "buck2 voltages not specified\n");
>> +			return -EINVAL;
>> +		}
>> +	}
>> +
>> +	if (of_get_property(pmic_np, "s5m8767,pmic-buck3-uses-gpio-dvs", NULL)) {
>>  		pdata->buck3_gpiodvs = true;
> 
> Another of_proeprty_read_bool candidate.

OK.

> 
>>  
>> -	if (of_get_property(pmic_np, "s5m8767,pmic-buck4-uses-gpio-dvs", NULL))
>> +		if (of_property_read_u32_array(pmic_np,
>> +				"s5m8767,pmic-buck3-dvs-voltage",
>> +				pdata->buck3_voltage, dvs_voltage_nr)) {
>> +			dev_err(iodev->dev, "buck3 voltages not specified\n");
>> +			return -EINVAL;
>> +		}
>> +	}
>> +
>> +	if (of_get_property(pmic_np, "s5m8767,pmic-buck4-uses-gpio-dvs", NULL)) {
>>  		pdata->buck4_gpiodvs = true;
> 
> Similarly.
> 
> In general I'm confused by the need for the *-uses-gpio-dvs properties.
> Are they not implied by the presence of *-dvs-voltage properties?
> 

The s5m8767 support the power control using DVS(Dynamic Voltage scaling) by using
GPIO or I2C interface between PMIC and AP. 

When controlling DVS by using GPIO interface, *-uses-gpio-dvs/*-dvs-voltage properties is used.
The s5m8767 pmic control DVS of one only buck using GPIO. If specific target use GPIO interface
to control buck2 DVS, must set buck2-uses-gpio-dvs as true and buck3/4-uses-gpio-dvs is false.
Also, GPIO interface include three gpio pin which select total 8 step voltage according to
three gpio state. *-dvs-voltage properties include the voltage table of 8 step.

Also, you can check the aim of *-uses-gpio-dvs/*-dvs-voltage properties
on Documentation/devicetree/bindings/regulator/s5m8767-regulator.txt.