dapm, device tree, clock control - questions
Eliot Blennerhassett
eliot at blennerhassett.gen.nz
Tue Oct 27 21:52:24 CET 2020
Hello again,
I got no answers last time I posted this
So, the meta-questions are
* Should I be asking somewhere else?
* Should I be reading something else to find my answers?
* Is there a better way to ask these questions?
thanks
On 21/10/20 10:50 pm, Eliot Blennerhassett wrote:
> Greetings,
>
> First the basic questions, so you can quickly stop reading if this is
> not your thing. Background details follow. Specific answers, or
> pointers to relevant reading both appreciated.
>
> 1) How do I force a DAI clock to keep running even when no audio is
> streaming? Can I do this with DT, or some runtime switch? OR how do I
> disable dapm so everything is powered on all the time?
>
> 2) Two devices: one DAC, one dumb digital audio transmitter both
> connected to same serial audio data - how to model the second one in DT
> (or not)? it has some settings needed after clock starts, must keep
> clock running.
> I have host CPU imx8mm
>
> DAI is sound/soc/fsl/fsl_sai.c
>
> The codec is sound/soc/codecs/ak4458.c
>
> The digital audio transmitter is CS8406 (no driver yet?).
>
> Both the above are connected to the same DAI signals (MCLK, TXC, TXFS, TXD)
--------.
imx8mm |
|
SAI *======+======[ ak4458 ]
| I [ ]->Line_Out
I2C2 *------I--+---[ ]
--------' I |
I `---[ cs8406 ]
"======[ ]->iec958_out
> CS8406 needs MCLK before it can be configured, and loses config if clock
> stops. Currently it is configured via I2C from userspace, but this is
> completely unaware of stream stop/start. Is there a way to hook into
> stream state events so that this setup can be done after the MCLK starts
> before audio starts?
>
> I desire for no discontinuity in analog or digital outputs as streams
> stop and start.
>
> I have read
> Documentation/sound/soc/dapm.rst and its brothers.
> Documentation/devicetree/*
> is there elsewhere I should read to get a newbies intro to this stuff?
>
> It must be about 8 years since I did any ALSA development, and then I
> wasn't in the ASoC space, so quite a learning curve here.
>
> many thanks
>
--
Eliot
More information about the Alsa-devel
mailing list