Hi André,
thanks for your patch!
mostly looks fine, given the MFD design is accepted.
Nitpicks below:
On Wed, Feb 26, 2025 at 6:51 PM André Draszik <andre.draszik@xxxxxxxxxx> wrote:
> +static irqreturn_t max77759_gpio_irqhandler(int irq, void *data)
> +{
> + int handled = 0;
bool handled = false;
(...)
> + for_each_set_bit(offset, &pending, MAX77759_N_GPIOS) {
> + unsigned int virq;
I usually just call this "irq", as it's not any more virtual than any other
Linux magic number, and it can confuse people working with
actual virtualization when we call things virtual like this.
> +
> + /*
> + * ACK interrupt by writing 1 to bit 'offset', all
> + * others need to be written as 0. This needs to be
> + * done unconditionally hence regmap_set_bits() is
> + * inappropriate here.
> + */
> + regmap_write(chip->map, MAX77759_MAXQ_REG_UIC_INT1,
> + BIT(offset));
> +
> + virq = irq_find_mapping(gc->irq.domain, offset);
> + handle_nested_irq(virq);
> +
> + handled = 1;
handled = true;
> + chip->gc.base = -1;
> + chip->gc.label = dev_name(&pdev->dev);
> + chip->gc.parent = &pdev->dev;
> + chip->gc.owner = THIS_MODULE;
The core already assigns THIS_MODULE so you can leave it
unassigned.
Yours,
Linus Walleij
