Re: [PATCH v2 9/9] media: i2c: tw9910: Remove soc_camera dependencies

[jacopo mondi]

Hi Fabio,

On Wed, Jan 03, 2018 at 04:14:35PM -0200, Fabio Estevam wrote:
> On Wed, Jan 3, 2018 at 3:37 PM, jacopo mondi <jacopo@xxxxxxxxxx> wrote:
>
> >> Initially the rest GPIO was doing:
> >>
> >> -       gpio_set_value(GPIO_PTT3, 0);
> >> -       mdelay(10);
> >> -       gpio_set_value(GPIO_PTT3, 1);
> >> -       mdelay(10); /* wait to let chip come out of reset */
> >
> > And that's what my driver code does :)
>
> No, on 5/9 you converted the original code to:
>
> GPIO_LOOKUP("sh7722_pfc", GPIO_PTT3, "rstb", GPIO_ACTIVE_HIGH)
>
> It should be GPIO_ACTIVE_LOW instead.
>
> > My point is that if I read the manual and I see an active low gpio (0
> > is reset state) then the driver code uses it as and active high one (1
> > is the reset state), that would be weird to me.
>
> Then on this patch you should do:
>
> gpiod_set_value(priv->rstb_gpio, 1);  ---> This tells the GPIO to go
> to its active state (In this case active == logic level 0)
> usleep_range(500, 1000);
> gpiod_set_value(priv->rstb_gpio, 0); ---> This tells the GPIO to go to
> its inactive state (In this case inactive == logic level 1)
>
> You can also look at Documentation/gpio/consumer.txt where the usage
> of the gpiod_xxx API is described.
>
> It seems you are confusing it with the legacy gpio_set_value() API
> (Documentation/gpio/gpio-legacy.txt)

It took you 3 email messages, but maybe I finally got it.

So, 1 and 0 do not actually represent the line level but the active
or inactive states, that's fine. This seems to me a bit inconsistent with
the existence of flags like GPIOD_OUT_HIGH/LOW meant to be used at gpiod_get()
time, where the actual line level has to be used instead, but that's a
discussion surely not pertinent to this series.

Thanks for your patience.
    j


>
> Hope this helps.