Re: [PATCH V7 1/1] drivers/gpio: Altera soft IP GPIO driver and devicetree binding

[Grant Likely <grant.likely@linaro.org>]

On Mon, 7 Apr 2014 12:11:15 -0500, Josh Cartwright <joshc@codeaurora.org> wrote:
> On Mon, Apr 07, 2014 at 04:00:43PM +0800, Tien Hock Loh wrote:
> > On Wed, Mar 19, 2014 at 6:09 PM, Tien Hock Loh <thloh@altera.com> wrote:
> > > On Fri, Mar 7, 2014 at 11:14 PM, Josh Cartwright <joshc@codeaurora.org> wrote:
> > >> On Mon, Mar 03, 2014 at 06:27:43PM +0800, thloh@altera.com wrote:
> > >>> From: Tien Hock Loh <thloh@altera.com>
> [..]
> > >>> +     altera_gc->mapped_irq = irq_of_parse_and_map(node, 0);
> > >>>
> > >>
> > >> platform_get_irq(pdev, 0);
> > >>
> > > OK.
> > >
> >
> > platform_get_irq doesn't create the irq mapping which is needed by the
> > driver. Since this driver is targeted at using of, should I be using
> > irq_of_parse_and_map or should I still redo the codes with
> > platform_get_irq and irq_create_mapping? I think the latter would be
> > introducing code redundancy. Please advice.
> 
> Yes, it is technically true that platform_get_irq() doesn't do the
> mapping directly, but that's because the mapping is setup earlier, when
> of_device_alloc() (drivers/of/platform.c) allocates resources for your
> platform device.
> 
> Calling irq_of_parse_and_map() should be unnecessary.

In the latest mainline, platform_get_irq() now does the mapping when
using an IRQ specified in the device tree.

g.