Re: [PATCH 2/4] ARM: shmobile: r8a7790: add I2C support in Device Tree mode

[Guennadi Liakhovetski <g.liakhovetski@gmx.de>]

Hi Magnus

On Thu, 26 Sep 2013, Magnus Damm wrote:

> Hi Guennadi,
> 
> On Thu, Sep 26, 2013 at 7:10 AM, Guennadi Liakhovetski
> <g.liakhovetski@gmx.de> wrote:
> > Hi Laurent
> >
> > On Wed, 25 Sep 2013, Laurent Pinchart wrote:
> >
> >> Hi Guennadi,
> >>
> >> Thank you for the patch.
> >>
> >> On Monday 09 September 2013 18:03:54 Guennadi Liakhovetski wrote:
> >> > This patch adds clocks and clock lookup entries for the four I2C
> >> > controllers on r8a7790 and respective Device Tree nodes.
> >> >
> >> > Signed-off-by: Guennadi Liakhovetski <g.liakhovetski+renesas@gmail.com>
> >> > ---
> >> >  arch/arm/boot/dts/r8a7790.dtsi         |   36 +++++++++++++++++++++++++++++
> >> >  arch/arm/mach-shmobile/clock-r8a7790.c |   10 ++++++++
> >> >  2 files changed, 46 insertions(+), 0 deletions(-)
> >> >
> >> > diff --git a/arch/arm/boot/dts/r8a7790.dtsi b/arch/arm/boot/dts/r8a7790.dtsi
> >> > index 885f9f4..a5021112 100644
> >> > --- a/arch/arm/boot/dts/r8a7790.dtsi
> >> > +++ b/arch/arm/boot/dts/r8a7790.dtsi
> >> > @@ -127,6 +127,42 @@
> >> >             interrupts = <0 0 4>, <0 1 4>, <0 2 4>, <0 3 4>;
> >> >     };
> >> >
> >> > +   i2c0: i2c@e6508000 {
> >> > +           #address-cells = <1>;
> >> > +           #size-cells = <0>;
> >> > +           compatible = "renesas,i2c-rcar-h2";
> >> > +           reg = <0 0xe6508000 0 0x40>;
> >> > +           interrupt-parent = <&gic>;
> >> > +           interrupts = <0 287 0x4>;
> >>
> >> Shouldn't you add state = "disabled" to all I2C controllers in order not to
> >> enable the unused controllers by default ?
> >
> > It would be logical, yes, and I seem to remember having discussed this
> > earlier with someone (with Magnus, IIRC), and the outcome was, that all
> > Renesas .dtsi files so far define all I2C directly in enabled mode and
> > that it's intentional, so, I just followed this pattern here. You can
> > indeed check in other .dtsi files - all seem to do exactly the same.
> 
> Uhm, I think you mix up platform device use case and DT use case.

I'm sure I don't.

> In
> the platform device case we define all devices by default, but that's
> not how it should be for the DT case. In the case of DT then default
> should most likely be "disabled". Please follow the direction of
> Laurent.

To put the record straight - originally I suggested to use status 
"disabled" for mmc devices in .dtsi, before that the status property 
wasn't used in Renesas .dts(i) files, and that's when all I2C nodes were 
already present in .dtsi in enabled state.

> Also, I mentioned this 25 times before already so once more cannot
> hurt: "renesas,i2c-rcar-h2" needs to be replaced with something more
> standard.

I didn't count, but yes, it has been discussed and that's been fixed in v2 
of my i2c patches. Now that I've got comments to the ARM patches I can 
also update them with a correct compatibility string.

Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/