On Mon, Sep 02, 2019 at 04:35:56PM +0300, Daniel Baluta wrote:
> On Mon, Sep 2, 2019 at 3:42 PM Mark Brown <broonie@kernel.org> wrote:

> > This patch seems to do this unconditionally.  This is fine for
> > configurations where the SoC is the only thing driving the bus but will
> > mean that for TDM configurations where something else also drives some
> > of the slots we'll end up with both devices driving simultaneously.  The
> > safest thing would be to set this only if TDM isn't configured.

> I thought that the SAI IP is the single owner of the audio data lines,
> so even in TDM
> mode SAI IP (which is inside SoC) is the only one adding data on the bus.

> Now, you say that there could be two devices driving some of he masked
> slots right?

Doing that is the major point of TDM modes.  It could even be another
SAI on the same bus.

> I'm not sure how to really figure out that SAI is running in TDM mode.

As a first approximation you could just check if set_tdm_slots() has
been called, it might still be the only device but it's a good first
guess.