Re: [PATCH v2 3/4] iio: accel: adxl345: Setup DATA_READY trigger

[Dmitry Torokhov]

On Sun, May 14, 2017 at 04:15:37PM +0100, Jonathan Cameron wrote:
> On 10/05/17 14:24, Eva Rachel Retuya wrote:
> >On Wed, May 03, 2017 at 12:05:00AM +0300, Andy Shevchenko wrote:
> >>On Tue, May 2, 2017 at 3:15 PM, Eva Rachel Retuya <eraretuya@xxxxxxxxx> wrote:
> >>>On Mon, May 01, 2017 at 02:31:00PM +0300, Andy Shevchenko wrote:
> >>>[...]
> >>>>>-int adxl345_core_probe(struct device *dev, struct regmap *regmap,
> >>>>>+int adxl345_core_probe(struct device *dev, struct regmap *regmap, int irq,
> >>>>>                        const char *name);
> >>>>
> >>>>I think I commented this once. Instead of increasing parameters,
> >>>>please introduce a new struct (as separate preparatory patch) which
> >>>>will hold current parameters. Let's call it
> >>>>strut adxl345_chip {
> >>>>  struct device *dev;
> >>>>  struct regmap *regmap;
> >>>>  const char *name;
> >>>>};
> >>>>
> >>>>I insisnt in this chage.
> >>>
> >>>I'm not sure if what you want is more simpler, is it something like what
> >>>this driver does?
> >>
> >>Nope. The driver you were referring to does the same you did.
> >>
> >>I'm proposing the above struct to be introduced along with changing
> >>prototype like:
> >>
> >>  -int adxl345_core_probe(struct device *dev, struct regmap *regmap,
> >>const char *name);
> >>  +int adxl345_core_probe(struct adxl345_chip *chip);
> >>
> >>In next patch adding interrupt would not touch prototypes at all!
> >>
> >
> >OK, got it. Thanks for clarifying.
> >
> >>>
> >>>http://lxr.free-electrons.com/source/drivers/iio/gyro/mpu3050.h#L41
> >>>http://lxr.free-electrons.com/source/drivers/iio/gyro/mpu3050-i2c.c#L34
> >>
> >>>>>+#include <linux/of_irq.h>
> >>>>
> >>>>Can we get rid of gnostic resource providers?
> >>>>
> >>>
> >>>I'm uninformed and still learning. Please let me know how to approach
> >>>this in some other way.
> >>
> >>I suppose something like platform_get_irq(); to use.
> >>But it would be nice to you to investigate more.
> >
> >I had a look and it seems I have to convert to platform_driver in order
> >to make use of that function. Is this correct?
> I believe Andy was suggesting a function 'similar to' that one rather
> than actually using platform_get_irq.
> 
> It's not an area I know all that much about either, but there
> are moves to try and move the boiler plat needed to get the same
> parameters from devicetree and acpi into core library code so
> that a single function can be called to get the parameter in
> either case.

Well, we of course could look into moving bulk of platform_get_irq()
into generic device_get_irq(), but I think the main problem is that ACPI
(as far as I know) does not have notion of interrupt names...

Thanks.

-- 
Dmitry