00:01 <brads> Im sorry to hear about the attacks, thoughts are with france.

00:05 <brads> jbrunet: minor issue in the yaml docuentation for dai-formats gx and upcomming axg, left-j should be left_j - https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/devicetree/bindings/sound/amlogic,gx-sound-card.yaml?h=v5.10-rc1#n51

00:17 <brads> Im actually trying to work out why my sound channels are swapped using i2s mode so attempted left_j but this seems to mix everything to the left speaker. maybe I have my mclk-fs = <128> a bit wrong (but this would maximise the supported formats /rates for the pcm5242)

00:36 <tuxd3v> when you let extremist cultures take possession of your own countries, bad things will happen

00:37 <tuxd3v> I don't know about any recent attacks, but for what I read there seems to be there some news, that I didn't red yet..

00:49 <brads> cool I can the channels unswapped using left_j and dai-tdm-slot-width = <16>; (rather than 32), wasn't able to try 24 as it it could not derive the clock "axg-tdm-iface audio-controller-1: can't derive sclk 2116800 from mclk 5644800" but maybe I could with different clock source

00:55 <brads> sound is a bit off with dai-tdm-slot-width = <16> (some bits must be out of wack) but channels are correct

08:51 <jbrunet> brads: left-j: you are right- feel free to submit a fix

08:52 <jbrunet> mixed channel usually means the format is not same on both side, often because of the i2s 1 clock cycle skew

08:56 <jbrunet> clock derivation: the problem is not the sources - notice the 3 PLLs:

08:56 <jbrunet> https://www.irccloud.com/pastebin/W1bTSIkW/

08:58 <jbrunet> the rate are 24 * 32 * {44100, 48000, 32000) * 2^n -> so the rates from usual clock freqs can be drived

09:00 <jbrunet> your problem is mclk-fs. if you left 256 then and your slot width is 24 it won't work because 256 cannot be divided by 24

09:01 <jbrunet> mclk-fs must be slotwidth * channel count * n

12:47 <brads> jbrunet: awsome thanks! I will do some more playing around