On Fri, Jul 10, 2015 at 08:03:50AM +0200, Alexandre Belloni wrote:
> > static const struct of_device_id at91_reset_of_match[] = {
> > { .compatible = "atmel,at91sam9260-rstc", .data = at91sam9260_restart },
> > { .compatible = "atmel,at91sam9g45-rstc", .data = at91sam9g45_restart },
> > + { .compatible = "atmel,sama5d3-rstc", .data = sama5d3_restart },
> > { /* sentinel */ }
> > };
> >
> > @@ -181,17 +189,21 @@ static int at91_reset_of_probe(struct platform_device
> > *pdev)
> > return -ENODEV;
> > }
> >
> > - for_each_matching_node(np, at91_ramc_of_match) {
> > - at91_ramc_base[idx] = of_iomap(np, 0);
> > - if (!at91_ramc_base[idx]) {
> > - dev_err(&pdev->dev, "Could not map ram controller
> > address\n");
> > - return -ENODEV;
> > + match = of_match_node(at91_reset_of_match, pdev->dev.of_node);
> > + at91_restart_nb.notifier_call = match->data;
> > +
> > + if (match->data != sama5d3_restart) {
>
> This doesn't scale well. I would create a structure with a pointer to
> the restart function and a boolean or a bitfield to store whether the
> workaround is needed. Use that structure in your match data. Then, you
> won't need to reorder anything.Maybe it simply doesn't need to scale (yet). You have a single exception here. Maybe you will have only this one in the future, maybe you won't, but for now, that solution looks a bit overkill. Maxime -- Maxime Ripard, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com
signature.asc
Description: Digital signature
