[PATCH 1/1] ASoC: sunxi: sun4i-i2s: fix LRCLK polarity in i2s mode

[Maxime Ripard]

On Tue, Jul 02, 2024 at 05:17:15PM GMT, Matteo Martelli wrote:
> Maxime Ripard wrote:
> > On Fri, Jun 07, 2024 at 10:04:43AM GMT, Matteo Martelli wrote:
> > > Maxime Ripard wrote:
> > > > > -	/*
> > > > > -	 * 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.
> > > > > -	 */
> > > > 
> > > > I think we should keep that comment somewhere.
> > > 
> > > I think that keeping that comment would be very misleading since the LRCLK
> > > setup would not contradict the datasheet anymore [1][2].
> > >
> > > Also, do you recall any details about the mentioned scope test setup? Was i2s
> > > mode tested in that occasion? It would help clarify the situation.
> > 
> > I can't remember if I tested i2s, I think I did though. But most of the
> > work was done on either TDM or DSP modes, and I remember very clearly
> > that the LRCK polarity was inverted compared to what Allwinner documents.
> > 
> > So the doc was, at best, misleading for these formats and we should keep
> > the comments.
> 
> Thanks for the reply Maxime, would you be able to point out the Allwinner
> document part that is (or was) misleading? The current datasheets (see links
> [1][2]) look correct, the current driver setup for TDM and DSP modes respects
> those datasheets and it's not "reversed compared to the expected polarity on
> the bus" as the comment states. 

I clearly remember having to debug something there, but I don't remember
much more, sorry.

I guess if you have tested on the H3 I2S, TDM and DSP and it all works
as expected with your changes, go ahead and ignore my comment then.

> Also I didn't find any related errata in their changelog. 

Yeah... Allwinner doesn't do errata.

> Could it be possible that during those mentioned tests you
> were still referring to the datasheets of other SoCs like A10 for
> instance? Or maybe that the misleading information was in another
> document rather than the main datasheets? If that's the case, would
> you still think that the comment should be kept as it is?

Possibly, or an older version of the datasheet, I really can't remember.

Maxime
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 273 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20240715/c9c778c3/attachment.sig>