backlight/ld9040.c: regulator control in the lcd driver

[Linus Walleij]

On Fri, Dec 2, 2011 at 9:57 AM, Kyungmin Park <kyungmin.park at samsung.com> wrote:
>>[leedonghwa]
> [Me]
>>> +         lcd->reg_vdd3 = regulator_get(lcd->dev, "vdd");
>>> +         if (IS_ERR(lcd->reg_vdd3)) {
>>> +                   dev_info(lcd->dev, "no %s regulator found\n", "vdd");
>>> +                   lcd->reg_vdd3 = NULL;
>>> +         }
>>> +
>>> +         lcd->reg_vci = regulator_get(lcd->dev, "vci");
>>> +         if (IS_ERR(lcd->reg_vci)) {
>>> +                   dev_info(lcd->dev, "no %s regulator found\n", "vci");
>>> +                   lcd->reg_vci = NULL;
>>> +         }
>>
>> As explained in earlier discussion with Mark regarding the SMSC911x
>> driver regulator, treat these as errors and do not fail
>> "gracefully" like this.
>>
>> Reference:
>> http://marc.info/?l=linux-netdev&m=131914562120725&w=2
>
> As mentioned at commit message, the lcd regulator is optional part and
> refer the mmc codes
>
>        host->vmmc = regulator_get(mmc_dev(mmc), "vmmc");
>        if (IS_ERR(host->vmmc)) {
>                pr_info("%s: no vmmc regulator found\n", mmc_hostname(mmc));
>                host->vmmc = NULL;
>        } else {
>                regulator_enable(host->vmmc);
>        }
>
> Previous time, these codes are located at board file, but more boards
> are used, it has same codes for all boards. so move it to drivers.

I know. This was brought up in the aforementioned discussion,
but the above is also wrong, simply. See:
http://marc.info/?l=linux-netdev&m=131914562120667&w=2
http://marc.info/?l=linux-netdev&m=131914562120690&w=2
http://marc.info/?l=linux-netdev&m=131914562120725&w=2
http://marc.info/?l=linux-netdev&m=131963332527416&w=2

> In our case, it has the regulator but some boards don't.
>
> Umm then how to handle the regulator gracefully?

Mark suggest using a fixed-voltage regulator for boards
where the power is always on. The voltage level itself
is optional. See:
http://marc.info/?l=linux-netdev&m=131963332527416&w=2

Other approaches is to use dummy regulators, or not
call regulator_has_full_constraints(), which means the
regulator core will provide dummy regulators anyways.
See:
http://marc.info/?l=linux-netdev&m=131973043527112&w=2
http://marc.info/?l=linux-netdev&m=131975178703166&w=2

Whole thread of discussion:
http://marc.info/?l=linux-netdev&w=2&r=1&s=smsc911x&q=b

Yours,
Linus Walleij