<chbgdn>
narmstrong: xdarklight: bc387887ae22d6ed6439e83968b5be3443a8e57d is the first bad commit when name changed from HDMI/DisplayPort to Analog output, but it not related to crashes. Need spend ~5 minutes to reproduce crash every of ~15 kernel rebuilds, so continue tomorrow.
Daanct12 has joined #linux-amlogic
vagrantc has joined #linux-amlogic
Daaanct12 has joined #linux-amlogic
Danct12 has quit [Ping timeout: 240 seconds]
vagrantc has quit [Quit: leaving]
<narmstrong>
This would probably need a change on alsa script on the Manjaro side since the driver name has changed with this commit
ldevulder has joined #linux-amlogic
Daanct12 has quit [Quit: Leaving]
<chbgdn>
narmstrong: xdarklight: e138233e56e9829e65b6293887063a1a3ccb2d68 is the first bad commit for hdmi-audio crash.
<chbgdn>
Will try to revert this commit in 5.17.3 and check how its going
<chbgdn>
Vanilla 5.17.3 with reverted e138233e56e9829e65b6293887063a1a3ccb2d68, crashes is gone :)
<chbgdn>
And I fixed my alsa card config for new common name of driver, name of output device now HDMI/DisplayPort again
<xdarklight>
chbgdn: thanks for investigating! can you please report this issue to Jerome (jbrunet@baylibre.com) and the mailing lists linux-amlogic@lists.infradead.org and alsa-devel@alsa-project.org
<xdarklight>
I am not familiar with that part of the code, so I am not sure about how to fix the underlying issue or the impact of a revert
<chbgdn>
xdarklight: Sorry, but I'm not familiar with mailing lists and how to use it. Will try to deal with it.
<xdarklight>
chbgdn: just send a plain-text email (no HTML) with a short summary of what you found. that's it :-)
<chbgdn>
xdarklight: thanks, will do it in few hours, want try to catch crash log
<xdarklight>
sure - thanks a lot for dealing with this!
<jbrunet>
I need to rework that. Making the link "non atomic" was a mistake since the period mechanism still rely on a irq.
<jbrunet>
You could revert the 2 commit related to the non-atomic series. The problem it adresses is making sure the backend (the TDM/i2s part) is started before the frontend (the fifo part). Failing to do so gives a random shift in channel assignment when there is more than 2 slots on the TDM link.
<jbrunet>
Unfortunately the FE/BE start order is not easy to configure and by default is the opposite of what we need. To properly solve this, I'll need to make the order configurable in ASoC. I need some time to that. I don't have it right now