17:22 <geraldog> narmstrong: So, I remembered that older Amlogic hardware has older version of dw-hdmi IP and I checked my hunch

17:22 <geraldog> narmstrong: I was wrong. insert_pcuv is not set in earlier version of dw-hdmi IP and it doesn't make a difference

17:23 <geraldog> narmstrong: I don't know what to make of this ...

17:27 <geraldog> narmstrong: I'm hoping that Christian Hewitt's testing goes well and the bad HDMI sound on RK3399 and beyond is solved. But don't ask me to explain all this

18:50 <geraldog> narmstrong: What I mean is: from my understanding, dw-hdmi hardware is generating valid PCUV bits for the HDMI audio stream, from early on in their IP version. But now they introduced this bit in 1.40a according to your research, and it only became active in 2.10a ?? And if you miss setting the bit to 1, you get no PCUV insertion ... ?

18:51 <geraldog> narmstrong: I appreciate that they give control to the user, so even from userspace, apparently, from 2.10a onwards you can open the interface in 32 bits and generate your own PCUV from userspace?

18:52 <geraldog> (who would want to do that is beyond my understanding)

19:17 <narmstrong> No I think it’s related to the GPAUD mode which is enabled on the dw-hdmi in rk3399, GPAUD is a « generic » audio interface unlike the I2S one we use and I think when there’s no GPAUD (like on amlogic) this bit is not required, but perhaps when GPAUD is available this bit is added and is required to be set when using the i2s interfaces

19:20 <narmstrong> geraldog: the IP must automatically generate PCUV from i2s streams when GPAUD isn’t enabled, and PCIV must be be in the spdif stream. So probably we should set that bit if GPAUD/GDOUBLE is in the CONF registers only

19:23 <narmstrong> geraldog: can you dump the CONFIGX_ID registers on the rk3399pro soc ?