Re: [PATCH] mfd/mc13783: near complete rewrite

["Uwe Kleine-König" <u.kleine-koenig@pengutronix.de>]

Hello Mark,

On Mon, Nov 02, 2009 at 11:51:01AM +0000, Mark Brown wrote:
> On Fri, Oct 23, 2009 at 10:38:43PM +0200, Uwe Kleine-K??nig wrote:
> 
> > -int mc13783_register_irq(struct mc13783 *mc13783, int irq,
> > -		void (*handler) (int, void *), void *data)
> > +int mc13783_irq_request_nounmask(struct mc13783 *mc13783, unsigned int irq,
> > +		irqreturn_t (*handler)(struct mc13783 *, unsigned int, void *),
> > +		const char *name, void *dev)
> 
> If you're changing the signature of the IRQ handler functions it'd be
> nice to change them to irq_handler_t - that way it'll be much easier to
> transition the driver to using genirq in future since it should end up
> being possible to just stub out the Atlas-specific calls in the header
> with calls to the standard IRQ functions when the core is transitioned,
> reducing cross-tree issues.
OK, will do.
 
> This might create issues with a request_nounmask() function, though a
> request plus mask is probably enough - I guess you're using this for
> your the RTC driver in which case a spurious periodic interrupt is
> unlikely to be an issue.
Yes, I use it to simplify things in the RTC driver a bit.  I register
the 1HZ irq unconditionally and only track if it's masked or not.  This way
I can save some case discrimination.

Best regards and thaks for your comments
Uwe

-- 
Pengutronix e.K.                              | Uwe Kleine-König            |
Industrial Linux Solutions                    | http://www.pengutronix.de/  |