Message-ID: <1870492.5CqhBlFY90@archbook>
Date:   Thu, 19 Aug 2021 19:01:25 +0200
From:   Nicolas Frattaroli <frattaroli.nicolas@...il.com>
To:     Mark Brown <broonie@...nel.org>
Cc:     Liam Girdwood <lgirdwood@...il.com>,
        Rob Herring <robh+dt@...nel.org>,
        Heiko Stuebner <heiko@...ech.de>,
        Robin Murphy <robin.murphy@....com>,
        alsa-devel@...a-project.org, devicetree@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org,
        linux-rockchip@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/4] dt-bindings: sound: add rockchip i2s-tdm binding

On Donnerstag, 19. August 2021 16:16:17 CEST Mark Brown wrote:
> On Thu, Aug 19, 2021 at 03:52:55PM +0200, Nicolas Frattaroli wrote:
> > On Donnerstag, 19. August 2021 14:08:36 CEST Robin Murphy wrote:
> > > > +  rockchip,no-dmaengine:
> > > > +    description:
> > > > +      If present, driver will not register a pcm dmaengine, only the
> > > > dai.
> > > > +      If the dai is part of multi-dais, the property should be
> > > > present.
> > > > +    type: boolean
> > > 
> > > That sounds a lot more like a policy decision specific to the Linux
> > > driver implementation, than something which really belongs in DT as a
> > > description of the platform.
> > 
> > I agree. Should I be refactoring this into a module parameter or
> > something along those lines? I'm unsure of where this goes.
> 
> Why is this even required?  What is "multi-dais" and why would
> registering the DMA stuff cause a problem?

After some research, it appears that multi-dais is a special driver that
downstream uses to allow multiple sub-DAIs to be combined into one DAI
that has all the channels of the sub-DAIs. This does not seem like
something that should be done at that level to me, because it seems
like it's pushing a sound driver configuration into the realm of
hardware description.

In retrospect, I should have stripped this out before submitting it,
because I should not be submitting things I don't understand completely.
I apologise.

> > The particular configuration may even vary per-board; an I2S/TDM
> > controller may be connected to an external codec which does not
> > support capture, whereas on another board it may be connected to
> > one that does.
> 
> If the external device doesn't support both directions then why does the
> driver for the I2S controller in the CPU care?  The constraint handling
> code in the core will ensure that nothing tries to start something that
> isn't supported

I went over the downstream text binding description again and from that
it appears that the playback/capture-only capability is something
specific to the controller, not to any device connected to it.

The downstream device tree for the rk3568 specifies playback-only for
I2S0, a.k.a. the one connected to the HDMI that I can't test because
we currently don't have a video clock. Another downstream device tree,
specific to what appears to be a robot demo for the px30 SoC, uses this
property on i2s1, which tells me that this is not an actual description
of the controller hardware but just a description of the application.

While not relevant to the device tree schema, the driver reacts to these
properties by setting the opposite directions _minimum_ channel number
to 0 (from the default of 2.)

My conclusion from this is that this reeks of nonsense and I will look
into what happens when I simply remove these properties and lower the
channels_min to 0 in the driver. If it turns out that on some SoC for
some I2S/TDM controller instance there is a limitation where specifying
that the controller only handles either capture or playback does make
sense, we can always add it later.

Thank you for your comments,
Nicolas Frattaroli

Powered by blists - more mailing lists