On 2022-08-18 11:25 AM, Icenowy Zheng wrote:
在 2022-08-17星期三的 15:19 +0200，Cezary Rojewski写道：
...
NHLT was left alone across all the updates. Updating it is one way of fixing problems but I don't believe it's necessary in your case. Topology update is more desirable approach.
BTW how could I fix the topology?
I now use topology files from GalliumOS (which, I assume, is extracted from ChromeOS).
(save)
That clarifies things out. Guess the kernel version used there is v4.4 (plus a ton of un-upstreamed patches). Again, will propagate the information up the chain. Perhaps one of the solutions for end-users would be providing working UCM files to alsa-topology-conf repo so users are not powerless in situations such as this one.
The message of mine above is the answer to this. Without some ADSP firmware knowledge it's almost impossible to write a topology file from scratch. During v4.4 -> v5.4 transition several problems with endpoints were detected as not all of them were behaving as expected. Long story short, I do not believe pure v4.4 on some of the designs (mainly I2S designs) works as intended. Some external patches are needed for that to happen.
The situation on v5.4 is much cleaner - the problem there are the topology files, as these are not updated automatically when you flash new kernel. In fact, these were never shared in alsa-topology-ucm repo. Only HDAudio topology has been fixed and shared. As long as you stick to the original ChromeOS the issues on your machine should cease to exist. I need some approvals before I2S designs can be shared. It's not a process that takes a day or two, unfortunately.
Regards, Czarek