[PATCH v3] gpio: UniPhier: add driver for UniPhier GPIO controller

[Masahiro Yamada]

Hi Linus,



2015-07-16 16:07 GMT+09:00 Linus Walleij <linus.walleij at linaro.org>:
> On Thu, Jul 16, 2015 at 5:44 AM, Masahiro Yamada
> <yamada.masahiro at socionext.com> wrote:
>> 2015-07-15 23:15 GMT+09:00 Linus Walleij <linus.walleij at linaro.org>:
>
>>>> +       for (i = 0; i < chip->ngpio; i += UNIPHIER_GPIO_PORTS_PER_BANK) {
>>>> +               bank = i / UNIPHIER_GPIO_PORTS_PER_BANK;
>>>> +               shift = i % BITS_PER_LONG;
>>>> +               bank_mask = (mask[BIT_WORD(i)] >> shift) &
>>>> +                                               UNIPHIER_GPIO_BANK_MASK;
>>>> +               bank_bits = bits[BIT_WORD(i)] >> shift;
>>>> +
>>>> +               uniphier_gpio_bank_write(chip, bank, UNIPHIER_GPIO_REG_DATA,
>>>> +                                        bank_mask, bank_bits);
>>>> +       }
>>>
>>> This looks like a piece of algorithm that we could make generic like in a
>>> static function in drivers/gpio/gpiolib.h or so, that it may be shared with
>>> other drivers. Do you see some clear way to break out the core of this?
>>> Or is it as generic as I think?
>>
>> I assume this comment has no intention to block my patch.
>
> Nah. Just thinking the code looks neat.
>
>>>> +       ret = of_property_read_u32(dev->of_node, "ngpio", &ngpio);
>>>> +       if (ret) {
>>>> +               dev_err(dev, "failed to get ngpio property\n");
>>>> +               return ret;
>>>> +       }
>>>
>>> This needs to be documented, plus I don't see why it's needed.
>>> The driver for this very specific hardware should already know
>>> how many GPIOs there are in this hardware, it should not come
>>> from the device tree.
>>
>> I want to use this driver on various SoCs, but
>> the number of GPIO pins varies by SoC.
>>
>> ngpio == 248 for some SoCs,
>> and ngpio == 136 for some, etc.
>
> That is the wrong way to handle different SoC. That should be handled
> by different compatible strings, and then you select the number of GPIOs
> for the version corresponding to that compatibe string.
>
>>>> +static const struct of_device_id uniphier_gpio_match[] = {
>>>> +       { .compatible = "socionext,uniphier-gpio" },
>>>> +       { /* sentinel */ }
>>>> +};
>
> i.e. you should use the .data field of of_device_id to carry variant-specific
> information.


Currently, I want to use this driver on 7 SoCs

PH1-sLD3:    ngpio == 136
PH1-LD4 :    ngpio == 136
PH1-Pro4:    ngpio == 248
PH1-sLD8:    ngpio == 136
PH1-Pro5:    ngpio == 248
ProXstream2: ngpio == 232
PH1-LD6b:    ngpio == 232

So, should I describe the OF match table like this?

static const struct of_device_id uniphier_gpio_match[] = {
       { .compatible = "socionext,ph1-sld3-gpio"       .data = (void *)136 },
       { .compatible = "socionext,ph1-ld4-gpio"        .data = (void *)136 },
       { .compatible = "socionext,ph1-pro4-gpio"       .data = (void *)248 },
       { .compatible = "socionext,ph1-sld8-gpio"       .data = (void *)136 },
       { .compatible = "socionext,ph1-pro5-gpio"       .data = (void *)248 },
       { .compatible = "socionext,proxstream2-gpio",   .data = (void *)232 },
       { .compatible = "socionext,ph1-ld6b-gpio",      .data = (void *)232 },
       { /* sentinel */ }
};


One disadvantage for this way is that
I need to touch the driver file every time I add a new SoC support.

If I could support "ngpio" property, I would only have to add a device
tree for a new SoC.




-- 
Best Regards
Masahiro Yamada