| arno11 | Wizzup: i tried 6.6.36 and now call audio works. in fact everything seems to work but something still seems to slowing down Qt5 | 10:28 |
|---|---|---|
| arno11 | conversations is sometimes very slow to launch and use, same for hamsterfiler and calendar. but that's totally random | 10:29 |
| arno11 | for calls, audio works but sphone is unusable: when you receive a call, the caller UI is too slow to appear | 10:31 |
| arno11 | (sometimes it's ok, but again things happen randomly) | 10:31 |
| arno11 | ah, and i got again an error during kernel install: dkms failed to find headers for 6.6 | 10:35 |
| arno11 | (but headers are ok) | 10:35 |
| arno11 | btw i didn't notice any process burning the cpu or using too much RAM | 10:41 |
| arno11 | (*troubles with sphone happen with qtloop, i didn't try without but i'll doublecheck later and i suppose it will be ok without any Qt stuff) | 10:46 |
| arno11 | bbl | 10:54 |
| Wizzup | arno_11: maybe you didn't install/upgrade the header package again | 11:29 |
| arno11 | Wizzup: nope, headers install/upgrade is ok | 12:14 |
| Wizzup | strange... | 12:14 |
| Wizzup | let me upgrade on my phone | 12:14 |
| arno11 | but i found on the web that 6.6 could cause troubles with swap | 12:14 |
| arno11 | and indeed, decreasing swappiness seems to be a workaround | 12:15 |
| arno11 | ok | 12:15 |
| Wizzup | hm, so they didn't solve it yet in 6.6.36 eh | 12:15 |
| Wizzup | you'd assume they would fix these swap issues with stable point releases | 12:16 |
| Wizzup | uvos: yeah so even on 6.6.36 the hotspot doesn't work, but it works on 6.1.y :D | 12:19 |
| Wizzup | arno11: ok I also see the header warning | 12:19 |
| arno11 | ah ok | 12:19 |
| Wizzup | strange | 12:20 |
| arno11 | weird, now with swappiness=10, the N900 seems very stable and fast like 6.1.76 | 12:24 |
| arno11 | swappiness=10 was previously unusable btw, too agressive | 12:25 |
| arno11 | maybe the way it works has been changed again (pourcentage vs agressivity) | 12:25 |
| arno11 | *i mean too agressive on 6.1 or older | 12:27 |
| Wizzup | what was your src on 6.6 changing swap behaviour? | 12:29 |
| Wizzup | keep in mind that on this latest version I also enabled a bunch of debugging stuff | 12:29 |
| Wizzup | I don't think it will really slow down anything (it's just some printks that are optionally enabled) but who knows | 12:29 |
| Wizzup | (but again, it is very unlikely to be the cause) | 12:29 |
| arno11 | ah ok | 12:29 |
| Wizzup | I enabled this so that I can try to debug call audio on d4 | 12:30 |
| Wizzup | brb ~10-15 mins | 12:31 |
| arno11 | ok | 12:31 |
| arno11 | maybe it is worth a try to disable the debugging stuff | 12:47 |
| Wizzup | arno11: it is 100% not that | 12:53 |
| Wizzup | since you saw this also with an earlier build that didn't have it | 12:53 |
| arno11 | ah yes indeed | 12:55 |
| arno11 | so we should have a deeper look @swap imo | 12:56 |
| Wizzup | if a different setting works, is that enough, or? | 12:57 |
| Wizzup | or is maybe zram not available even though you normally use it, or? | 12:57 |
| arno11 | good questions | 12:57 |
| arno11 | swappiness=10 is enough atm but multitasking with heavy apps is impossible | 12:58 |
| arno11 | and for zram, i don't know | 12:58 |
| arno11 | i've never understood what's going on with zram and Leste tbh | 13:00 |
| Wizzup | well, you didn't use it right? | 13:03 |
| Wizzup | it would show up as additional swap them | 13:04 |
| arno11 | ok so it is not activated by default | 13:05 |
| Wizzup | it might be :) | 13:05 |
| Wizzup | I can check in a bit - have to go for a bit again | 13:06 |
| arno11 | ah | 13:06 |
| arno11 | ok no probs | 13:06 |
| arno11 | according to syslog, zram starts on boot and add /dev/zram0 | 13:20 |
| arno11 | but it doesn't appear in fstab | 13:21 |
| arno11 | it appears in /dev/ | 13:22 |
| arno11 | hm, according to irc logs (fmg, uvos), not using zram is normal on N900 | 13:43 |
| arno11 | so i'll rather have a look on possible kernel changes with vm stuff | 13:46 |
| Wizzup | back | 13:58 |
| Wizzup | arno11: ok, well, I want to really try to take another look at call audio and at least write a report to tmlind and others | 13:58 |
| Wizzup | hopefully we can figure it out | 13:59 |
| Wizzup | call audio n d4 that is | 13:59 |
| arno11 | yeah that's definitely a priority | 14:00 |
| arno11 | i'll continue to 'investigate' on swap/vm stuff and LYK | 14:03 |
| Wizzup | ok, | 14:10 |
| arno11 | Wizzup: ok, for 6.6 the culprit is h-s-m again and again...so no new issue. it is just worse | 19:13 |
| arno11 | i mean, usually i killall h-s-m after boot to restart it properly. it doesn't work anymore | 19:14 |
| sicelo | btw hsm is? | 19:14 |
| arno11 | hildon-status-menu | 19:15 |
| sicelo | um, what's wrong with it? | 19:15 |
| arno11 | don't know | 19:15 |
| sicelo | i mean - what do you observe? | 19:15 |
| sicelo | i don't recall having issues with it | 19:16 |
| arno11 | pin entry freeze, availibility or other plugin buttons not visible | 19:16 |
| Wizzup | the pin entry is not part of h-s-m | 19:16 |
| arno11 | it is now according to fmg | 19:17 |
| Wizzup | buttons not being visible means that it crashed prior and loaded only 'safe' plugins | 19:17 |
| arno11 | yep | 19:17 |
| sicelo | yes, i have the pin entry issues. but haven't yet had problems with status | 19:17 |
| Wizzup | I think we just need a core dump to see what happens | 19:17 |
| Wizzup | I think fmg was working on enabling those | 19:17 |
| Wizzup | he should be back tomorrow | 19:17 |
| arno11 | ok | 19:17 |
| sicelo | arno11: what plugins do you have on hsm? | 19:17 |
| sicelo | i have none. | 19:17 |
| arno11 | i just do have FB | 19:18 |
| arno11 | but even without it, it is buggy | 19:18 |
| Wizzup | that's not a hsm plugin fwiw, that's just part of the general status plugin | 19:18 |
| arno11 | oh yes indeed | 19:18 |
| arno11 | i just have original plugins | 19:19 |
| sicelo | i want to work on wpa_supplicant stuff at some point | 19:19 |
| Wizzup | better error messages? :D | 19:20 |
| sicelo | hehe, well first, signal strength | 19:20 |
| Wizzup | wait, this works, doesn't it? | 19:21 |
| sicelo | icd doesn't report correct signal strength - only reports whatever was the strength when the connection was established | 19:21 |
| Wizzup | are you sure? | 19:21 |
| Wizzup | that's very strange, I've seen that number change frequently | 19:21 |
| sicelo | quite sure, yes | 19:21 |
| Wizzup | maybe this broke from beowulf->chimaera | 19:22 |
| sicelo | https://github.com/maemo-leste/bugtracker/issues/730 | 19:23 |
| sicelo | i don't think it ever worked correctly. there are some dbus signals we should subscribe to | 19:23 |
| Wizzup | I've definitely seen that number change in the status applet | 19:24 |
| Wizzup | but I believe you | 19:24 |
| Wizzup | shouldn't be too hard to check I think | 19:24 |
| sicelo | so yes, the number will seem to change because when making a new connectio , the correct strength will ne recorded | 19:24 |
| Wizzup | do you use the wifi signal applet? | 19:24 |
| sicelo | but as long as the connection is established, it won't change | 19:24 |
| sicelo | no, i dont use the applet | 19:24 |
| Wizzup | ok, might be worth checking to see if it works with that just in case | 19:24 |
| Wizzup | in src/wlan.c see | 19:25 |
| sicelo | i think it won't... i looked at its source. i *think* it asks icd | 19:25 |
| Wizzup | static void wlan_statistics(const gchar * network_type, | 19:25 |
| Wizzup | enum icd_nw_levels signal = map_rssi(info->signal); | 19:25 |
| Wizzup | stats_cb(link_stats_cb_token, network_type, network_attrs, network_id, 0, /* time_active */ | 19:25 |
| Wizzup | hm, yeah, ok, it doesn't look like info->signal gets updated | 19:26 |
| sicelo | i meant to update that issue with the signal we should subscribe to, but got sidetracked | 19:27 |
| Wizzup | current_bss_info should contain a new signal though | 19:27 |
| arno11 | Wizzup: so excepting h-s-m troubles we already had before, i don't see any issue with 6.6 atm :) | 19:28 |
| Wizzup | arno11: right, well, there's no audio at all on d4 for calls | 19:28 |
| Wizzup | so that needs to be solved | 19:28 |
| Wizzup | after that we're probably fine | 19:28 |
| Wizzup | sicelo: see get_bss_info in src/wpaicd.c | 19:28 |
| arno11 | oh yes ofc, i meant on n900 | 19:28 |
| Wizzup | _BSS_SIMPLE_INFO_FROM_DICT(&bss_info_dict, "Signal", &info->signal, G_VARIANT_TYPE_INT16, "n") | 19:29 |
| Wizzup | this is where it gets set every time get_bss_info is called | 19:29 |
| sicelo | I'm on android atm ... so i can't easily check | 19:29 |
| Wizzup | I'll just install the applet and test it now, easy enough to remove it afterwards | 19:29 |
| sicelo | yes, good idea | 19:29 |
| Wizzup | I wouldn't have packaged it if it never worked I would hope | 19:30 |
| Wizzup | :D | 19:30 |
| sicelo | that get_bss_info ... does it get called again if connection remains intact? what calls it? | 19:30 |
| sicelo | iirc we're not subscribing to fi.w1.wpa_supplicant1.BSS.PropertiesChanged signal. | 19:31 |
| sicelo | so we don't know when signal strength of currently connected network changes | 19:31 |
| sicelo | anyway, i might be wrong :-) | 19:32 |
| sicelo | i *think* i actually noticed the problem when i installed signal applet (first time ever) | 19:32 |
| Wizzup | well doing that made icd2 use 50-60% cpu | 19:33 |
| Wizzup | it seems to repeatedly issue dbus calls | 19:33 |
| Wizzup | sicelo: yes it gets called frequently | 19:33 |
| Wizzup | at least when statistics are asked for | 19:33 |
| sicelo | ah! what triggers calling it | 19:33 |
| Wizzup | your dbus call I would say | 19:34 |
| Wizzup | and having the conndlgs open | 19:34 |
| Wizzup | and opening the status menu | 19:34 |
| Wizzup | iirc | 19:34 |
| Wizzup | and the wifi signal applet | 19:34 |
| sicelo | ok. I'll retest again. | 19:35 |
| Wizzup | well the wifi signal applet for me just caused havoc | 19:37 |
| Wizzup | it's gone now that I uninstalled it | 19:37 |
| sicelo | hehe, what havoc? 🤪 | 19:38 |
| arno11 | Wizzup: btw did you see my msg about the 3 missing deps for drnoksnes ? | 19:53 |
| arno11 | cause i expect to be able to push drnoksnes in the next couple of days :) | 19:54 |
| Wizzup | arno11: I did see the message | 20:11 |
| Wizzup | do you need me to package them, or just to make the repos and such? | 20:11 |
| arno11 | they are easy to pkg so i can do it if you want, i just need you to make the repos. | 20:15 |
| arno11 | sicelo: your battery patches work, got a fully charge msg + green led | 20:20 |
| arno11 | i mean, we know it works lol but works ootb now with 6.6 | 20:21 |
| arno11 | Wizzup: *i mean empty repos | 20:25 |
| arno11 | bbl | 20:26 |
| Wizzup | arno11: ok, will make it | 20:38 |
| Wizzup | sicelo: icon for otp would be nice btw :D | 20:39 |
| Wizzup | arno11: https://github.com/maemo-leste-extras/x11proto-xsp https://github.com/maemo-leste-extras/libxsp https://github.com/maemo-leste-extras/sdlhaa | 20:40 |
| Wizzup | uvos: even with my regtool I can't get call audio to work | 20:46 |
| Wizzup | uvos: I think it might be somehow dapm related | 20:47 |
| Wizzup | hmm | 20:50 |
| Wizzup | https://bpa.st/6GXA | 20:50 |
| Wizzup | the above is with | 20:51 |
| Wizzup | echo 'file cpcap.c +p' > /sys/kernel/debug/dynamic_debug/control | 20:51 |
| Wizzup | echo 'file motmdm.c +p' > /sys/kernel/debug/dynamic_debug/control | 20:51 |
| Wizzup | echo 'file n_gsm.c +p' > /sys/kernel/debug/dynamic_debug/control | 20:51 |
| Wizzup | echo 'file soc-dapm.c +p' > /sys/kernel/debug/dynamic_debug/control | 20:51 |
| Wizzup | I wonder if the hifi muting interferes with our dapm forcing | 20:51 |
| Wizzup | uvos: I should note that my regtool doesn't necessarily write all regs I think, just a few that I picked | 20:52 |
| Wizzup | uvos: the dapm entries definitely seem *off* | 20:55 |
| Wizzup | very strange to see what when I accept the call the Hifi is unmuted | 20:56 |
| Wizzup | that has to be sphone doing something | 20:57 |
| Wizzup | with 6.6 I seem to regularly lose wifi btw, it just disconnects | 20:58 |
| Wizzup | much like the n900 does frequently | 20:59 |
| arno11 | Wizzup: ty for the 3 repos | 21:14 |
| arno11 | btw i've never seen wifi disconnections on my device | 21:15 |
| arno11 | and weird at hifi unmuted when call is accepted | 21:15 |
| arno11 | seems really wrong | 21:16 |
| arno11 | could be sphone or ucm | 21:16 |
| arno11 | btw what is the call audio sample rate on D4 modem ? | 21:19 |
| sicelo | arno11: nice @ applet | 21:20 |
| arno11 | yep | 21:20 |
| sicelo | Wizzup: @icon ... i'm not good with that kind of stuff, but can try | 21:20 |
| arno11 | ok D4 call audio rate seems 8000 | 16000 | 21:27 |
| arno11 | so if hifi is unmuted and use 44100 | 48000, it is normal to have no audio during calls imo | 21:29 |
| freemangordon | that could be the ringing tone | 21:42 |
| Wizzup | freemangordon: hi | 21:42 |
| freemangordon | there is a race between ringing and call audio it seems | 21:42 |
| Wizzup | freemangordon: based on what? | 21:42 |
| freemangordon | hi | 21:42 |
| freemangordon | gut feeling | 21:42 |
| Wizzup | I am drafting a email, and I also thought that | 21:43 |
| freemangordon | the only thing that unmutes hifi during call is ringing tone | 21:43 |
| Wizzup | let me send it | 21:43 |
| Wizzup | sending | 21:43 |
| Wizzup | sicelo: maybe just pick some generic otp icon, inspired by what android does or something | 21:44 |
| freemangordon | anyway, as I said I just came home from ~1000km driving. tomorrow :) | 21:45 |
| Wizzup | sure :) | 21:45 |
| Wizzup | arno11: hifi is not involved in d4 audio, it's all internal | 21:48 |
| arno11 | i understand that but i suppose you can't use hifi and voice at the same time, or ? | 21:50 |
| Wizzup | don't know | 21:50 |
| Wizzup | my assumption is that our dapm hack is being nullified due to hifi being messed with during call setup | 21:51 |
| freemangordon | Wizzup: try to disable ringiong tone and see if you can repro | 21:56 |
| Wizzup | freemangordon: this is with ringtone disabled | 22:08 |
| Wizzup | profile set to silent | 22:09 |
| Wizzup | can't we make a dapm "Switch" widget hooked up in such a way that the all the call audio components get power? | 22:20 |
| Wizzup | freemangordon: to me it looks like just switching ucm makes the hifi muted but the dapm (or whatever it is) turns it off only after 5 seconds (presumably a timeout) | 22:24 |
| Wizzup | (on "Switch", looks like probably not...) | 22:27 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!