Re: [PATCH v7 2/2] ASoC: fsl: Add S/PDIF machine driver

[Philipp Zabel <p.zabel@pengutronix.de>]

Am Montag, den 19.08.2013, 11:01 +0100 schrieb Mark Rutland:
> On Mon, Aug 19, 2013 at 10:50:43AM +0100, Nicolin Chen wrote:
> > Hi,
> > 
> > On Mon, Aug 19, 2013 at 10:24:58AM +0100, Mark Rutland wrote:
> > > Is this used semantically, or is it a completely arbitrary string?  In
> > > either case I don't see why the compatible string doesn't give the
> > > driver enough to have a sensible value.
> > > 
> > > I'm confused as to why we need this. The phrase "user-visible" in a
> > > device description seems very odd.
> > 
> > The string would be in the ALSA device list:
> > ALSA device list:
> >   #0: imx-spdif
> > 
> > I think it can be a sort of arbitrary as long as users know which this 
> > device exactly is when they catch the name by 'aplay -l' or 'arecord -l'
> > 
> > The phrase "user-visible" is being used in many current docs, I don't 
> > dare to change it unless a sage gives me a suggestion.
> 
> I can see that there is entrenched usage, but this really seems to be
> embedding Linux-specific implementation details into the dt. I don't see
> why the driver cannot select a sensible name, but perhaps I'm missing
> something.
> 
> Mark, is there any reason we need to handle the user-visible name of the
> device this way?
> 
> > 
> > > > +
> > > > +  - spdif-controller : The phandle of the i.MX S/PDIF controller
> > > > +
> > > > +
> > > > +Optional properties:
> > > > +
> > > > +  - spdif-transmitter : The phandle of the spdif-transmitter dummy codec
> > > > +
> > > > +  - spdif-receiver : The phandle of the spdif-receiver dummy codec
> > > > +
> > > > +* Note: At least one of these two properties should be set in the DT binding.
> > > 
> > > Are all four units (comlpex,controller,transmitter,receiver) really
> > > separate blocks?
> > 
> > At least they are separate drivers as I mentioned in the commit comments.
> 
> I'm not sure that the boundary of Linux drivers should necessarily
> determine the way we carve up the description of IP blocks, though
> presumably it's a pretty sensible way of carving it up or we wouldn't
> have done it.

The transmitter and receiver can be real external codec devices with
S/PDIF input or output pads connected to the i.MX S/PDIF. One example
would be the Analog Devices ADV7612 HDMI receiver, which can output
audio taken from HDMI input via S/PDIF (just as via I2S).

The dummy codec devices are just needed if the S/PDIF pads are directly
routed to externally accessible connectors.

regards
Philipp