Hi all,

On Thu, Jul 19, 2018 at 04:10:20PM +0300, sakari.ailus@iki.fi wrote:
> On Thu, Jul 19, 2018 at 03:14:06PM +0300, Laurent Pinchart wrote:
> > On Thursday, 19 July 2018 11:42:08 EEST Wolfram Sang wrote:
> > > > > -static int ov772x_mask_set(struct i2c_client *client, u8  command, u8
> > > > > mask,
> > > > > -			   u8  set)
> > > > > -{
> > > > > -	s32 val = ov772x_read(client, command);
> > > > > -
> > > > > -	if (val < 0)
> > > > > -		return val;
> > > > > -
> > > > > -	val &= ~mask;
> > > > > -	val |= set & mask;
> > > > > -
> > > > > -	return ov772x_write(client, command, val);
> > > > > -}
> > > > > -
> > > >
> > > > If I were you I would have kept these functions and wrapped the regmap
> > > > operations there. This is not an issue though if you prefer it this
> > > > way :)
> > >
> > > I have suggested this way. It is not a show stopper issue, but I still
> > > like this version better.
> >
> > Wrapping the regmap functions minimizes the diff and makes it easier to
> > backport the driver.

This was my reasoning too, but I'm happy with the current
implementation. Thanks Akinobu for handling this!

>
> May be, but using the regmap functions directly makes the driver cleaner.
> Most drivers have some kind of wrappers around the I²C framework (or
> regmap) functions; this one is one of the few to get rid of them.
>
> The two could be done in a separate patch, too, albeit I think the current
> one seems fine as such.
>
> --
> Sakari Ailus
> e-mail: sakari.ailus@iki.fi