Hi,

On Wed, Apr 29, 2020 at 06:33:00PM +0200, Clément Péron wrote:
> On Wed, 29 Apr 2020 at 14:35, Maxime Ripard <maxime@cerno.tech> wrote:
> >
> > On Tue, Apr 28, 2020 at 10:55:47AM +0200, Clément Péron wrote:
> > > > > +static int sun50i_i2s_set_soc_fmt(const struct sun4i_i2s *i2s,
> > > > > +                              unsigned int fmt)
> > > >
> > > > The alignment is off here
> > > >
> > > > > +{
> > > > > +     u32 mode, val;
> > > > > +     u8 offset;
> > > > > +
> > > > > +     /*
> > > > > +      * DAI clock polarity
> > > > > +      *
> > > > > +      * The setup for LRCK contradicts the datasheet, but under a
> > > > > +      * scope it's clear that the LRCK polarity is reversed
> > > > > +      * compared to the expected polarity on the bus.
> > > > > +      */
> > > >
> > > > Did you check this or has it been copy-pasted?
> > >
> > > copy-pasted, I will check this.
> >
> > It's not going to be easy to do this if you only have a board with HDMI. If you
> > can't test that easily, just remove the comment (or make it explicit that you
> > copy pasted it?), no comment is better than a wrong one.
> 
> I have talked with Marcus Cooper it may be able to test this this week-end.
> Also this can explain why we need the "
> simple-audio-card,frame-inversion;" in the device-tree.
> 
> If think this fix has been introduced by you, correct? Could you say
> on which SoC did you see this issue?

This was seen on an H3

Maxime