Re: [PATCH v6 10/12] input: keyboard: Add support for MAX7360 keypad

["Mathieu Dubois-Briand" <mathieu.dubois-briand@bootlin.com>]

On Wed Apr 9, 2025 at 8:40 PM CEST, Dmitry Torokhov wrote:
> Hi Mathieu,
>
> On Wed, Apr 09, 2025 at 04:55:57PM +0200, Mathieu Dubois-Briand wrote:
> ...
>> +static irqreturn_t max7360_keypad_irq(int irq, void *data)
>> +{
>> +	struct max7360_keypad *max7360_keypad = data;
>> +	unsigned int val;
>> +	unsigned int row, col;
>> +	unsigned int release;
>> +	unsigned int code;
>> +	int ret;
>
> int error;
>

Ok using error on all similar cases.

>> +
>> +	do {
>> +		ret = regmap_read(max7360_keypad->regmap, MAX7360_REG_KEYFIFO, &val);
>> +		if (ret) {
>> +			dev_err(&max7360_keypad->input->dev, "Failed to read max7360 FIFO");
>
> This will return name pf the input device, whereas logging name of the
> platform device (representing the hardware device) would be much more
> interesting. You can either use max7360_keypad->input->dev.parent, or,
> better yet, add *dev pointer to struct max7360_keypad.
>

Makes sense, thanks.

> ...
>
>> +static int max7360_keypad_build_keymap(struct max7360_keypad *max7360_keypad)
>> +{
>> +	struct input_dev *input_dev = max7360_keypad->input;
>> +	struct device *dev = input_dev->dev.parent->parent;
>> +	struct matrix_keymap_data keymap_data;
>> +	const char *propname = "linux,keymap";
>> +	unsigned int max_keys;
>> +	int size;
>> +	int ret;
>> +
>> +	size = device_property_count_u32(dev, propname);
>> +	if (size <= 0) {
>> +		dev_err(dev, "missing or malformed property %s: %d\n", propname, size);
>> +		return size < 0 ? size : -EINVAL;
>> +	}
>> +
>> +	max_keys = max7360_keypad->cols * max7360_keypad->rows;
>> +	if (size > max_keys) {
>> +		dev_err(dev, "%s size overflow (%d vs max %u)\n", propname, size, max_keys);
>> +		return -EINVAL;
>> +	}
>> +
>> +	u32 *keys __free(kfree) = kmalloc_array(size, sizeof(*keys), GFP_KERNEL);
>> +	if (!keys)
>> +		return -ENOMEM;
>> +
>> +	ret = device_property_read_u32_array(dev, propname, keys, size);
>> +	if (ret) {
>> +		dev_err(dev, "failed to read %s property: %d\n", propname, ret);
>> +		return ret;
>> +	}
>> +
>> +	keymap_data.keymap = keys;
>> +	keymap_data.keymap_size = size;
>> +	ret = matrix_keypad_build_keymap(&keymap_data, NULL, max7360_keypad->rows, max7360_keypad->cols,
>> +					 max7360_keypad->keycodes, max7360_keypad->input);
>
> What if it fails? Error handling please.

Yes, forgot to return ret just below. Not adding logs as in most cases
matrix_keypad_build_keymap() will already print some logs. OK with that?

>
> Also, it looks like you are repeating what matrix_keypad_build_keymap()
> is already doing. If you pass NULL as keymap data, won't it do the right
> thing?
>

No, because matrix_keypad_parse_keymap() is using input_dev->dev.parent
and this device will not have any associated device tree node. It should
use input_dev->dev.parent->parent to get correct values.

There is a discussion ongoing about using device_set_of_node_from_dev(),
so the MFD child device reuse the node of the parent. But I tried to
avoid using it here, as I was able to come with another solution.
Discussion is in the thread of the pinctrl driver (patch #3).

Another solution would be to modify matrix_keypad_parse_keymap(),
allowing to pass a custom device. But again, I tried to avoid doing this
modification just for my own need.

> ...
>
> Thanks.

Ok with all other comments.

Thanks for your review.
Mathieu

-- 
Mathieu Dubois-Briand, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com