dec behavior question

[David Gibson]

On Tue, Oct 25, 2011 at 08:41:55AM -0500, Kumar Gala wrote:
> 
> On Oct 25, 2011, at 7:18 AM, Yoder Stuart-B08248 wrote:
> 
> > 
> > 
> >> -----Original Message-----
> >> From: devicetree-discuss-bounces+stuart.yoder=freescale.com at lists.ozlabs.org
> >> [mailto:devicetree-discuss-bounces+stuart.yoder=freescale.com at lists.ozlabs.org] On Behalf Of
> >> David Gibson
> >> Sent: Monday, October 24, 2011 6:15 PM
> >> To: Kumar Gala
> >> Cc: devicetree-discuss
> >> Subject: Re: dec behavior question
> >> 
> >> On Mon, Oct 24, 2011 at 09:01:49AM -0500, Kumar Gala wrote:
> >>> If I have something like:
> >>> 
> >>> ---------------------------------------
> >>> spi.dtsi:
> >>> 
> >>> &spi0 {
> >>>        #address-cells = <1>;
> >>>        #size-cells = <0>;
> >>>        compatible = "fsl,mpc8536-espi";
> >>>        reg = <0x7000 0x1000>;
> >>>        interrupts = <59 0x2 0 0>;
> >>> };
> >>> ---------------------------------------
> >>> 
> >>> foo.dts:
> >>> 
> >>> spi0: spi at f00bar {
> >>> 	compatible = "FOOBAR";
> >>> };
> >>> 
> >>> /include/ "spi.dtsi"
> >>> 
> >>> ---------------------------------------
> >>> 
> >>> What do you expect the resulting dtb to look like w/regards to the
> >>> compatible node?
> >> 
> >> It's always last tree wins.  So it will end up with compatible = "FOOBAR".
> > 
> > If last wins, then it would be the included file
> > (compatible = "fsl,mpc8536-espi"), no?

Uh, yes, sorry.  I misread and assumed the /include/ came first.

> I don't believe so because the include of spi.dtsi doesn't actually
> instantiate a node.  This is a little different than use of
> 'include' for merging of nodes.

No, it really isn't.  /include/ is just a textual include.  If a node
is defined multiple times, properties in the last definition always
win.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson