Re: [PATCH v20 2/3] Input: mt6779-keypad - Add MediaTek keypad driver

From: Mattijs Korpershoek <mkorpershoek@baylibre.com>
To: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>,
	Marco Felsch <m.felsch@pengutronix.de>,
	Rob Herring <robh+dt@kernel.org>,
	Matthias Brugger <matthias.bgg@gmail.com>,
	Fengping Yu <fengping.yu@mediatek.com>,
	Yingjoe Chen <yingjoe.chen@mediatek.com>,
	Fabien Parent <fparent@baylibre.com>,
	Kevin Hilman <khilman@baylibre.com>,
	linux-input@vger.kernel.org, devicetree@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org,
	linux-mediatek@lists.infradead.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v20 2/3] Input: mt6779-keypad - Add MediaTek keypad driver
Date: Fri, 28 Jan 2022 11:03:08 +0100	[thread overview]
Message-ID: <87v8y4p483.fsf@baylibre.com> (raw)
In-Reply-To: <YfK4UcuCfF7JfI7H@smile.fi.intel.com>

On Thu, Jan 27, 2022 at 17:20, Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote:

> On Thu, Jan 27, 2022 at 12:15:25PM +0100, Mattijs Korpershoek wrote:
>> From: "fengping.yu" <fengping.yu@mediatek.com>
>> 
>> This patch adds matrix keypad support for Mediatek SoCs.
>
> Some comments which may be addressed now or in the follow-up patch(es).
> Up to you.
Hi Andy,
Thank you for your review and your suggestions.

>
> ...
>
>> +static const struct regmap_config mt6779_keypad_regmap_cfg = {
>> +	.reg_bits = 32,
>> +	.val_bits = 32,
>
>> +	.reg_stride = sizeof(u32),
>
> I'm wondering if we need this when we have reg_bits = 32 already.

Per my understanding, .reg_stride is mainly used to check for invalid register
addresses in regmap_{read,write}():

    if (!IS_ALIGNED(reg, map->reg_stride))
            return -EINVAL;

If .reg_stride is not set, regmap core will default it to 1.
It's not computed from reg_bits.

So I think we still need it.
>
>> +	.max_register = 36,
>> +};
>
> ...
>
>> +	regmap_write(keypad->regmap, MTK_KPD_DEBOUNCE,
>> +		     (debounce * 32) & MTK_KPD_DEBOUNCE_MASK);
>
> I'm wondering if << 5 is more specific to show that the value
> is based on 2^5 units.

The datasheet I've seen states: "De-bounce time = KP_DEBOUNCE / 32ms"
But rewriting it as 1 << 5 seems reasonable as well:
regmap_write(keypad->regmap, MTK_KPD_DEBOUNCE,
            (debounce * (1 << 5)) & MTK_KPD_DEBOUNCE_MASK);

I don't have any preference on this one.
If I have to send a v21, I will rewrite it using (1 << 5)

>
> ...
>
>> +	error = devm_add_action_or_reset(&pdev->dev, mt6779_keypad_clk_disable, keypad->clk);
>
> You have this long line...
>
>> +	error = devm_request_threaded_irq(&pdev->dev, irq,
>> +					  NULL, mt6779_keypad_irq_handler,
>> +					  IRQF_ONESHOT,
>> +					  MTK_KPD_NAME, keypad);
>
> ...at the same time you may reduce LOCs here...
Ack. will join lines to reduce LOCs if I have to send v21.
>
>> +	if (error) {
>> +		dev_err(&pdev->dev, "Failed to request IRQ#%d:%d\n",
>> +			irq, error);
>
> ...and here.
Ack. will join lines to reduce LOCs if I have to send v21.

>
>> +		return error;
>> +	}
>
> -- 
> With Best Regards,
> Andy Shevchenko