Re: [PATCH v3 04/15] iio: accel: adxl345: use regmap cache for INT mapping

[Jonathan Cameron <jic23@kernel.org>]

On Sun, 2 Mar 2025 11:45:03 +0000
Jonathan Cameron <jic23@kernel.org> wrote:

> On Thu, 20 Feb 2025 10:42:23 +0000
> Lothar Rubusch <l.rubusch@gmail.com> wrote:
> 
> > Use regmap cache to replace maintaining the member variable intio
> > for the interrupt mapping state. The interrupt mapping is initialized
> > when the driver is probed, and it is perfectly cacheable.
> > 
> > The patch will still leave the function set_interrupts(). A follow up
> > patch takes care of it, when cleaning up the INT enable register
> > variable.
> > 
> > Signed-off-by: Lothar Rubusch <l.rubusch@gmail.com>
> > ---
> >  drivers/iio/accel/adxl345.h      |  4 ++
> >  drivers/iio/accel/adxl345_core.c | 63 ++++++++++++++++++++------------
> >  drivers/iio/accel/adxl345_i2c.c  |  2 +
> >  drivers/iio/accel/adxl345_spi.c  |  2 +
> >  4 files changed, 48 insertions(+), 23 deletions(-)
> > 
> > diff --git a/drivers/iio/accel/adxl345.h b/drivers/iio/accel/adxl345.h
> > index bc6d634bd85c..a2a81caa292a 100644
> > --- a/drivers/iio/accel/adxl345.h
> > +++ b/drivers/iio/accel/adxl345.h
> > @@ -8,6 +8,8 @@
> >  #ifndef _ADXL345_H_
> >  #define _ADXL345_H_
> >  
> > +#include <linux/regmap.h>  
> 
> Why add this include?
> 
> The file should have a forwards def of
> struct regmap;
> which is currently missing.  If you clean that up in this patch that
> is fine (mention it in the patch description though as it isn't
> directly related) but I don't see a reason to include regmap.h here.
> 
> Given rest if fine I'll tweak this whilst applying. Applied to the
> togreg branch of iio.git, pushed out for now as testing for 0-day
> to poke at it.
> 
> Also move to a newer kernel tree. The changes in export symbol
> should be causing you build errors for this path. I'll fix that up.
> Quotes now needed around IIO_ADXL345 in the EXPORT_SYMBOL_NS_GPL()
> calls. I fixed that up.
Dropped again after reviewing later patch.  I think the volatile
stuff should specify all registers that are volatile from the start
not add them as we go.
> 
> Jonathan