Let's formally track the status in this issue, please help with answering. If we are confident, that everything is working, we can close this issue and have it show up in the milestone as resolved.
Works fine but when active it emits a high pitched noise which is quite unpleasant. It's not as annoying because the speaker gets switched off automatically when not in use (after 5 seconds or so).
z3ntu: headphone jack detection should be reading a single register (0x318 HMIC Status Register -> Bit 6 -> 0: no jack, 1: jack!). Also 0x1D Jack & Mic detect Control register has to have Bit 7 set to 1 to enable the detection.
MartijnBraam: yes but the jack detect interrupt is in the wrong subsystem in linux
anarsoul: z3ntu: IIRC the problem is that interrupt is in digital codec and actual jack status is in analog codec
anarsoul: so one should export interrupt-controller and another should use it
anarsoul: it's not elegant but I don't see another way to do that
I have no audio output on a fresh install of pmOS with phosh on the PinePhone CE, installed over JumpDrive. I've restarted the device but I have no audio output at all. I notice that Speaker is not here reported as working. Should this "just work" on boot/reboot with the current state of the OS or do I still need to apply some settings?
https://wiki.postmarketos.org/wiki/Phosh would suggest I need to change the configuration but the option it speaks of (Make a phone call) is not present. The available Output Devices are Multichannel Output - Built-in Audio and Headphones - Built-in Audio. The available Input Devices are Multichannel Input - Built-in Audio and Analog Input - Built-in Audio.
Within dmesg, my 1.2 CE board is reported as a 1.1 Braveheart board.
I have the same issue as @thwright on my braveheart (1.1), both on phosh and plasma-mobile on the edge branch (master).
I could also verify that the audio works on the stable branch (v20.05) on both phosh and plasma-mobile but i was unable to pinpoint the package causing the problem.
I can confirm it's fixed, I just don't understand why "introducing a typo" (audio -> audi) fixes it. Shouldn't the proper fix be fixing the typo somewhere else?
ALSA a while ago introduced rule in UCM parser where profile names shouldn't exceed 16 characters (16 with null-terminator) and now Pulseaudio added this rule too.
When I enable AIF1 Loopback (WARNING: loud) in alsamixer, I hear the microphone, so the audio should be getting through at least up until the AIF1 interface.
From the above observations, I suspect that the issue is somewhere in the kernel, and I am kind of stuck on this right now. Any help would be appreciated :)
Does the microphone work in calls? I'm not sure anyone has tested recording audio outside of calls yet - I certainly did not. For calls, the audio profile automatically gets switched in Phosh to the "Make a phone call" profile, which will route the microphone through the modem.
The APKBUILD shows where they get installed. Note that the path is slightly different in the stable channel due to a different pulseaudio version ("audio" instead of "audi"):
I am specifically interested in audio recording by the CPU. (I have not tested it in calls, but I have no reason to believe that it would not work, since AFAICT the microphone is routed directly to the modem on AIF2.)
I tried many different configurations in alsamixer, and was unable to find any combination that would get the capture working, so the issue is probably not with the audio profiles. (Yes, I am aware of the "audi" quirk.)
I have created an issue for the upstream kernel: pine64-org/linux#28