| T42_ | <adampigg> No, someone asked me | 07:55 |
|---|---|---|
| T42_ | <b100dian> Did anything change w.r.t qti autodetection of audiosystem-passthrough between 4.4 and 4.5 SFOS? | 10:12 |
| T42_ | <b100dian> I made a bizzare "fix" to set the env variable manually today, but am pretty sure I didn't touch that when I started with the 4.4 SFOS port and audio in calls "worked". | 10:16 |
| T42_ | <b100dian> But I didn't have any of these files either then https://github.com/mer-hybris/pulseaudio-modules-droid-hidl/blob/master/src/hidl/module-droid-hidl.c#L356 | 10:16 |
| mal | @b100dian what kind of fix did you make? | 11:27 |
| T42_ | <b100dian> This https://github.com/sailfishos-on-sake/droid-config-sake/commit/1060eef714153228b0c9bd64fe48dd86bac037bb | 11:28 |
| mal | do you have those manifest.xml files anywhere on the device? | 11:29 |
| mal | @b100dian how can that help, qti is the default https://github.com/mer-hybris/audiosystem-passthrough/blob/master/src/passthrough-helper.c#L100 you can se it checks the env var and if not found uses qti | 11:32 |
| T42_ | <b100dian> Yes, but pulseaudio-modules-droid-hidl seems to set the env to "af" because it doesnt find those manifests. Still, I don't see how can that be overriden by the pulseaudio env, I'm also surprised | 11:43 |
| T42_ | <b100dian> I have `/vendor/etc/vintf/manifest/manifest.xml` (notice the extra /manifest/) and `/system/etc/vintf/manifest.xml` | 11:44 |
| T42_ | <b100dian> Neither contain IQcRilAudio though | 11:45 |
| mal | hmm | 11:45 |
| mal | which android base is that? | 11:45 |
| T42_ | <TheVancedGamer> have you tried grepping for it? | 11:45 |
| mal | also check /odm just in case | 11:46 |
| T42_ | <b100dian> I started a grep now. I'ts A11 (lineage 18.1) | 11:46 |
| T42_ | <TheVancedGamer> it will never be in /system, fyi | 11:46 |
| T42_ | <TheVancedGamer> so only check vendor and odm | 11:46 |
| T42_ | <b100dian> It's `/vendor/etc/vintf/manifest_lahaina.xml` where it finds it | 11:47 |
| T42_ | <TheVancedGamer> yeah, so you do have iQcRilAudio | 11:47 |
| T42_ | <TheVancedGamer> and split manifests, hm | 11:47 |
| T42_ | <b100dian> Yes, that's why the "fix" works. Now, I don't know why the env works when setting it in pulseaudio, that I would have to dig a bit, but I am 99% sure it worked on sfos 4.4.0.72 without this.. | 11:48 |
| mal | @b100dian I would just use the fix for now, not sure what is the best way to handle split manifests | 11:48 |
| mal | @b100dian it works because of this https://github.com/mer-hybris/pulseaudio-modules-droid-hidl/blob/master/src/hidl/module-droid-hidl.c#L381C9-L381C22 that sets the env in the helper I linked earlier | 11:50 |
| mal | well not exactly the, env you define actually overrides what pulseaudio tells | 11:51 |
| mal | so pulseaudio hidl module thinks we want af, the env you set tells helper to not use what pulseaudio sets | 11:52 |
| mal | @b100dian that is exactly why the env var exists, to force correct mode if something goes wrong, it's not ugly way at all | 11:53 |
| T42_ | <b100dian> Agreed, this is the way to go, for now. I was just surprised that I seem to recall it worked, and wondered if there was anything changed that I didn't detect in the changelogs for 4.5.0.x | 11:54 |
| T42_ | <b100dian> I even tested bluetooth calls on SFOS 4.4.0.72 back then and they worked too, that's why I say I'm pretty sure. | 11:55 |
| T42_ | <b100dian> Anyway, a quest for another time | 11:56 |
| T42_ | <b100dian> Wonder if @k1gen had the same thing | 11:57 |
| mal | @b100dian did you change anything in your adaptation? the android side build of things | 11:57 |
| T42_ | <b100dian> Mostly additions, like hwcrypt for community encryption, other bits for community fingerprint, droidmedia changes we talked about before.. But using the same lineage base system and vendor built on the first day.. | 12:03 |
| T42_ | <elros34> you are building pulseaudio-modules-droid-hidl so even if that was updated you still use own version (whether that was in you 4.4.0 release) | 12:04 |
| T42_ | <b100dian> Let me check, maybe it was changed when I started using OBS | 12:05 |
| T42_ | <b100dian> It seems I was building 1.5.0 locally, checked out on March 2023, but I have 1.4.0 in OBS. Also, today, on device, there is 1.5.0 installed from "(System Packages)" instead of 1.4.0 from "adaptation-community" | 12:12 |
| T42_ | <elros34> so nothing changed regarded this package | 12:21 |
| T42_ | <elros34> but probably you are building from old ks file with local repo | 12:23 |
| mal | also only difference between 1.4.0 and 1.5.0 is the build system, autotools vs meson | 12:26 |
| T42_ | <b100dian> Yeah, tried to dig a little bit more through the diffs and the install history and ks file but don't know what happened. Let's leave it like this, better it working with a reason than accidentally. | 13:37 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!