Re: [PATCH v3 1/2] dt-bindings: at24: Add address-width property
From: Andy Shevchenko
Date: Wed Jun 27 2018 - 07:41:46 EST
On Wed, 2018-06-27 at 12:40 +0300, Sakari Ailus wrote:
> On Wed, Jun 27, 2018 at 10:19:38AM +0200, Bartosz Golaszewski wrote:
> > 2018-06-27 7:46 GMT+02:00 <alanx.chiang@xxxxxxxxx>:
> > > From: Alan Chiang <alanx.chiang@xxxxxxxxx>
> > >
> > > The AT24 series chips use 8-bit address by default. If some
> > > chips would like to support more than 8 bits, the at24 driver
> > > should be added the compatible field for specfic chips.
> > >
> > > Provide a flexible way to determine the addressing bits through
> > > address-width in this patch.
> > >
> > > Signed-off-by: Alan Chiang <alanx.chiang@xxxxxxxxx>
> > > Signed-off-by: Andy Yeh <andy.yeh@xxxxxxxxx>
> > >
> > > ---
> > > since v1:
> > > -- Remove the address-width field in the example.
> > > since v2:
> > > -- Remove redundant space.
> > >
> > > ---
> > > Documentation/devicetree/bindings/eeprom/at24.txt | 2 ++
> > > 1 file changed, 2 insertions(+)
> > >
> > > diff --git a/Documentation/devicetree/bindings/eeprom/at24.txt
> > > b/Documentation/devicetree/bindings/eeprom/at24.txt
> > > index 61d833a..aededdb 100644
> > > --- a/Documentation/devicetree/bindings/eeprom/at24.txt
> > > +++ b/Documentation/devicetree/bindings/eeprom/at24.txt
> > > @@ -72,6 +72,8 @@ Optional properties:
> > >
> > > - wp-gpios: GPIO to which the write-protect pin of the chip is
> > > connected.
> > >
> > > + - address-width: number of address bits (one of 8, 16).
> > > +
> > > Example:
> > >
> > > eeprom@52 {
> > > --
> > > 2.7.4
> > >
> >
> > Rob,
> >
> > we only have two possibilities here and the default is 8 bits.
> >
> > What do you think about introducing a boolean property here called:
> > 'address-width-16' instead of an integer?
>
> I'd have thought the same, but it turns out address-width is already
> being
> used by the at25 bindings:
>
> Documentation/devicetree/bindings/eeprom/at25.txt
>
Agree with Sakari, there is no need to evolve a chaos in DT bindings. We
have too many semi-hemi-duplications in bindings. Especially in this
case we have already established property by a similar driver.
--
Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
Intel Finland Oy