Re: [PATCH v2 15/18] dt-bindings: sound: Add bindings for Cirrus Logic Madera codecs

From: Mark Brown <broonie@kernel.org>
To: Richard Fitzgerald <rf@opensource.wolfsonmicro.com>
Cc: lee.jones@linaro.org, linus.walleij@linaro.org, gnurou@gmail.com,
	robh+dt@kernel.org, tglx@linutronix.de, jason@lakedaemon.net,
	alsa-devel@alsa-project.org, patches@opensource.wolfsonmicro.com,
	linux-gpio@vger.kernel.org, devicetree@vger.kernel.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 15/18] dt-bindings: sound: Add bindings for Cirrus Logic Madera codecs
Date: Sun, 14 May 2017 19:04:03 +0900	[thread overview]
Message-ID: <20170514100403.nocjlb77q5r3nvo2@sirena.org.uk> (raw)
In-Reply-To: <1493137664.4826.75.camel@rf-debian.wolfsonmicro.main>

[-- Attachment #1: Type: text/plain, Size: 1101 bytes --]

On Tue, Apr 25, 2017 at 05:27:44PM +0100, Richard Fitzgerald wrote:
> On Tue, 2017-04-25 at 16:52 +0100, Mark Brown wrote:

> > > +Required properties:
> > > +  - compatible : One of the following chip-specific strings:
> > > +        "cirrus,cs47l35-codec"
> > > +        "cirrus,cs47l85-codec"
> > > +        "cirrus,cs47l90-codec"

> > You shouldn't have compatible strings for subfunctions of a MFD unless
> > these represent meaningful reusable IPs that can exist separately from
> > the parent chip, that's clearly not the case here.  All you're doing
> > here is encoding Linux internal abstractions which aren't OS neutral and
> > might change in future (for example clocking might move more into the
> > clock API).

> While that's nice, the of_node doesn't get populated if there isn't a
> compatible string. And people don't like workarounds for the missing
> of_node.

What workarounds are you referring to?  Why would this need any kind of
workaround, there is no requirement for magic broken nodes like this in
the subsystem and if there is we should fix that rather than bodge the
ABI.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]