Re: [PATCH v2] backlight: lp855x: use private data for regulator control

[Sean Paul]

On Fri, Jul 10, 2015 at 4:43 PM, Kim, Milo <milo.kim@xxxxxx> wrote:
> Hi Paul,
>
>
> On 7/11/2015 12:01 AM, Sean Paul wrote:
>>
>> On Fri, Jul 10, 2015 at 4:26 AM, Milo Kim <milo.kim@xxxxxx> wrote:
>>>
>>> LP855x backlight device can be enabled by external VDD input.
>>> The 'supply' data is used for this purpose.
>>> It's kind of private data which runs internally, so there is no reason to
>>> expose to the platform data.
>>>
>>> And devm_regulator_get() is moved from _parse_dt() to _probe().
>>> Regulator consumer(lp855x) can control regulator not only from DT but
>>> also
>>> from platform data configuration in a source file such like board-*.c.
>>>
>>> If 'power' regulator driver is not ready, lp855x should continue to work
>>> because the power supply control is optional. So -EPROBE_DEFER return
>>> code
>>> is removed.
>>>
>>> v1->v2:
>>>    Keeps optional property '<name>-supply' in LP855x DT binding.
>>>
>>> Cc: Sean Paul <seanpaul@xxxxxxxxxxxx>
>>> Cc: Jingoo Han <jingoohan1@xxxxxxxxx>
>>> Cc: Lee Jones <lee.jones@xxxxxxxxxx>
>>> Cc: linux-kernel@xxxxxxxxxxxxxxx
>>> Signed-off-by: Milo Kim <milo.kim@xxxxxx>
>>> ---
>>>   drivers/video/backlight/lp855x_bl.c  | 20 +++++++++-----------
>>>   include/linux/platform_data/lp855x.h |  2 --
>>>   2 files changed, 9 insertions(+), 13 deletions(-)
>>>
>>> diff --git a/drivers/video/backlight/lp855x_bl.c
>>> b/drivers/video/backlight/lp855x_bl.c
>>> index a26d3bb..277d5ca 100644
>>> --- a/drivers/video/backlight/lp855x_bl.c
>>> +++ b/drivers/video/backlight/lp855x_bl.c
>>> @@ -73,6 +73,7 @@ struct lp855x {
>>>          struct device *dev;
>>>          struct lp855x_platform_data *pdata;
>>>          struct pwm_device *pwm;
>>> +       struct regulator *supply;       /* regulator for VDD input */
>>>   };
>>>
>>>   static int lp855x_write_byte(struct lp855x *lp, u8 reg, u8 data)
>>> @@ -384,13 +385,6 @@ static int lp855x_parse_dt(struct lp855x *lp)
>>>                  pdata->rom_data = &rom[0];
>>>          }
>>>
>>> -       pdata->supply = devm_regulator_get(dev, "power");
>>> -       if (IS_ERR(pdata->supply)) {
>>> -               if (PTR_ERR(pdata->supply) == -EPROBE_DEFER)
>>> -                       return -EPROBE_DEFER;
>>> -               pdata->supply = NULL;
>>> -       }
>>> -
>>>          lp->pdata = pdata;
>>>
>>>          return 0;
>>> @@ -431,8 +425,12 @@ static int lp855x_probe(struct i2c_client *cl, const
>>> struct i2c_device_id *id)
>>>          else
>>>                  lp->mode = REGISTER_BASED;
>>>
>>> -       if (lp->pdata->supply) {
>>> -               ret = regulator_enable(lp->pdata->supply);
>>> +       lp->supply = devm_regulator_get(lp->dev, "power");
>>> +       if (IS_ERR(lp->supply))
>>> +               lp->supply = NULL;
>>
>>
>>
>> Hi Milo,
>> You removed the probe deferral handling on the regulator and broke
>> probe deferral in cases where the regulator isn't ready.
>
>
> This power supply is optional. Even if lp855x can not get regulator driver,
> it should work. (And I saw same comment in the DT. The 'power-supply'
> property is optional). So -EPORBE_DEFER is not necessary in _probe().
>

I respectfully disagree. devm_regulator_get can return EPROBE_DEFER if
the regulator is valid (and specified in the dt), but not ready to be
used yet. In this case, your patch will assume it doesn't exist and
will never use it. This Is Bad.

Sean

> Best regards,
> Milo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/