devicetree.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: David Gibson <david-xT8FGy+AXnRB3Ne2BGzF6laj5H9X9Tb+@public.gmane.org>
To: John Linn <John.Linn-gjFFaj9aHVfQT0dZR+AlfA@public.gmane.org>
Cc: devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org
Subject: Re: dtsi vs dts files, where to better understand this
Date: Thu, 22 Mar 2012 12:01:45 +1100	[thread overview]
Message-ID: <20120322010145.GH15997@truffala.fritz.box> (raw)
In-Reply-To: <d09d039e-88ba-41d3-ac51-8de1a11d806b-+Ck8Kgl/v0/HU/wHWYriTLjjLBE8jN/0@public.gmane.org>

On Wed, Mar 21, 2012 at 02:50:09PM -0600, John Linn wrote:
> > -----Original Message-----
> > From: Stephen Warren [mailto:swarren-3lzwWm7+Weoh9ZMKESR00Q@public.gmane.org]
> > Sent: Wednesday, March 21, 2012 1:37 PM
> > To: John Linn
> > Cc: devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org
> > Subject: Re: dtsi vs dts files, where to better understand this
> > 
> > On 03/21/2012 12:36 PM, John Linn wrote:
> > > Hi,
> > >
> > > I've done some digging and not found any info, maybe I looked in the
> > > wrong places.
> > >
> > > I'm trying to better understand how the dtsi file works with the dts
> > > file and what can go in each file.
> > 
> > The basic idea is that what's inside the SoC is identical across all
> > boards using that SoC. This information is put into the .dtsi file so
> > it
> > can be included by the .dts file for any boards using the SoC.
> > 
> > Anything that's board-specific goes into the board's individual .dts
> > file. This might be new nodes to e.g. define which chips are connected
> > to an I2C bus hosted by the SoC, and much more.
> > 
> > Where properties existing in both, the most recent value in parsing
> > order overrides any earlier values. So for example, if the .dtsi file
> > said that 'status = "disabled"' for a particular HW module, the board
> > .dts could later override it by saying 'status = "okay"' within the
> > same
> > node.
> > 
> > One other factor: Something like an SDHCI controller may have
> > properties
> > that are defined by the SoC (e.g. compatible, reg, interrupts), and
> > properties that are defined by the board (e.g. the GPIO ID to use for
> > CD, WP, etc.). Just set the relevant properties in the .dtsi and .dts
> > files and they'll get merged together to form the final device tree.
> > 
> > Was that what you were looking for?
> 
> That was very helpful.  I was hoping there were details written
> somewhere I had missed.  My apologies if I'm just not getting it yet :)
> 
> It's not clear to me how the hierarchy is maintained across both files
> such as an i2c controller would be in the dtsi, while the i2c eeprom on
> the bus is in the board file.
> 
> Do I have to duplicate the i2c controller (and the bus it's on also) in
> the board file? It's not clear how smart the merge is.
> 
> The examples in the 3.3 tree don't seem that clear to me yet.

The model used by dtc for this is a "stack of overlays".  Once the
/include/s are processed you effectively have one big dts file with
several partial trees specified one after another.  The first one
always starts at /, the later ones can start at / but more likely
start at a subtree of the first tree, specified by a path or a label.

dtc then starts with the first tree and lays the later trees over it
in order.  If the same property is specified in a later tree, it
replaces the value from an earlier tree.  If the same node is
specified, its properties and subnodes are merged into the existing
version.

-- 
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

      parent reply	other threads:[~2012-03-22  1:01 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-03-21 18:36 dtsi vs dts files, where to better understand this John Linn
     [not found] ` <89736f85-54aa-4f71-814e-91ae5ce33bdc-+Ck8Kgl/v0+44QFJ4H5SYbjjLBE8jN/0@public.gmane.org>
2012-03-21 20:36   ` Stephen Warren
     [not found]     ` <4F6A3BD3.4040805-3lzwWm7+Weoh9ZMKESR00Q@public.gmane.org>
2012-03-21 20:50       ` John Linn
     [not found]         ` <d09d039e-88ba-41d3-ac51-8de1a11d806b-+Ck8Kgl/v0/HU/wHWYriTLjjLBE8jN/0@public.gmane.org>
2012-03-22  1:01           ` David Gibson [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20120322010145.GH15997@truffala.fritz.box \
    --to=david-xt8fgy+axnrb3ne2bgzf6laj5h9x9tb+@public.gmane.org \
    --cc=John.Linn-gjFFaj9aHVfQT0dZR+AlfA@public.gmane.org \
    --cc=devicetree-discuss-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).