| sicelo | oh, this is read only. maybe likely after all? | 00:00 |
|---|---|---|
| arno11 | unrealistic imo | 00:01 |
| arno11 | testing sd is tricky, too much parameters involved imo | 00:07 |
| arno11 | but again, dd seems pretty accurate compared to other 'safe' tools | 00:11 |
| tmlind | pushed out updated droid4-kexecboot images with battery charging and oops while trying to kexec fixed: https://github.com/tmlind/droid4-kexecboot/commits/master | 06:10 |
| tmlind | we no only umount kernel partition in kexecboot if kexec to the new kernel fails, see https://github.com/tmlind/buildroot/commits/mapphone-kexecboot-2017.11 | 06:11 |
| tmlind | uvos: i did not see the kexecboot up and down keys swapped issue with d4 at least, can you please check with the new images? | 06:14 |
| freemangordon | arno11: I think you didn't do proper experiment: | 08:20 |
| freemangordon | 1. start OMP and verify tehre is sound | 08:21 |
| freemangordon | 2. close OMP | 08:21 |
| freemangordon | 3. call the device and do not pick-up (just hang-up) | 08:21 |
| freemangordon | 4. close sphone | 08:21 |
| freemangordon | 5. start OMP and check if tehre is sound | 08:21 |
| freemangordon | oh, and in few minutes it recovers :) | 08:45 |
| freemangordon | yeah, the issue is not specific to omp | 08:58 |
| freemangordon | mplayer and gst-launch also do not work | 08:58 |
| freemangordon | mpv as well | 08:59 |
| freemangordon | yeah, ok, after a phone call all sounds are suspended for several minutes | 09:03 |
| freemangordon | switching profiles from pavucontrol is ok | 09:13 |
| freemangordon | so either sphone does something wrong or someone else | 09:14 |
| freemangordon | Wizzup: why? https://github.com/maemo-leste/hildon-connectivity-meta/commit/fd2394e8a63bbc73f0cf5d946db7b247ed3c5e3a | 09:17 |
| tmlind | freemangordon: voice calls currently change the routing in the kernel driver, maybe that's the issue? | 09:24 |
| tmlind | for mapphone devices that is | 09:25 |
| freemangordon | ok, switching profile in pavucontrol hifi->call->hifi recovers it | 09:25 |
| freemangordon | tmlind: could be, but what to check? | 09:26 |
| freemangordon | yes, confirmed | 09:27 |
| freemangordon | dunno what sphone is doing, but it breaks sound until I do hifi->call->hifi or until some (unknown) timeout expiores | 09:27 |
| freemangordon | tmlind: there is no voice call active | 09:28 |
| freemangordon | just calling the device (wihtout pick-up) is enough to break all other sounds | 09:28 |
| tmlind | hmm ok, maybe see sound/soc/codecs/motmdm.c for the tdm (time divsion multiplexing) stuff | 09:28 |
| tmlind | sounds like something is not getting restored after a voice call? | 09:29 |
| freemangordon | again, sphone just plays ringing sound without switching profiles (iiuc)\ | 09:29 |
| freemangordon | there is no voice call :) | 09:29 |
| arno11 | freemangordon: i just did exactly what you said and still absolutely no troubles | 09:29 |
| tmlind | yeah but the kernel driver changes routing based on the modem notification | 09:29 |
| freemangordon | like, I am calling the device, but just hang | 09:29 |
| freemangordon | tmlind: ah | 09:30 |
| freemangordon | arno11: yeah, seems like issue with d4 only | 09:30 |
| tmlind | yup incoming call, kernel driver activates the modem tdm | 09:30 |
| freemangordon | tmlind: do you use voice calls on your d4? | 09:30 |
| freemangordon | like, could you confirm that? | 09:30 |
| freemangordon | BTW, by "sound not work" I mean it just pauses the process trying to do playback | 09:31 |
| tmlind | yup i don't use much other audio though | 09:31 |
| tmlind | freemangordon: see motmdm_voice_get_state() | 09:31 |
| freemangordon | wait | 09:31 |
| freemangordon | need some tme to find kernel source :) | 09:31 |
| tmlind | motmdm_enable_primary_dai() gets called on incoming call, and then motmdm_disable_primary_dai() should get called after the call | 09:32 |
| freemangordon | but it is not? | 09:32 |
| tmlind | maybe there's some missing handling in motmdm_voice_get_state() if incoming call is not answered? | 09:33 |
| tmlind | other than disconnected or call failed | 09:33 |
| freemangordon | hmm, lemme pick-up | 09:34 |
| tmlind | yup that's a good test :) | 09:34 |
| freemangordon | yeah, seems that's it :) | 09:35 |
| freemangordon | if I pick-up the call there is no issue it seems | 09:35 |
| tmlind | ok nice, enable n_gsm debug via /sys and see what the missing message might be in dmesg | 09:35 |
| tmlind | modprobe n_gsm debug=0xff | 09:36 |
| tmlind | the missing variant likely is something "1,0,0" like, maybe "1,0,1" as that's missing | 09:38 |
| freemangordon | tmlind: https://pastebin.com/wKmWKi0u | 09:40 |
| freemangordon | seems like 1,0,0 is processed | 09:41 |
| freemangordon | "[46624.579315] VOICE CALL 0" | 09:44 |
| freemangordon | no issue there | 09:44 |
| freemangordon | IIUC | 09:44 |
| tmlind | that's after you don't answer but hangup? | 09:48 |
| freemangordon | yes | 09:48 |
| tmlind | so is the failing test case just don't even hang up just let it ring? | 09:49 |
| freemangordon | no, I hang-up | 09:50 |
| freemangordon | and there is no sound afterworkd | 09:51 |
| tmlind | ok | 09:51 |
| freemangordon | but it seems there is more that one way it breaks | 09:51 |
| freemangordon | 1st variant - there is no sound at all, switching the profiles in pavucontrol fixes it | 09:51 |
| freemangordon | 2nd variant - touch sounds work, but no media sounds | 09:52 |
| freemangordon | like, gstreamer, mplayer, mpv, etc | 09:52 |
| tmlind | sounds like sound/soc/codecs/motmdm.c maybe not restoring state completely? | 09:53 |
| freemangordon | perhaps | 09:54 |
| freemangordon | but how to test? | 09:54 |
| freemangordon | alsactl? | 09:54 |
| tmlind | or actually mcbsp tdm configuration | 09:54 |
| tmlind | yeah alsactl diff before and after? | 09:54 |
| freemangordon | [47095.422180] abe-clkctrl:0030:0: failed to disable | 09:54 |
| freemangordon | is that ok? | 09:54 |
| freemangordon | [44018.005493] omap-mcbsp 40126000.mcbsp: RX Frame Sync Error! | 09:55 |
| freemangordon | and https://pastebin.com/swB6YmK6 | 09:56 |
| freemangordon | I guess none of those are to blame, but still | 09:56 |
| tmlind | the clkctrl errors are ok for devices that have autoclocking capability enabled | 09:56 |
| tmlind | the rx frame sync could be some race in the mcbsp.c configuration for the bus | 09:56 |
| tmlind | like if during a voice call alsa tweaks some settings too? | 09:57 |
| tmlind | sound/soc/ti/omap-mcbsp.c that is, the set_tdm_slot() changes the settings for mcbsp for voice call | 09:59 |
| tmlind | hmm yeah no idea where the %V comes from.. | 10:00 |
| freemangordon | but those should not be the reason for the issue | 10:00 |
| freemangordon | right? | 10:00 |
| tmlind | right, sounds like you have this reproducable | 10:01 |
| freemangordon | yes | 10:01 |
| freemangordon | happens every time | 10:01 |
| freemangordon | ok, I see nothing suspicious in alsamixer | 10:02 |
| freemangordon | maybe we have issues in both pulseaudio and the driver :) | 10:03 |
| freemangordon | oh, maybe the frequency is changed | 10:04 |
| tmlind | so in omap-mcbsp.c we set up multiple dais because of the tdm, maybe that's racy between the voice call and audio dais | 10:06 |
| freemangordon | ok, starting mafw-gst-renderer makes gst produce no sound :) | 10:07 |
| freemangordon | that's total mess | 10:07 |
| freemangordon | ok, restarting pulseaudio makes it work again :( | 10:34 |
| freemangordon | tmlind: arno11: if anyone has an idea wtf is going on: https://pastebin.com/ | 10:44 |
| freemangordon | this is what happens on hangup | 10:44 |
| arno11 | plz update the link :P | 10:48 |
| freemangordon | ugh: "You have exceeded the maximum size of 512 kilobytes per Paste." | 10:49 |
| freemangordon | arno11: https://pastebin.com/PTsEpNkt | 10:50 |
| freemangordon | arno11: "I: [alsa-sink-40124000.mcbsp-dai0-cpcap-hifi cpcap-hifi-0] alsa-sink.c: Scheduling delay of 0.56 ms > 0.25 ms, you might want to investigate this to improve latency..." repeats till ethernity | 10:51 |
| freemangordon | with various values for "delay of 0.xx ms" | 10:51 |
| freemangordon | tmlind: ^^^ | 10:51 |
| arno11 | freemangordon: could you plz have a look in daemon.conf and check what's the default resample method ? | 11:00 |
| freemangordon | arno11: /etc/pulse/daemon.conf? | 11:01 |
| arno11 | yes | 11:01 |
| freemangordon | nothing in there | 11:01 |
| arno11 | wow | 11:02 |
| freemangordon | arno11: https://pastebin.com/5x1UneUK | 11:02 |
| arno11 | ah speex-float-1 | 11:03 |
| freemangordon | yeah, but isn't that commented out? | 11:03 |
| freemangordon | we have ; at the beginning of the line | 11:03 |
| arno11 | yes so that's the default value | 11:04 |
| freemangordon | ok | 11:04 |
| arno11 | so i think we must try to switch to speex-fixed-2 | 11:05 |
| arno11 | and use 48000hz as default sample rate | 11:05 |
| freemangordon | I think it does already | 11:05 |
| freemangordon | I: [pulseaudio] resampler.c: Speex appears to be compiled with --enable-fixed-point. Switching to a fixed-point resampler because it should be faster. | 11:06 |
| freemangordon | D: [pulseaudio] resampler.c: rate 24000 -> 48000 (method speex-fixed-1) | 11:06 |
| arno11 | yes so that's why we should switch to fixed resampler by default | 11:06 |
| freemangordon | I think the issue happens after "Mhmm, actually there is nothing to rewind." | 11:07 |
| freemangordon | then it spits hundreds of those messages per seconfd | 11:07 |
| arno11 | ok but according to errors, something is wrong with the sink, then source-sink synchro is bad then we have latency issues. | 11:13 |
| arno11 | then everything goes wrong | 11:14 |
| arno11 | i think we should try a basic loopback to see if it works | 11:14 |
| freemangordon | I must admit I have no clue about PA | 11:18 |
| freemangordon | so, if you want me to try something in particular, I can do it | 11:19 |
| freemangordon | but that's all :) | 11:19 |
| arno11 | and definitely check resampling stuff (it could crash PA) | 11:19 |
| arno11 | ah ok | 11:19 |
| arno11 | i have to go now but will be available in 1h | 11:19 |
| freemangordon | ok | 11:19 |
| freemangordon | maybe in the evening | 11:20 |
| arno11 | yes if you want | 11:20 |
| freemangordon | yep | 11:20 |
| arno11 | ok | 11:20 |
| Wizzup | freemangordon: just synced devel, I think for some pkg we didn't have kernel headers at the time | 12:12 |
| Wizzup | freemangordon: we can revert and see if things break again | 12:15 |
| freemangordon | Wizzup: you mean this? https://github.com/maemo-leste/hildon-connectivity-meta/commit/fd2394e8a63bbc73f0cf5d946db7b247ed3c5e3a | 12:37 |
| Wizzup | hmh | 13:00 |
| freemangordon | I think this shall be reverted | 13:01 |
| rafael2k | hey, which piece of software does the opengl magic, rotation control, etc? | 13:12 |
| rafael2k | need yelp | 13:14 |
| rafael2k | the lock screen works beatifully | 13:14 |
| freemangordon | hildon-desktop | 13:16 |
| freemangordon | which uses clutter | 13:16 |
| freemangordon | to do gl stuff | 13:16 |
| freemangordon | it takes the resolution from xrandr | 13:16 |
| rafael2k | this one right: https://github.com/maemo-leste/hildon-desktop | 13:49 |
| rafael2k | but xrandr is giving correct resolution... | 13:49 |
| rafael2k | if there a way to get the stdout of it? | 13:50 |
| rafael2k | I'll take a look, will try to understand why it is wrong | 13:58 |
| freemangordon | rafael2k: yes, you can | 14:35 |
| freemangordon | /usr/sbin/dsmetool -k /usr/bin/hildon-desktop, to stop it | 14:35 |
| freemangordon | and then use maemo-summoner to start /usr/bin/hildon-desktop.launch | 14:36 |
| freemangordon | send SIGUSR1 to it to enable debug output | 14:37 |
| freemangordon | maybe you'll need G_MESSAGES_DEBUG=all when you start it | 14:37 |
| arno11 | Wizzup: (just to understand) you moved all stuff from devel to stable, right ? | 17:11 |
| Wizzup | mosly everything | 17:12 |
| Wizzup | mostly | 17:12 |
| arno11 | ok. that's 2 different kernels in use, how it works for droid4linux stuff ? | 17:13 |
| Wizzup | what do you mean with 2 different kernels in use | 17:14 |
| arno11 | devel is 6.1.47 and stable is 6.1.9, no ? | 17:15 |
| arno11 | *6.1.67 | 17:15 |
| Wizzup | https://maedevu.maemo.org/pkgweb/search?q=linux-image-omap | 17:16 |
| Wizzup | devel is Version: 6.1.67.1-1+m7 | 17:16 |
| Wizzup | looks like stable is still Version: 6.1.8.0-1+m7 | 17:16 |
| arno11 | yes | 17:16 |
| Wizzup | hm, yeah, so maybe I didn't do that somehow | 17:16 |
| Wizzup | repodiff doesn't mention it | 17:16 |
| Wizzup | weird | 17:16 |
| arno11 | ah | 17:16 |
| Wizzup | $ python3 repodiff.py chimaera-devel chimaera -t newer -f source | 17:16 |
| Wizzup | Packages in chimaera-devel that are newer than chimaera | 17:16 |
| Wizzup | charge-mode 1.3.1+m7 (> 1.2.6+m7 ) | 17:16 |
| Wizzup | python-mafw 0.2+m7.1 (> 0.2+m7 ) | 17:16 |
| Wizzup | sphone 0.8.0+m7 (> 0.7.5+m7 ) | 17:16 |
| Wizzup | hildon-connectivity-meta 1.10+m7.1 (> 1.10+m7 ) | 17:16 |
| Wizzup | xorg-server 2:1.20.11.4-1+m7 (> 2:1.20.11.2-1+m7.1 ) | 17:16 |
| Wizzup | will debug a bit later | 17:16 |
| arno11 | ok | 17:17 |
| arno11 | on fresh img stable is 6.1.9 btw | 17:18 |
| arno11 | freemangordon: your kernel is 6.1.67 on your d4 ? | 17:23 |
| freemangordon | ъес | 17:43 |
| freemangordon | oops | 17:43 |
| freemangordon | arno11: yes | 17:43 |
| arno11 | ok cool because 6.1.y stuff has been merged with 6.1.67, not with 6.1.8 (or 9) so i suppose we'll have troubles with stable | 17:45 |
| freemangordon | umm, can;t parse | 17:45 |
| freemangordon | 6.1.y is upstream branch | 17:45 |
| freemangordon | we do our patches on top | 17:45 |
| freemangordon | so, why should we have troubles? | 17:46 |
| freemangordon | ah, you mean because stable is on older kernel? | 17:46 |
| arno11 | no newer kernel | 17:46 |
| freemangordon | well, Wizzup said will debug later | 17:47 |
| freemangordon | I guess he'll fix that | 17:47 |
| arno11 | yep | 17:47 |
| freemangordon | do you have som etime now to try to debug PA issues? | 17:47 |
| arno11 | yes i think | 17:47 |
| freemangordon | ok, what do you want me to do? | 17:48 |
| arno11 | i think we can try to simulate a call with a loopback to see if it works | 17:49 |
| freemangordon | but why call? | 17:49 |
| arno11 | arecord -c 2 -r 24000 -f S16_LE - | aplay - | 17:49 |
| freemangordon | I don't pick-up the call | 17:49 |
| arno11 | with voicecall profile enable in pavucontrol | 17:50 |
| freemangordon | ok, but it is not enabled when the issue happens | 17:50 |
| arno11 | anyway | 17:50 |
| freemangordon | ok | 17:50 |
| freemangordon | will try | 17:50 |
| freemangordon | for how long? | 17:52 |
| arno11 | if you talk, you should hear your voice in speaker with around 1sec latency | 17:52 |
| arno11 | just try few seconds to test if sound works | 17:53 |
| freemangordon | no | 17:53 |
| freemangordon | I hear nothing | 17:53 |
| arno11 | ah | 17:53 |
| arno11 | try with 44100 instead of 24000 | 17:54 |
| freemangordon | ummm | 17:54 |
| freemangordon | wait | 17:54 |
| freemangordon | it works with hifi profile | 17:54 |
| freemangordon | but not with call | 17:54 |
| arno11 | oh | 17:54 |
| arno11 | so ucm has troubles lol | 17:54 |
| freemangordon | but phone calls work | 17:55 |
| arno11 | ok | 17:55 |
| freemangordon | lemme try with 44100 | 17:55 |
| arno11 | ok | 17:55 |
| freemangordon | nope | 17:56 |
| freemangordon | lemme try phone call to confirm | 17:56 |
| arno11 | ok | 17:56 |
| freemangordon | umm, no | 17:58 |
| arno11 | what is weird is that your mic works with hifi profile and apparently not with voicecall | 17:59 |
| freemangordon | ok, I stoppped pulseaudio and started it from the shell, could that be the reason? | 17:59 |
| arno11 | i doubt | 18:00 |
| freemangordon | ok, now music plays | 18:00 |
| arno11 | could you check during a call, in alsamixer, if mic is activated ? | 18:01 |
| freemangordon | ok | 18:01 |
| arno11 | and speaker/earphone | 18:01 |
| freemangordon | 3 mics are enabled | 18:02 |
| arno11 | ok | 18:02 |
| freemangordon | Mic1, Mic2 and Call | 18:02 |
| arno11 | and speaker/earphone ? | 18:02 |
| freemangordon | Call is enabled | 18:03 |
| freemangordon | HiFi is enabled | 18:03 |
| freemangordon | and Voice is enabled | 18:03 |
| freemangordon | well, sliders are > 0 (around 30-40) | 18:03 |
| freemangordon | not sure if they are enabled really | 18:03 |
| arno11 | ok let me check in ucm | 18:04 |
| freemangordon | shall I rollback the kernel? | 18:06 |
| arno11 | nope | 18:06 |
| arno11 | could you check (using hifi profile) if mics are still enabled during music playback | 18:08 |
| freemangordon | all mics ar at 0 | 18:09 |
| arno11 | ok | 18:09 |
| arno11 | so how is it possible to do a loopback using the hifi profile...weird | 18:10 |
| freemangordon | I see microphone when hifi profile is enabled | 18:11 |
| arno11 | you mean activated ? | 18:11 |
| freemangordon | yes | 18:12 |
| freemangordon | do you want me to do some pa dump? | 18:12 |
| arno11 | no i think the problem is with ucm | 18:12 |
| arno11 | because what you see is a bit different that what ucm is supposed to do on mapphone | 18:13 |
| freemangordon | https://pastebin.com/c1xEQ2P5 | 18:13 |
| arno11 | plz check files/folders in /usr/share/alsa/ucm2 | 18:14 |
| freemangordon | umm, why do we have earpiece in hifi profile? | 18:16 |
| freemangordon | https://github.com/maemo-leste/leste-config/blob/master/leste-config-mapphone/usr/share/alsa/ucm2/Mapphone_Audio/HiFi.conf.leste#L50 | 18:16 |
| freemangordon | arno11: what do you want me to check there? | 18:16 |
| arno11 | i just want to check if ucm path is still correct | 18:17 |
| freemangordon | /usr/share/alsa/ucm2/Mapphone_Audio | 18:18 |
| arno11 | for hifi profile and earpiece, i don't know | 18:18 |
| freemangordon | so the above dir contains 3 files and 3 links | 18:18 |
| arno11 | ok path is correct | 18:18 |
| freemangordon | files are dated 15 of December | 18:19 |
| freemangordon | ok, shall I reboot the device to check it behaves the same on fresh boot? | 18:19 |
| arno11 | yes | 18:20 |
| freemangordon | ok | 18:20 |
| freemangordon | on fresh boot it works | 18:24 |
| freemangordon | lemme try loopback | 18:24 |
| arno11 | ok | 18:24 |
| arno11 | loopback is supposed to not work with hifi profile) | 18:25 |
| arno11 | only with voicecall | 18:25 |
| freemangordon | no | 18:25 |
| freemangordon | it does not work with call profie | 18:26 |
| arno11 | wtf | 18:26 |
| arno11 | and works with hifi i suppose | 18:26 |
| freemangordon | nope | 18:26 |
| freemangordon | now it does not work at all | 18:27 |
| arno11 | ah | 18:27 |
| arno11 | and real calls ? | 18:27 |
| freemangordon | thats' after a phone call | 18:27 |
| arno11 | ah ok | 18:27 |
| freemangordon | call worked | 18:27 |
| arno11 | ok | 18:27 |
| arno11 | and if you switch profile twice, it works ? | 18:27 |
| freemangordon | it? | 18:27 |
| arno11 | i mean voicecall to hifi to voicecall | 18:28 |
| freemangordon | underrun!!! (at least 16000.183 ms long) | 18:28 |
| arno11 | for loopback | 18:28 |
| freemangordon | this is the result from loopback | 18:28 |
| arno11 | wow | 18:29 |
| freemangordon | and after a timeout it started to work | 18:29 |
| freemangordon | now loopback works with hifi | 18:29 |
| freemangordon | lemme switch to call | 18:29 |
| arno11 | ok | 18:29 |
| freemangordon | nope | 18:30 |
| freemangordon | I guess I should hear loopback from the earpiece, right? | 18:30 |
| arno11 | yes | 18:31 |
| freemangordon | notheing | 18:31 |
| freemangordon | *nothing | 18:31 |
| arno11 | are you ok to modify daemon.conf a bit ? | 18:31 |
| freemangordon | sure | 18:31 |
| freemangordon | I can modify it as much as you tell me to :) | 18:32 |
| arno11 | ok :) so plz replace 44100 default sample rate by 48000 | 18:32 |
| arno11 | and the contrary for alternate sample rate | 18:33 |
| arno11 | i mean exchange the 2 values | 18:33 |
| freemangordon | ok | 18:33 |
| arno11 | and replace speex-float-1 by speex-fixed-2 | 18:34 |
| arno11 | then you need to reboot | 18:34 |
| freemangordon | can I restart pulseaudio instead? | 18:35 |
| freemangordon | no difference | 18:37 |
| freemangordon | arno11: I think the approach is wrong | 18:38 |
| freemangordon | afaik d4 does not route mic through cpu when in call | 18:38 |
| freemangordon | the modem does some magic instead | 18:38 |
| freemangordon | and mic is routed directly to the modedm | 18:38 |
| freemangordon | *modem | 18:38 |
| Wizzup | yes | 18:39 |
| Wizzup | well the modem doesn't do magic, the board does | 18:39 |
| Wizzup | I think | 18:39 |
| freemangordon | well... | 18:39 |
| freemangordon | still | 18:39 |
| arno11 | ok but how could it be enable during a loopback | 18:39 |
| freemangordon | we can;t | 18:39 |
| freemangordon | because it does not go through cpu | 18:39 |
| arno11 | so something is really wrong | 18:39 |
| freemangordon | arno11: agree, but the issue is with fifi profile | 18:40 |
| freemangordon | *hifi | 18:40 |
| arno11 | yes so ucm is involved | 18:40 |
| freemangordon | but not only it seems | 18:40 |
| arno11 | and the high latency means something is wrong with PA too | 18:40 |
| freemangordon | modem driver does some things as well | 18:40 |
| arno11 | yes | 18:41 |
| freemangordon | PA could just wait for some kernel event | 18:41 |
| freemangordon | which never comes | 18:41 |
| arno11 | indeed | 18:41 |
| arno11 | plz reboot, i really want to see if errors are different with different daemon.conf parameters | 18:43 |
| freemangordon | arno11: wait, please explain a bit what are you trying to achieve | 18:44 |
| freemangordon | we dont; have issues with the calls it seems | 18:45 |
| arno11 | according to errors, something is wrong with sink when you hangup | 18:45 |
| freemangordon | whch sink? | 18:45 |
| freemangordon | *which | 18:45 |
| arno11 | then we have resampling errors and high latency | 18:46 |
| arno11 | for sink, no idea | 18:46 |
| freemangordon | IIUC, what happens is: after hangup, sphone creates notification that seems to be played over voice profile | 18:47 |
| arno11 | ok | 18:47 |
| freemangordon | but voice profile is not really enabled | 18:47 |
| arno11 | so ucm sucks | 18:48 |
| arno11 | but why | 18:48 |
| freemangordon | not really, it is the kernel that calls motmdm_enable_primary_dai() | 18:48 |
| arno11 | ah yes | 18:48 |
| freemangordon | see what tmlind said ^^^ | 18:48 |
| freemangordon | so, PA thinks HiFi profile is enabled | 18:48 |
| freemangordon | but in reality it is some frankenstein profile | 18:49 |
| arno11 | kind of lol | 18:49 |
| arno11 | with no d4 around, that's not easy to help you :( | 18:50 |
| freemangordon | tmlind: why motmdm has to do what it does, in terms of motmdm_enable_primary_dai()/ motmdm_disable_primary_dai()? | 18:50 |
| freemangordon | like, on incomming call, motmdm_enable_primary_dai() is called unconditionally, but we want ringing through the speaker | 18:51 |
| freemangordon | isn;t there some conflict? | 18:51 |
| freemangordon | like, why shall we call enable on "1,4,0"? | 18:53 |
| arno11 | but why a conflict now and not few days/weeks ago ? | 18:55 |
| arno11 | what has been changed recently ? | 18:55 |
| freemangordon | are you sure there was no issue back then? | 18:55 |
| arno11 | nope | 18:56 |
| freemangordon | we have mafw/OMP from few weeks | 18:56 |
| arno11 | we must ask d4 users :) | 18:56 |
| freemangordon | Wizzup: said there is no issue | 18:56 |
| arno11 | and PA user mode | 18:56 |
| freemangordon | that should make no diff | 18:56 |
| arno11 | indeed | 18:56 |
| freemangordon | I will rather try to not enable dai on incomming call | 18:57 |
| freemangordon | but after it is connected | 18:57 |
| arno11 | maybe you should rollback the kernel afterall, just to be sure | 19:06 |
| freemangordon | will wait for tmlind first :) | 19:07 |
| tmlind | freemangordon: i think for voice calls the modem is the mcbsp bus master rather than the soc clocking the bus | 19:09 |
| freemangordon | tmlind: ok, but what will happen if we call motmdm_enable_primary_dai() and then someone tries to play audio? | 19:10 |
| freemangordon | because that's what happens on call reject, IIUC | 19:11 |
| freemangordon | brb, dinner | 19:11 |
| tmlind | i think one of the two soc audio dais works voice or hifi, one gets disabled, might be worth tracing what android does there | 19:11 |
| sicelo | may i briefly interrupt you gentlemen - what are the values of EVICGABS in https://gitlab.freedesktop.org/hadess/iio-sensor-proxy/-/blob/7a12f6bbc91e372a4aa0bc1344531dca7239a016/tests/input-accel-capture.ioctl and how could I generate/find them myself? | 19:12 |
| sicelo | *EVIOCGABS | 19:12 |
| sicelo | i need them because upstream wants tests for the N900's proximity sensor/switch support in iio-sensor-proxy, and i don't really have a clue how to do it :-P | 19:14 |
| tmlind | freemangordon: i think hifi should still be functional during a voice call, see https://lkml.org/lkml/2018/3/27/1225 | 19:15 |
| freemangordon | hmm, ok, so we have 2 independent DAIs | 19:26 |
| freemangordon | tmlind: ok, but enable/disable reprogram master DAI | 20:34 |
| freemangordon | so that must affect hifi profile, no? | 20:35 |
| freemangordon | it is still functional, just with wrong frequency | 20:35 |
| tmlind | hmm should not affect hifi, they're on separate mcbsp controllers | 20:36 |
| tmlind | unless there's a bug somewhere | 20:37 |
| tmlind | maybe a bug in the cpcap driver that touches wrong registers affecting also hifi? | 20:38 |
| freemangordon | snd_soc_dai_set_sysclk(ddata->master_dai, 1, 19200000, SND_SOC_CLOCK_OUT); in enable and snd_soc_dai_set_sysclk(ddata->master_dai, 0, 26000000, SND_SOC_CLOCK_OUT); in disable | 20:38 |
| freemangordon | this does not affect hifi profile? | 20:38 |
| freemangordon | also, is it ok that clk_ids differ? | 20:39 |
| freemangordon | I am shooting in the dark, but still :) | 20:40 |
| tmlind | should not affect hifi but maybe i'm wrong, when a voice call is active we set the modem to be the bus master, then change back after the call | 20:46 |
| freemangordon | but, at the point of "incomming call', there is still no call | 20:46 |
| tmlind | yeah i guess it could be changed a bit later | 20:46 |
| freemangordon | I am going to try it | 20:47 |
| tmlind | ok | 20:47 |
| freemangordon | and do not enable on 1,4,0 | 20:47 |
| tmlind | also see the dts files for mcbsp and cpcap | 20:48 |
| freemangordon | hmm, why? | 20:48 |
| tmlind | hifi and voice are separate instances at mcbsp and and cpcap dtsi | 20:48 |
| freemangordon | ok, but what I see in dmesg is: | 20:49 |
| freemangordon | mot-mdm6600-codec 4806a000.serial:modem:audio-codec@2: Master DAI is cpcap-codec.0 | 20:49 |
| freemangordon | anyway, let me try the modified driver | 20:49 |
| tmlind | ok could be the audio driver is structured wrong if some parts are not separate | 20:50 |
| tmlind | for voice and hifi | 20:50 |
| freemangordon | tmlind: this https://pastebin.com/Q5ZRMwMn fixes the issue | 21:00 |
| freemangordon | and for the first time I heard "missed call notification" sound | 21:00 |
| freemangordon | no idea what the conclusion would be, but to me we have issue in the modem driver | 21:01 |
| freemangordon | also, why kernel enables the dai and not userspace through some also control? | 21:01 |
| freemangordon | *alsa control | 21:02 |
| freemangordon | isn't that supposed to be part of ucm? | 21:05 |
| freemangordon | ah, where is uvos | 21:05 |
| freemangordon | ok, I think what happens here is that kernel changes dai frequencies and/or rate behind PA's back | 21:17 |
| tmlind | freemangordon: ok great. sure could be alsa too, not sure where though, ofono maybe? not sure if ofono knows anythign about alsa | 21:17 |
| freemangordon | it is sphone that switches the profiles | 21:17 |
| freemangordon | but only after you pick-up the call | 21:17 |
| freemangordon | not on reject | 21:17 |
| tmlind | not generic enough imo, there could be other phone clients like a command line one in ofono | 21:18 |
| freemangordon | and immediately after the reject sphone issues "missed call" notification | 21:18 |
| freemangordon | which hits the still reprogrammed for call $something | 21:18 |
| tmlind | might be worth checking if there's some standard way of doing this nowadays | 21:19 |
| freemangordon | all this are wild guesses ofc | 21:19 |
| freemangordon | but my gut feeling tells me I am rgiht | 21:19 |
| freemangordon | *right | 21:19 |
| tmlind | sure, just don't make it limited to sphone | 21:20 |
| freemangordon | ah, sure | 21:20 |
| freemangordon | I am not blaming sphone here | 21:20 |
| freemangordon | it just triggers the bug | 21:20 |
| tmlind | right, but what i'm talking about is alsa controlling the tdm stuff | 21:20 |
| freemangordon | ok, I admit it is out of my area of expertise | 21:21 |
| freemangordon | I know what TDM is, but tht's all :) | 21:21 |
| tmlind | so all the phone using clients would have to implement alsa controls and some knowledge of the modem? | 21:21 |
| freemangordon | ah, I see | 21:21 |
| tmlind | maybe ofono or modemmanager whatever have already sorted this issue out | 21:21 |
| freemangordon | how to check? | 21:22 |
| tmlind | grep alsa | 21:22 |
| tmlind | :) | 21:22 |
| freemangordon | heh | 21:22 |
| tmlind | in ofono and modemmanager | 21:22 |
| tmlind | anyways, it's time for ttyl here, later | 21:23 |
| freemangordon | later | 21:23 |
| sicelo | afaik both ofono and modemmanager have chosen to leave the audio alone, so unlikely to be them | 21:24 |
| sicelo | ofono does provide a CallAudio interface, but that's just a notification for an audio client to start handling routing. | 21:25 |
| sicelo | 'modern' distros handle the routing via `callaudiod` | 21:25 |
| sicelo | and i think the long-term plan/hope is for pipewire to transparently handle it | 21:27 |
| freemangordon | ok, but isn;t ucm supposed to do that? | 21:28 |
| sicelo | yes, callaudiod basically toggles ucm | 21:29 |
| freemangordon | ok, so it is what sphone already does | 21:30 |
| freemangordon | the point is that we ahve some driver issue here | 21:30 |
| sicelo | i agree | 21:30 |
| freemangordon | driver does things without being instructed by userspace | 21:30 |
| freemangordon | and without letting userspace know, iiuc | 21:30 |
| apoage | hello, i would like to ask, trying to get runing maemo leste on droid4 by maemo leste wiki but failing to get kexec menu ... is unlocked bootloader necesary to get it runing?? since till now i had there safestrap to hijack boot process | 22:50 |
| apoage | ¯\_(ツ)_/¯ | 23:01 |
| Wizzup | apoage2: hmmm | 23:05 |
| Wizzup | argh | 23:05 |
| Wizzup | apoage: check your android version | 23:05 |
| Wizzup | you don't need to unlock the bootloader | 23:05 |
| Wizzup | check our wiki page on the supported versions | 23:05 |
| Wizzup | you might need to upgrade/downgrdae | 23:05 |
| Wizzup | upgrade/downgrade | 23:06 |
| Wizzup | either that or the image isn't flashed correctly to the microsd | 23:06 |
| Wizzup | apoage2: pls check irc log | 23:07 |
| Wizzup | but check the wiki page and see if your android version is one of the supported ones | 23:07 |
| Wizzup | and/or check if you dd'd the image correctly | 23:08 |
| Wizzup | (you need to decompress it first) | 23:08 |
| apoage2 | Wizzup: ok | 23:09 |
| apoage2 | Wizzup:First I would like to apologize for my irc client.. had to go to phone and it's well kind of unstable app.. the thing is I ran completely by wiki except android install script I ran manually since was on windows mashine but at first nor second droid boot there was no kexec menu.. will try different SD and complete process under Linux.. had to ask for that unlocking because I have Verizon droid 4 and it's kind of hard locked | 23:25 |
| sicelo | no worries, and yes, there's only the Verizon Droid 4. that's what all of us have ;-) | 23:27 |
| apoage2 | Ok so no hard stops.. thank you every one | 23:30 |
| sicelo | and you can use https://libera.irclog.whitequark.org/maemo-leste/ to see any replies to your questions that you might have missed at any time | 23:30 |
| apoage2 | Sicelo: thanks.. really appreciate that log | 23:32 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!