[PATCH 7/7] watchdog: orion: Update device-tree binding documentation

[ezequiel.garcia@free-electrons.com (Ezequiel Garcia)]

On Fri, Aug 23, 2013 at 07:04:51AM -0300, Ezequiel Garcia wrote:
> On Thu, Aug 22, 2013 at 09:26:21PM -0400, Jason Cooper wrote:
> > On Thu, Aug 22, 2013 at 11:41:58AM -0300, Ezequiel Garcia wrote:
> > > Change the 'reg' property meaning, by defining three required cells.
> > > It's important to note this commit breaks DT-compatibility for this
> > > device.
> > > 
> > > Cc: devicetree at vger.kernel.org
> > > Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
> > > ---
> > >  Documentation/devicetree/bindings/watchdog/orion-wdt.txt | 9 +++++++--
> > >  1 file changed, 7 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/Documentation/devicetree/bindings/watchdog/orion-wdt.txt b/Documentation/devicetree/bindings/watchdog/orion-wdt.txt
> > > index 5dc8d30..bb7f1a2 100644
> > > --- a/Documentation/devicetree/bindings/watchdog/orion-wdt.txt
> > > +++ b/Documentation/devicetree/bindings/watchdog/orion-wdt.txt
> > > @@ -3,7 +3,10 @@
> > >  Required Properties:
> > >  
> > >  - Compatibility : "marvell,orion-wdt"
> > > -- reg		: Address of the timer registers
> > > +- reg		: Three cells are required.
> > > +		  First  cell contains the global timer control register.
> > > +		  Second cell contains the watchdog counter register.
> > > +		  Third  cell contains the RSTOUT register.
> > >  
> > >  Optional properties:
> > >  
> > > @@ -13,7 +16,9 @@ Example:
> > >  
> > >  	wdt at 20300 {
> > >  		compatible = "marvell,orion-wdt";
> > > -		reg = <0x20300 0x28>;
> > > +		reg = <0x20300 0x4
> > > +		       0x20324 0x4
> > > +		       0x20108 0x4>;
> > 
> > I don't like this.  It reaches outside of the wdt register.  I think a
> > more clean way to do this is to do a provider/consumer relationship as
> > in reset.txt.  eg, here you would retain the original reg binding, and
> > add a reset phandle.
> > 
> 
> Mmm... I can't see how this fits a reset-controller usage.
> 
> The watchdog simply "enables" the RSTOUT bit that allows the whole SoC
> to be reset when the watchdog counter expires.
> 
> The reset-controller seems to be meant to send reset signals to devices,
> which is not this case.
> 
> What am I missing?

Another possible solution is to simply "enable" the RSTOUT bit for
watchdog somewhere in mach-{kirkwood,mvebu,...} at board boot-up time.

Do you think that would have any drawbacks?
-- 
Ezequiel Garc?a, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com