[PATCH v2 1/5] ARM: shmobile: r8a7740: Add CMT1 device to DT

[Simon Horman]

On Tue, Aug 19, 2014 at 07:17:33PM +0900, Magnus Damm wrote:
> On Tue, Aug 19, 2014 at 4:29 PM, Geert Uytterhoeven
> <geert at linux-m68k.org> wrote:
> > On Tue, Aug 19, 2014 at 8:39 AM, Magnus Damm <magnus.damm at gmail.com> wrote:
> >>>  arch/arm/boot/dts/r8a7740.dtsi | 12 ++++++++++++
> >>>  1 file changed, 12 insertions(+)
> >>>
> >>> diff --git a/arch/arm/boot/dts/r8a7740.dtsi b/arch/arm/boot/dts/r8a7740.dtsi
> >>> index 6cd6a46..1067a96 100644
> >>> --- a/arch/arm/boot/dts/r8a7740.dtsi
> >>> +++ b/arch/arm/boot/dts/r8a7740.dtsi
> >>> @@ -41,6 +41,18 @@
> >>>                 interrupts = <0 83 IRQ_TYPE_LEVEL_HIGH>;
> >>>         };
> >>>
> >>> +       cmt1: timer at e6138000 {
> >>> +               compatible = "renesas,cmt-48";
> >>
> >> Do we want to have the "r8a7740" in the compatible string somewhere?
> >
> > Documentation/devicetree/bindings/timer/renesas,cmt.txt does not
> > mention compatible properties containing SoC names.
> 
> Right, that seems to be missing from the timers. Is there any reason
> why we need to treat those in a special way?
> 
> In my mind this timer case is similar to various SCIF variants
> documented in Documentation/devicetree/bindings/serial/renesas,sci-serial.txt
> - in the case of SCIF we have SoC name in the binding documentation.

FWIW, now you mention it I agree with your point.
But I am interested to hear if Laurent had a specific reason for this.

If it is an oversight then I recommend correcting it as a series
of add-on patches, even for the r8a7740. That will avoid adding
more dependencies to this series.