[PATCH v2 1/2] fsl: Add binding for RCPM

[Scott Wood]

On Tue, 2015-09-15 at 21:30 -0500, Wang Dongsheng-B40534 wrote:
> Hi Scott,
> 
> > -----Original Message-----
> > From: Wood Scott-B07421
> > Sent: Wednesday, September 16, 2015 10:19 AM
> > To: Wang Dongsheng-B40534
> > Cc: devicetree at vger.kernel.org; linuxppc-dev at lists.ozlabs.org;
> > robh+dt at kernel.org; linux-arm-kernel at lists.infradead.org; Wang Huan-
> > B18965; Jin
> > Zhengxiong-R64188; Zhao Chenhui-B35336; Tang Yuantian-B29983
> > Subject: Re: [PATCH v2 1/2] fsl: Add binding for RCPM
> > 
> > On Tue, 2015-09-15 at 21:15 -0500, Wang Dongsheng-B40534 wrote:
> > > Hi Scott,
> > > 
> > > > -----Original Message-----
> > > > From: Wood Scott-B07421
> > > > Sent: Wednesday, September 16, 2015 7:57 AM
> > > > To: Wang Dongsheng-B40534
> > > > Cc: devicetree at vger.kernel.org; linuxppc-dev at lists.ozlabs.org;
> > > > robh+dt at kernel.org; linux-arm-kernel at lists.infradead.org; Wang Huan-
> > > > B18965; Jin
> > > > Zhengxiong-R64188; Zhao Chenhui-B35336; Tang Yuantian-B29983
> > > > Subject: Re: [PATCH v2 1/2] fsl: Add binding for RCPM
> > > > 
> > > > On Tue, 2015-09-15 at 16:55 +0800, Dongsheng Wang wrote:
> > > > > +* Freescale RCPM Wakeup Source Device Tree Bindings
> > > > > +-------------------------------------------
> > > > > +Required rcpm-wakeup property should be added to a device node if
> > > > > +the
> > > > > device
> > > > > +can be used as a wakeup source.
> > > > > +
> > > > > +  - rcpm-wakeup: The value of the property consists of 3 cells.
> > > > > + The
> > > > > first
> > > > > cell
> > > > > +     is a pointer to the rcpm node, the second cell is the bit
> > > > > + mask
> > > > > that
> > > > > +     should be set in IPPDEXPCR0, and the last cell is for 
> > > > > IPPDEXPCR1.
> > > > > +     Note: If the platform has no IPPDEXPCR1 register, put a zero 
> > > > > here.
> > > > 
> > > > What if a future platform has more than two of these registers?
> > > 
> > > Those registers are only used for wakeup device, we have a lot of
> > > available bit for feature. For example, In LS1021a platform only 7bits
> > > has used in the registers, and 57bits is reserved.
> > 
> > Still, it'd be better to for the rcpm node to advertise the number of 
> > cells it
> > expects.
> 
> For the foreseeable future it should be enough to use, even if not enough 
> to use in
> the future at that time we can update the binding.

That's the whole point.  Device tree is stable ABI.  Updating it later to not 
be fixed to two cells would be a lot harder than getting it right from the 
beginning.  Putting the number of cells in the phandle target is a standard 
device tree idiom.

-Scott