Re: [PATCH v3 3/5] power: supply: max17040: Config alert SOC low level threshold from FDT

From: Matheus Castello <matheus@castello.eng.br>
To: Krzysztof Kozlowski <krzk@kernel.org>
Cc: sre@kernel.org, robh+dt@kernel.org, mark.rutland@arm.com,
	"Chanwoo Choi" <cw00.choi@samsung.com>,
	"Bartłomiej Żołnierkiewicz" <b.zolnierkie@samsung.com>,
	lee.jones@linaro.org, linux-pm@vger.kernel.org,
	devicetree@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v3 3/5] power: supply: max17040: Config alert SOC low level threshold from FDT
Date: Sun, 2 Jun 2019 19:26:20 -0300	[thread overview]
Message-ID: <69a4f003-4413-1316-6145-f8bef2171e86@castello.eng.br> (raw)
In-Reply-To: <CAJKOXPd2F6iy0ZqYf+X2k=eQ=tY1zG4gVbrr68XqE9+w4HK6dw@mail.gmail.com>

On 5/29/19 11:46 AM, Krzysztof Kozlowski wrote:
> On Mon, 27 May 2019 at 04:46, Matheus Castello <matheus@castello.eng.br> wrote:
>>
>> For configuration of fuel gauge alert for a low level state of charge
>> interrupt we add a function to config level threshold and a device tree
>> binding property to set it in flatned device tree node.
>>
>> Now we can use "maxim,alert-low-soc-level" property with the values from
>> 1% up to 32% to configure alert interrupt threshold.
>>
>> Signed-off-by: Matheus Castello <matheus@castello.eng.br>
>> ---
>>   drivers/power/supply/max17040_battery.c | 52 +++++++++++++++++++++++--
>>   1 file changed, 49 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/power/supply/max17040_battery.c b/drivers/power/supply/max17040_battery.c
>> index b7433e9ca7c2..2f4851608cfe 100644
>> --- a/drivers/power/supply/max17040_battery.c
>> +++ b/drivers/power/supply/max17040_battery.c
>> @@ -29,6 +29,9 @@
>>   #define MAX17040_DELAY         1000
>>   #define MAX17040_BATTERY_FULL  95
>>
>> +#define MAX17040_ATHD_MASK             0xFFC0
>> +#define MAX17040_ATHD_DEFAULT_POWER_UP 4
>> +
>>   struct max17040_chip {
>>          struct i2c_client               *client;
>>          struct delayed_work             work;
>> @@ -43,6 +46,8 @@ struct max17040_chip {
>>          int soc;
>>          /* State Of Charge */
>>          int status;
>> +       /* Low alert threshold from 32% to 1% of the State of Charge */
>> +       u32 low_soc_alert_threshold;
>>   };
>>
>>   static int max17040_get_property(struct power_supply *psy,
>> @@ -99,6 +104,28 @@ static void max17040_reset(struct i2c_client *client)
>>          max17040_write_reg(client, MAX17040_CMD, 0x0054);
>>   }
>>
>> +static int max17040_set_low_soc_threshold_alert(struct i2c_client *client,
>> +       u32 level)
>> +{
>> +       int ret;
>> +       u16 data;
>> +
>> +       /* check if level is between 1% and 32% */
>> +       if (level > 0 && level < 33) {
>> +               level = 32 - level;
>> +               data = max17040_read_reg(client, MAX17040_RCOMP);
>> +               /* clear the alrt bit and set LSb 5 bits */
>> +               data &= MAX17040_ATHD_MASK;
>> +               data |= level;
>> +               max17040_write_reg(client, MAX17040_RCOMP, data);
>> +               ret = 0;

I will put the return of max17040_write_reg on ret, instead of ret = 0.

>> +       } else {
>> +               ret = -EINVAL;
>> +       }
> 
> This is unusual way of handling error... when you parse DTS, you
> accept any value for "level" (even incorrect one). You validate the
> value later when setting it and show an error... however you ignore
> the error of max17040_write_reg() here... This is correct but looks
> unusual.
> 

Ok, so would it be better to check the level value in 
"max17040_get_of_data" and return an error there if the input is wrong?

Best Regards,
Matheus Castello

> Best regards,
> Krzysztof
> 