Re: [PATCH v2] mfd: lubbock_io: add lubbock_io board

[Robert Jarzmik <robert.jarzmik@free.fr>]

Mark Rutland <mark.rutland@arm.com> writes:

> Hi Robert,
>
> Given the addition of an of_device_id table and some (implicit) property
> parsing, this requires a device tree binding document.
Yes, you're right. I'll prepare that.

>> +	ret = devm_request_irq(&pdev->dev, cot->irq, lubbock_irq_handler, 0,
>> +			       dev_name(&pdev->dev), cot);
>> +	if (ret == -ENOSYS)
>> +		return -EPROBE_DEFER;
>> +	if (ret) {
>> +		dev_err(&pdev->dev, "Couldn't request GPIO : ret = %d\n", ret);
>> +		return ret;
>> +	}
>> +	irq_set_irq_type(cot->irq, IRQ_TYPE_EDGE_FALLING);
>
> Shouldn't that be in the interrupt-specifier when using DT?
Maybe. But what about the platform data case ? I think I wanted to express the
multiple-NOR gate binding all incoming level interrupt into the output cot->irq
interrupt line going to GPIO0.

Do you have an idea for the platform data case ?
>> +static const struct of_device_id lubbock_id_table[] = {
>> +	{ .compatible = "marvell,lubbock_io", },

> When PXA25x it was Intel, not Marvell. So I think the vendor prefix
> should be "intel".
>
> Also s/_/-/ in property names and compatible strings please.
Ok for v3.

Thanks for the review.

-- 
Robert