Re: [PATCH v3] PCI: imx6: Replace legacy gpio interface for gpiod interface

[Lucas Stach]

Hi Linus,

Am Donnerstag, dem 21.04.2022 um 01:24 +0200 schrieb Linus Walleij:
> Hi Maira and sorry for being slow on reviews.
> 
> On Tue, Nov 2, 2021 at 2:04 AM Maíra Canal <maira.canal@xxxxxx> wrote:
> 
> > -               gpio_set_value_cansleep(imx6_pcie->reset_gpio,
> > +               gpiod_set_raw_value_cansleep(imx6_pcie->reset_gpio,
> >                                         !imx6_pcie->gpio_active_high);
> 
> Hm I see you got advised to use the raw api. I'm not so sure about
> that I like v1 better.
> 
> > +       imx6_pcie->reset_gpio = devm_gpiod_get_optional(dev, "reset",
> > +                       imx6_pcie->gpio_active_high ?  GPIOD_OUT_HIGH : GPIOD_OUT_LOW);
> > +       if (IS_ERR(imx6_pcie->reset_gpio))
> > +               return dev_err_probe(dev, PTR_ERR(imx6_pcie->reset_gpio),
> > +                               "unable to get reset gpio\n");
> 
> Where is this descriptor coming from? Device trees? Can't we just fix the
> DTS file(s) in that case given how wrong they are if they don't set
> GPIO_ACTIVE_LOW flag on this IRQ.

The binding explicitly describes the GPIO as not polarity aware and has
a separate property "reset-gpio-active-high" to avoid breaking old
DTBs. I don't think it's helpful to dismiss this explicit backward
compat just because the driver code looks nicer that way.

Regards,
Lucas