MIME-Version: 1.0
References: <20230313113308.157930-1-francesco@dolcini.it> <20230313113308.157930-3-francesco@dolcini.it>
In-Reply-To: <20230313113308.157930-3-francesco@dolcini.it>
From:   Andy Shevchenko <andy.shevchenko@gmail.com>
Date:   Mon, 13 Mar 2023 14:24:47 +0200
Message-ID: <CAHp75VePJ76nhwX-1cyo-uD=2+w4dbyGNhazJzKkkQOfDma6aQ@mail.gmail.com>
Subject: Re: [PATCH v2 2/2] gpio: fxl6408: add I2C GPIO expander driver
To:     Francesco Dolcini <francesco@dolcini.it>
Cc:     linux-gpio@vger.kernel.org,
        Linus Walleij <linus.walleij@linaro.org>,
        Bartosz Golaszewski <brgl@bgdev.pl>,
        Emanuele Ghidoli <emanuele.ghidoli@toradex.com>,
        linux-kernel@vger.kernel.org,
        Francesco Dolcini <francesco.dolcini@toradex.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Mon, Mar 13, 2023 at 1:33=E2=80=AFPM Francesco Dolcini <francesco@dolcin=
i.it> wrote:
>
> From: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
>
> Add minimal driver for Fairchild FXL6408 8-bit I2C-controlled GPIO expand=
er
> using the generic regmap based GPIO driver (GPIO_REGMAP).
>
> The driver implements setting the GPIO direction, reading inputs
> and writing outputs.
>
> In addition to that the FXL6408 has the following functionalities:
> - allows to monitor input ports for data transitions with an interrupt pi=
n
> - all inputs can be configured with pull-up or pull-down resistors

Thank you for the update, my comments below.

...

+ Co-developed-by: your name + email ?

> Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>

...

> +config GPIO_FXL6408
> +       tristate "FXL6408 I2C GPIO expander"
> +       select GPIO_REGMAP

> +       select REGMAP_I2C

Somebody pointed out that this might require

    depends on I2C

being added as well.

...

> +#include <linux/gpio/regmap.h>
> +#include <linux/i2c.h>
> +#include <linux/module.h>
> +#include <linux/regmap.h>

Missing kernel.h for ARRAY_SIZE(), but I prefer the array_size.h to
appear (nevertheless, the latter is out of scope of this one).
Missing err.h for error handling macros.

...

> +#define FXL6408_MAX_REGISTER           0x13

This is used as a range, but why? If we can have a proper name for
this register, why bother dumping all this or even having access to?

...

> +static int fxl6408_identify(struct device *dev, struct regmap *regmap)
> +{
> +       int val, ret;
> +
> +       ret =3D regmap_read(regmap, FXL6408_REG_DEVICE_ID, &val);
> +       if (ret) {
> +               dev_err(dev, "error %d reading DEVICE_ID\n", ret);
> +       } else if (val >> FXL6408_MF_SHIFT !=3D FXL6408_MF_FAIRCHILD) {
> +               dev_err(dev, "invalid device id 0x%02x\n", val);
> +               ret =3D -ENODEV;
> +       }
> +
> +       return ret;

This function is only used at ->probe(), you may refactor it like

       ret =3D regmap_read(regmap, FXL6408_REG_DEVICE_ID, &val);
       if (ret)
               return dev_err_probe(dev, ret, "error reading DEVICE_ID\n");
       if (val >> FXL6408_MF_SHIFT !=3D FXL6408_MF_FAIRCHILD)
               return dev_err_probe(dev, -ENODEV, "invalid device id
0x%02x\n", val);

       return 0;

> +}

--=20
With Best Regards,
Andy Shevchenko