| freemangordon | dsc_: do you plan to make a new conversations release? is there a stopper or you just lack time? | 07:27 |
|---|---|---|
| freemangordon | Wizzup: the guaranteed way to break audio in mafw is to start call from the dialer and to hangup before other party has answered | 07:42 |
| freemangordon | right after restart it works ok | 07:42 |
| freemangordon | so it is not some config that's broken | 07:42 |
| freemangordon | but, it is PA that fails | 07:43 |
| freemangordon | restarting PA makes mafw work again | 07:52 |
| dsc_ | freemangordon: do you mean a stable release? | 07:57 |
| dsc_ | or just a release in general | 07:57 |
| freemangordon | just a release | 07:57 |
| dsc_ | ah | 07:57 |
| dsc_ | i didnt make a release because the menu doesnt give control back to the user | 07:57 |
| dsc_ | the context menu for a message | 07:58 |
| dsc_ | need to fix it | 07:58 |
| dsc_ | right now working on adblock in jib | 07:58 |
| freemangordon | ah, right | 07:58 |
| freemangordon | ok | 07:58 |
| freemangordon | Wizzup: https://pastebin.com/F8UJYcci | 08:03 |
| freemangordon | this is a trace of the issue | 08:03 |
| freemangordon | seems sphone(?) leaves wrong UCM profile on hangup | 08:04 |
| freemangordon | hmm, no, HiFi profile is selected | 08:10 |
| freemangordon | but, "Selected mapping 'Headset Microphone + Internal Microphone' (HiFi: hw:Audio,1: source)." | 08:10 |
| freemangordon | oh, that's source | 08:11 |
| freemangordon | switching profiles from PA volume control makes it work again | 08:13 |
| Wizzup | 07:43 < freemangordon> but, it is PA that fails | 09:36 |
| Wizzup | I'm pretty sure it's a kernel problem still | 09:36 |
| Wizzup | and switching the ucm solves it | 09:36 |
| freemangordon | mhm | 10:17 |
| freemangordon | maybe I shall try to fix that | 10:17 |
| freemangordon | maybe you can help by giving me hints which registers to check and how | 10:17 |
| Wizzup | I think it would make more sense to start from my email regarding 6.6 | 10:19 |
| Wizzup | since it's all the same problem more or less | 10:19 |
| freemangordon | yes, but it would be easier for me to use 6.1 for debugging | 10:20 |
| Wizzup | ok | 10:20 |
| freemangordon | because we have something working ther | 10:20 |
| Wizzup | true... | 10:20 |
| freemangordon | while in 6.6 I have no base to compare with (working vs non) | 10:20 |
| freemangordon | ok, I'll ping you later on (after working hours) | 10:21 |
| Wizzup | even on 6.1 I think you should probably try to dump the registers before the call and after the call | 10:28 |
| Wizzup | probably with hifi playing (or both with it not playing) | 10:28 |
| Wizzup | https://github.com/MerlijnWajer/droid4-cpcap-audio-reg-tool | 10:28 |
| freemangordon | uvos: we discuss if it makes sense to have cpu call recording always active during call, in order to create alsa control that enables call DAPM (or whatever needs to be active) | 13:30 |
| freemangordon | Wizzup: ^^^ basically that's it, right? | 13:30 |
| Wizzup | we would still need userspace to actually record to activate the path, but yes | 13:30 |
| Wizzup | and I linked uvos' call recording txt https://uvos.xyz/maserati/stockinfo/call/ | 13:30 |
| freemangordon | no, we can just turn the swicth on in UCM, no? | 13:30 |
| Wizzup | no, there must be an active path in alsa for dapm to do anything | 13:31 |
| freemangordon | ah | 13:31 |
| freemangordon | hmm | 13:31 |
| Wizzup | it makes sense but at the same time it's annoying | 13:31 |
| freemangordon | what was the otehr option? | 13:31 |
| Wizzup | we could make a switch in alsa that force triggers dapm | 13:31 |
| Wizzup | have to go, back later | 13:31 |
| freemangordon | later | 13:32 |
| Wizzup | there would be a clear impact on battery of course if we record all the time to keep it active, but: | 13:53 |
| Wizzup | 1. we'll have call recording which is nice | 13:53 |
| Wizzup | 2. it'll work without dapm hacks | 13:53 |
| freemangordon | we can disregard the data | 13:53 |
| Wizzup | yes sur | 13:53 |
| Wizzup | the alternative, a switch that forces dapm for call on, is possible too | 13:53 |
| Wizzup | but it would still be somewhat race sensitive if something else messes with alsa too | 13:53 |
| freemangordon | we can use such switch in UCM, no? | 13:54 |
| freemangordon | also, do we have some d4 audio diagram? | 13:54 |
| Wizzup | yes but it if is a force dapm it can be overriden by say hifi also being enabled | 13:54 |
| freemangordon | is there any issue of hifi and voice being active simultaneously? | 13:55 |
| freemangordon | they use different codecs, tdm slots, etc | 13:55 |
| freemangordon | (maybe) | 13:55 |
| freemangordon | tmlind: ^^^ please, help :) | 13:55 |
| freemangordon | Wizzup: hmm, wait, wait | 13:56 |
| freemangordon | can't we create a DAI for voice, but produce only zeroes unless monitoring is enabled? | 13:56 |
| freemangordon | maybe I am talking nonsense, I admit ALSA is not a thing I know much about | 13:58 |
| freemangordon | so, we may have 3 UCM profiles: | 13:59 |
| freemangordon | hifi, voice, voice+recording | 14:00 |
| Wizzup | yes we would need another profile for recording | 14:00 |
| Wizzup | I don't know about DAI for voice | 14:00 |
| freemangordon | hmm, we already have DAI | 14:01 |
| freemangordon | see cpcap_set_sysclk() | 14:01 |
| Wizzup | there are two codecs, snd cpcap and motmdm irc | 14:02 |
| Wizzup | iirc | 14:02 |
| Wizzup | see my dynamic debug statemtents | 14:02 |
| Wizzup | bbiab again sry | 14:02 |
| freemangordon | Wizzup: another idea: can we treat call state as a jack? | 14:21 |
| freemangordon | also, see https://kernel.org/doc/html/latest/sound/soc/codec-to-codec.html | 14:41 |
| freemangordon | I thinkt this is what we need, no? | 14:42 |
| freemangordon | *think | 14:42 |
| freemangordon | heh, we don;t have c2c_params in 6.1 :) | 14:56 |
| freemangordon | Wizzup: see https://elixir.bootlin.com/linux/v6.6.37/source/sound/soc/samsung/aries_wm8994.c#L483 | 15:01 |
| Wizzup | freemangordon: do you mean ignore suspend, or? | 15:35 |
| Wizzup | do we not have c2c already? | 15:36 |
| freemangordon | we don't | 15:43 |
| freemangordon | it is not supported in 6.1 | 15:43 |
| Wizzup | oh... | 15:43 |
| Wizzup | ok, that looks like a great thing to try then | 15:43 |
| freemangordon | mhm | 15:44 |
| freemangordon | but, I still lack the audio diagram :) | 15:44 |
| Wizzup | might also be worth mentioning on the mail thread we have with Tony | 15:44 |
| Wizzup | he might be able to help | 15:44 |
| freemangordon | right | 15:44 |
| freemangordon | I'll reply | 15:45 |
| freemangordon | Wizzup: do we have 6.6? | 15:46 |
| freemangordon | well, *where* do we have 6.6? | 15:46 |
| Wizzup | yes and yes | 15:47 |
| Wizzup | freemangordon: maemo-6.6.y | 15:47 |
| freemangordon | ok, lemme try to build it | 15:47 |
| Wizzup | it's in -experimental | 15:48 |
| freemangordon | ok | 15:48 |
| freemangordon | so, no sound at all or only calls don;t work? | 15:48 |
| Wizzup | only calls | 15:48 |
| freemangordon | with 6.6 that is | 15:48 |
| freemangordon | ok | 15:48 |
| freemangordon | both sides? | 15:48 |
| Wizzup | neither sides :) | 15:49 |
| freemangordon | ok | 15:49 |
| freemangordon | ok, building 6.6 | 15:54 |
| freemangordon | Linux devuan-droid4 6.6.36 #2 SMP PREEMPT Fri Jul 5 18:00:45 EEST 2024 armv7l GNU/Linux | 17:10 |
| freemangordon | ok, lets see | 17:10 |
| freemangordon | Wizzup: it seems we will have to switch to audio-graph-card2, see https://elixir.bootlin.com/linux/v6.6.37/source/sound/soc/generic/audio-graph-card2-custom-sample.dtsi | 17:40 |
| freemangordon | not that I understand what is explained there :) | 17:41 |
| inky | Wizzup, thank you! are you sure you have added comics_daily in phoenix for the user norayr? | 19:56 |
| Wizzup | not yet @ jenkins | 19:57 |
| inky | yes, whenever convenient, just reminding. i guess i am ready for it. | 19:59 |
| Wizzup | I will do it today (moving, emptying my entire apartment) | 20:00 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!