[alsa-devel] No device for DAI %s

[Mark Brown]

On Sun, Sep 05, 2010 at 01:59:27PM -0400, Patrick Doyle wrote:

Please always remember to CC maintainers on kernel related mail.

> I am curious -- what is the intent of that warning message?  If the
> device "should become mandatory over time" when will that happen
> (obviously, when the developer of a particular device makes it
> happen), but what would motivate the developer to do so?  If I am that

Eventually the motivation would be that the device would become
mandatory and the driver would stop working.  And indeed this has
already happened with current ASoC.

> developer, and I am adapting code from another device, what do I need
> to do (and why should I do it) to associate a device with a DAI?

Fill in that pointer with the struct device you probed off.

> Actually, in a more general sense, is there any good
> article/book/tutorial/blog explaining the design and architecture of
> the SoC code in ALSA?  Right now, I am in a maze of little twisty
> passages, all alike, composed of calls to snd_soc_register_dai(),
> platform_register_driver(), platform_register_device, "soc-audio"
> platform drivers, i2s drivers, i2c drivers, etc... and would like to
> find something to help me pull my head back above water.  (Have I
> mixed enough metaphors here?)

Not outside the code.  Basically it's just the same as any other
subsystem: you probe off whatever bus your device is physically attached
to and then register with the subsystem.