Grant Likely wrote at Wednesday, July 20, 2011 12:32 PM:
> On Wed, Jul 20, 2011 at 05:03:55PM +0100, Mark Brown wrote:
> > On Wed, Jul 20, 2011 at 08:37:19AM -0700, Stephen Warren wrote:
> > > Grant Likely wrote at Tuesday, July 19, 2011 5:43 PM:
> >
> > > > +	sound {
> > > > +		compatible = "nvidia,harmony-sound", "nvidia,tegra-wm8903";
> >
> > > I thought the sound bindings were still somewhat experimental and not
> > > completely agreed upon. One issue I see is that Device Tree is
> > > supposed to represent pure HW, rather than driver-required abstractions,
> > > and at least the compatible name here is pretty Linux-driver-specific.
> >
> > The current decision is that the schematic for embedded audio hardware
> > is sufficiently interesting to be considered hardware in its own right
> > separately to the chips contained within it.
> 
> Correct.  For complex composite devices like audio, it is completely
> appropriate to have a root node that represents the entire complex and
> how it is wired together.

Sure, that makes sense.

> The compatible property here definitely represents the hardware
> because it reflects the sound infrastructure on the harmony board.

I can see that argument for "nvidia,harmony-sound".

But "nvidia,tegra-wm8903" is pretty generic; I can certainly see there
being Tegra 20 systems that use a WM8903 but are so different from either
Harmony or anything supported by sound/soc/tegra/tegra_wm8903.c that the
existing driver isn't applicable. And hence, there may be a different
ASoC driver for such board(s), and hence choosing such a generic name as
"tegra-wm8903" for Harmony/Seaboard's audio layout seems like it might
cause problems in the future.

I'm fine with such a generic name for the platform driver, since that name
can fairly easily be modified just by editing the driver and the board file.
But since *.dts files are at least logically separate from the kernel, such
naming future-proofing is a little more important.

-- 
nvpublic