| *** halfline has quit IRC | 00:23 | |
| *** halfline has joined #automotive | 00:39 | |
| smurray | gkiagia: one more thing to look at in the morning, when uncorking AFAICT pwaudiosink is not sending GST_MESSAGE_REQUEST_STATE at all, so we don't know we're uncorked | 01:00 |
|---|---|---|
| smurray | gkiagia: I see that on both what's in halibut atm and the backported latest PW | 01:01 |
| *** User3209 has joined #automotive | 01:23 | |
| *** RP has quit IRC | 01:24 | |
| *** User3209 has quit IRC | 01:43 | |
| *** mranostay has quit IRC | 01:55 | |
| *** LiXiaoming has joined #automotive | 02:08 | |
| *** buspirate has quit IRC | 02:33 | |
| *** buspirate has joined #automotive | 02:33 | |
| *** syborg has quit IRC | 03:35 | |
| *** phc has quit IRC | 05:17 | |
| *** Newami has joined #automotive | 06:16 | |
| *** Newami has quit IRC | 06:31 | |
| *** vrubiolo has joined #automotive | 07:24 | |
| LiXiaoming | dl9pf: Pi 4 & Icefish RC3, Bt connect to Android is unstable, is it a know issue? | 08:01 |
| LiXiaoming | Dec 15 23:41:22 raspberrypi4 bluetoothd[331]: Agent /org/bluez/agent333 replied with an error: org.freedesktop.DBus.Error.UnknownMethod, No such method “RequestPinCode” | 08:01 |
| LiXiaoming | [/w/workspace/release-jjb-icefish-snapshot/MACHINE/raspberrypi4/label/agl-test-slave/repoclone/output/tmp/work/aarch64-agl-linux/agl-service-bluetooth-map/1.0+gitAUTOINC+012d6bf1cb-r0/git/binding/bluetooth-map-api.c:885,map_create_session] | 08:01 |
| LiXiaoming | org.freedesktop.DBus.Error.UnknownMethod, No such method “RequestPinCode” | 08:02 |
| LiXiaoming | [/w/workspace/release-jjb-icefish-snapshot/MACHINE/raspberrypi4/label/agl-test-slave/repoclone/output/tmp/work/aarch64-agl-linux/agl-service-bluetooth-map/1.0+gitAUTOINC+012d6bf1cb-r0/git/binding/bluetooth-map-api.c:885,map_create_session] | 08:02 |
| *** vrubiolo has quit IRC | 08:04 | |
| *** Newami has joined #automotive | 08:04 | |
| *** Newami has quit IRC | 08:06 | |
| *** leon-anavi has joined #automotive | 08:08 | |
| leon-anavi | hi | 08:09 |
| *** kbroulik has joined #automotive | 08:17 | |
| *** RP has joined #automotive | 08:22 | |
| *** skz81 has joined #automotive | 08:34 | |
| *** vrubiolo has joined #automotive | 08:44 | |
| *** jaillet has joined #automotive | 08:49 | |
| *** [AD]Turbo has joined #automotive | 08:51 | |
| *** toscalix has joined #automotive | 08:54 | |
| *** [AD]Turbo has left #automotive | 09:08 | |
| *** toscalix has quit IRC | 09:15 | |
| *** toscalix has joined #automotive | 09:38 | |
| *** toscalix has quit IRC | 10:11 | |
| *** LiXiaoming has quit IRC | 10:17 | |
| *** LiXiaoming has joined #automotive | 10:17 | |
| *** kooltux_ has joined #automotive | 10:29 | |
| LiXiaoming | dl9pf: do we have a icmp configure change in icefish RC3? I can't ping or ssh to a board from my host pc. RC1 is OK. | 10:33 |
| *** toscalix has joined #automotive | 10:45 | |
| *** toscalix has quit IRC | 10:46 | |
| *** LiXiaoming has quit IRC | 11:08 | |
| *** toscalix has joined #automotive | 11:24 | |
| *** toscalix has quit IRC | 11:35 | |
| *** vrubiolo has quit IRC | 11:35 | |
| *** toscalix has joined #automotive | 11:37 | |
| *** toscalix has quit IRC | 11:42 | |
| dl9pf | no | 11:56 |
| *** gkiagia has quit IRC | 13:23 | |
| *** vrubiolo has joined #automotive | 13:52 | |
| smurray | gkiagia: so the delays when radio gets uncorked seem related to the alsasrc and/or pwaudiosink being unhappy: https://www.irccloud.com/pastebin/BP4pCTfx/ | 13:57 |
| smurray | doh, he's not here | 13:57 |
| dl9pf | kooltux: appfw call ? is jose availabe ? | 14:09 |
| *** skz81_ has joined #automotive | 14:38 | |
| *** skz81 has quit IRC | 14:39 | |
| *** skz81 has joined #automotive | 14:56 | |
| *** skz81_ has quit IRC | 14:57 | |
| *** gkiagia has joined #automotive | 15:15 | |
| gkiagia | I built the tip of master now, it all looks good to me | 15:20 |
| dl9pf | gkiagia: I'm on halibut + your wireplumber update. | 15:21 |
| dl9pf | observations: usb soundcard is not preferred over built-in | 15:22 |
| gkiagia | did you build the last rev that I uploaded ~1.30 h ago? | 15:22 |
| gkiagia | there was a bug with priorities, it's fixed now | 15:23 |
| dl9pf | ok, good point, need to check ... probably I'm still rev1 | 15:23 |
| dl9pf | updated, let me try. | 15:24 |
| smurray | gkiagia: re the delays when the radio gets uncorked after having been corked for a while, it seems related to alsasrc and/or pwaudiosink being unhappy: https://www.irccloud.com/pastebin/jbYnPeDU/ | 15:26 |
| gkiagia | smurray: I was about to test that now, I've only tried rtl-sdr so far | 15:27 |
| smurray | gkiagia: I'm wondering if maybe some state isn't being reset on un-pause in pwaudiosink? | 15:27 |
| smurray | gkiagia: also, mentioned it here last night, I'm pretty sure GST_MESSAGE_REQUEST_STATE (with state GST_STATE_PLAYING) isn't being sent on uncork | 15:28 |
| gkiagia | smurray: how do you test the "uncork" exactly ? | 15:32 |
| gkiagia | so, with kf radio, this is what I see: initial playback starts fast; if I switch to the media player and then pause the media player, I get back to radio fast; if I stop the radio and try to start it again, I get no audio | 15:33 |
| smurray | gkiagia: I start the radio, then go to mediaplayer and start it, wait a bit, then stop mediaplayer. If it's a few seconds, radio starts right away, if it's longer, it delays for minutes, then those messages get printed | 15:33 |
| smurray | gkiagia: let it sit for a couple of minutes, it will start, it seems the pipeline is in a bad state... | 15:34 |
| gkiagia | hmm, ok, let me check the code | 15:34 |
| *** gkiagia has quit IRC | 15:46 | |
| *** gkiagia has joined #automotive | 15:51 | |
| gkiagia | ah I see what the problem is | 15:51 |
| gkiagia | the code should handle GST_MESSAGE_REQUEST_STATE and actually change the state to PAUSED/PLAYING | 15:52 |
| gkiagia | because alsasrc is a live element, which means it runs on its own clock | 15:52 |
| gkiagia | so the pipeline keeps going, but pwaudiosink blocks and the queue fills up and then alsasrc blocks and the timestamps are in a bad state | 15:53 |
| gkiagia | it's generally ok if you don't handle this message, except in the case where the pipeline is live (which is the case here) | 15:54 |
| gkiagia | smurray: I can fix it | 15:55 |
| smurray | gkiagia: ah, I'd figured that pwaudiosink was changing to PAUSED itself, and that would propagate up the pipeline... | 15:56 |
| gkiagia | no, the elements are not supposed to change the state of the whole pipeline, which is why GST_MESSAGE_REQUEST_STATE exists in the first place... it's for the elements to ask the app to change the state of the pipeline | 15:58 |
| smurray | gkiagia: ah, okay, I'd figured it was informational. I'm not a gstreamer expert ;) | 15:59 |
| smurray | gkiagia: and it "just worked" before | 15:59 |
| gkiagia | it is only used in pulsesink, though, so it's almost synonymous to the corking mechanism... although I think we might need something more advanced in the future, in order to deal with more advanced policy | 15:59 |
| gkiagia | smurray: also, afaict, there is no GMainLoop running in the radio binding... so this is not going to work: gst_bus_add_watch(gst_element_get_bus(pipeline), (GstBusFunc) handle_message, NULL); | 16:04 |
| gkiagia | that's probably why you don't see the cork message | 16:04 |
| smurray | gkiagia: yeah, I have a change to add that, currently slightly blocked by a power outage here | 16:04 |
| gkiagia | ah good | 16:04 |
| smurray | gkiagia: with it, I do see the REQUEST_STATE with PAUSED, but not one with PLAYING, same in both mediaplayer and radio | 16:05 |
| smurray | gkiagia: which is problematic since I was hoping to use that to ignore steering wheel button events when corked | 16:06 |
| dl9pf | there is some UI blockage when BT connection active as well | 16:09 |
| dl9pf | need to check for logs | 16:09 |
| gkiagia | smurray: well, it works in gst-launch | 16:10 |
| gkiagia | stop everything, start this in a terminal: gst-launch-1.0 audiotestsrc ! pwaudiosink stream-properties="p,media.role=Multimedia" -v | 16:11 |
| gkiagia | then start the media player from the UI... gst-launch will say: Setting state to PAUSED as requested by /GstPipeline:pipeline0/GstPwAudioSink:pwaudiosink0... | 16:11 |
| gkiagia | then stop the media player... gst-launch will say: Setting state to PLAYING as requested by /GstPipeline:pipeline0/GstPwAudioSink:pwaudiosink0... | 16:11 |
| smurray | maybe it doesn't trigger if it's not been paused? | 16:12 |
| gkiagia | ah, hmm.. could be, I've never checked that scenario | 16:12 |
| smurray | I can hack in whacking the pipeline and test once I have power back here | 16:12 |
| gkiagia | gst-launch handles it like this: https://gitlab.freedesktop.org/gstreamer/gstreamer/blob/master/tools/gst-launch.c#L809 | 16:13 |
| smurray | isn't there a some potential for issues with WP policy here? If move the pipeline to PAUSED state, and then something else goes to play over top on the same role, won't it be in a bad way due to the endpoint still being present? | 16:14 |
| smurray | might take 3 users of a role to get there, maybe | 16:14 |
| dl9pf | gkiagia: with your latest wireplumber, any bluetooth connection will freeze the UI (if not in mediaplayer) | 16:19 |
| smurray | like the whole UI? homescreen, etc? | 16:25 |
| dl9pf | yes | 16:34 |
| dl9pf | will unfreeze as soon as i kill the bt connection on the phone | 16:34 |
| dl9pf | cannot switch apps or do anything | 16:34 |
| gkiagia | wtf | 16:35 |
| gkiagia | testing | 16:35 |
| dl9pf | env here atm: rpi4, halibut+latest wireplumber (v2)+moto g6 | 16:36 |
| smurray | cough rpi4 cough | 16:37 |
| dl9pf | yes | 16:37 |
| dl9pf | if it does not show that elsewhere, fine | 16:38 |
| *** vrubiolo has quit IRC | 16:38 | |
| smurray | dl9pf: estimated restoration time for power here is "Under review", which doesn't sound promising | 16:42 |
| gkiagia | I cannot reproduce on h3, but I am on master | 16:43 |
| gkiagia | ah you said not in mediaplayer, hmm | 16:44 |
| gkiagia | still, it works fine | 16:46 |
| gkiagia | I'll try halibut again | 16:46 |
| dl9pf | building master to compare | 16:47 |
| dl9pf | lol ... yes, 'Under review' sounds quite promising ... much like 'maybe this year' | 16:47 |
| *** kbroulik has quit IRC | 16:48 | |
| dl9pf | wrt pi3/pi4: slow/crappy hardware can show issues masked by the powerfull cpu's - we've seen this with the gpu memory that also applies to m3 when under virtualization ... | 16:50 |
| smurray | dl9pf: we've seen some issues on and off where scanning in the BT settings seems to bog down, but not when connected | 16:51 |
| gkiagia | I can check pw/wp memory usage, but I doubt that would be an issue | 16:52 |
| waltminer | gkiagia can you be on the SAT call tomorrow? Also, we are planning a final CES integration session the weekend of Jan 4-5 in Vegas. Would you be available on-line at least part of those days? | 16:54 |
| dl9pf | don't think its the memory. | 16:55 |
| smurray | dl9pf: heh, now saying 2 more hours, finally | 16:55 |
| smurray | I knew I should have made coffee earlier ;) | 16:56 |
| waltminer | smurray just forwarded you proposed icon designs. Do you have a fav? | 16:56 |
| gkiagia | waltminer: I will be on the SAT, yes | 16:57 |
| waltminer | do you at least have heat? | 16:57 |
| waltminer | thanks gkiagia | 16:57 |
| gkiagia | waltminer: I don' t have plans yet for that weekend, so I suppose I can arrange to be partly available | 16:57 |
| smurray | waltminer: top right hand corner one for lane departure, 3rd one down in the leftmost column for cruise control | 16:58 |
| waltminer | Great. I would guess the Sunday is more important because Scott will not arrive in Vegas until late your time. So if you can be available on Sunday I would really appreciate it | 16:58 |
| smurray | waltminer: those come closest to matching the steering wheel buttons imo | 16:58 |
| waltminer | ok. Will let her know | 16:59 |
| gkiagia | ok | 16:59 |
| *** jaillet has quit IRC | 17:24 | |
| *** skz81 has quit IRC | 17:49 | |
| *** leon-anavi has quit IRC | 17:56 | |
| dl9pf | ui blocking is also on master - can play some choppy music over bt. | 18:01 |
| dl9pf | well, if it is only pi4 ... let's add to known issues | 18:01 |
| *** RzR has joined #automotive | 18:03 | |
| *** kooltux_ has quit IRC | 18:14 | |
| smurray | heh, they're toying with me, power was back on for a couple of min, then back off | 18:18 |
| dl9pf | likely overloaded somewhere | 18:18 |
| mdp | smurray: PG&E take over your grid? | 18:55 |
| *** vrubiolo has joined #automotive | 19:59 | |
| fury | Can anyone recommend a good Bluetooth headset that supports wide band speech (WBS / 16khz)? | 20:59 |
| fury | The last one I had died and took the company that made it with it (MadCatz..May they rest in pieces) | 20:59 |
| dl9pf | smurray: settings ui looks fine after bump | 21:17 |
| dl9pf | changed your commit to use the new srcrev as well. | 21:17 |
| smurray | dl9pf: okay. I was going to do that separately, but works just as well | 21:18 |
| dl9pf | will merge the bump and then yours, so we have it aligned. | 21:18 |
| dl9pf | at least for master ... halibut needs the whole libqtappfw stuff cherry-picked if we want it. | 21:19 |
| dl9pf | and probably more | 21:19 |
| smurray | Yeah, think it’d libqtappfw, vshl, and settings | 21:21 |
| *** vrubiolo has quit IRC | 21:27 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!