uvos__ | so xorg cpu usage is probubly some x client makeing x do work rather than x doing something by itself | 12:08 |
---|---|---|
uvos__ | for instance if the volume applet where to constantly be getting pa events | 12:08 |
uvos__ | this might cause it to constantly wake up and query some xatom or something like that | 12:08 |
uvos__ | so it would make sense to xtrace all pa clients you have | 12:10 |
uvos__ | se if someone is constantly calling xgetprop or something | 12:10 |
arno11 | uvos: thx | 12:28 |
Wizzup | uvos__: yeah I was thinking also it could be volume applet | 13:08 |
Wizzup | uvos__: in any case if we have 4 PA servers we have other problems | 13:08 |
uvos__ | idealy we should not use system pa at all | 13:21 |
uvos__ | we now have elogind session | 13:21 |
uvos__ | so session pa should work | 13:21 |
uvos__ | should save some non trival amount of ram for n900 too, when no one is playing sound | 13:22 |
Wizzup | in any case I am not sure if this is the problem | 13:22 |
uvos__ | no | 13:22 |
uvos__ | you have other issues | 13:22 |
uvos__ | but just going for the session pa setup might make sense | 13:22 |
uvos__ | vs figureing out why system pa is being launched multiple times | 13:23 |
Wizzup | I don't think the system one is launched multiple times | 13:26 |
Wizzup | it seems to me like cmt_pulse or whatever it is just launches its own | 13:26 |
Wizzup | which probably just means the env is not correct? | 13:26 |
Wizzup | not sure, I would have to check | 13:26 |
Wizzup | uvos__: do you know, how can we know if all regular modules are loaded | 13:31 |
Wizzup | maybe some udev trigger | 13:31 |
Wizzup | or some udev init script that finishes | 13:31 |
Wizzup | the idea being that we need to load nokia-modem as last module | 13:33 |
Wizzup | arno11: is pa still set to do this autospawn thing on your device? | 13:40 |
arno11 | autospawn is deactivated | 13:41 |
arno11 | no change with pa cpu if activated | 13:41 |
arno11 | btw when i kill one of these weird pa daemons i get specific errors from cmtspeech | 13:42 |
arno11 | like pa_get_latency error, sink error, so it seems cmtspeech is creating its own daemons for specific tasks | 13:43 |
Wizzup | do you know why it would do that? or where the code is that does this? | 13:44 |
arno11 | i think it's because cmt_pulse has been split into 3 different progs and each ones are using different PA api commands | 13:47 |
arno11 | my idea is to create a new prog with all code in the same file | 13:48 |
arno11 | like it was before the split. | 13:49 |
Wizzup | hmmmm | 13:49 |
Wizzup | do you know where the code is that doe sit? | 13:49 |
Wizzup | also probably a good idea to ask pavel | 13:49 |
Wizzup | and it could indeed be the volume applet or some other maemo program that acts on the presence of other PA daemons that causes xorg cpu usage | 13:49 |
arno11 | really not sure | 13:50 |
arno11 | yes for Pavel | 13:51 |
arno11 | at the beginning cmtspeech_ofono_test was the main app | 13:52 |
arno11 | and then Pavel split it to different files | 13:53 |
arno11 | to be ale to easely switch between alsa, dsp and pulse | 13:53 |
Wizzup | hmm | 13:53 |
arno11 | *able | 13:54 |
arno11 | according to gh cmtspeech history | 13:54 |
arno11 | and if i understand correctly lol | 13:55 |
Wizzup | mhm | 13:57 |
arno11 | that's a lot of mhm, hmm and mmmhmmm :D | 14:02 |
bencoh | mhm! | 14:03 |
uvos__ | as long as its not mhh | 14:03 |
Wizzup | arno11: heh | 14:04 |
Wizzup | arno11: well I think what I need to do is power up my n900 tonight and see if I can reproduce what you have | 14:05 |
Wizzup | and then see if I can see what's happening with all the pulse servers | 14:05 |
Wizzup | or maybe the x11 cpu usage first | 14:05 |
arno11 | ok, should be easy to reproduce | 14:07 |
arno11 | thx for your help | 14:16 |
arno11 | if you just build calls-test-2 branch and set alternate-sample-rate to 4000 in daemon.conf, it should be enough to reproduce | 14:19 |
Wizzup | ok | 14:20 |
Wizzup | last time I didn't really know who to call :D | 14:20 |
arno11 | lol not easy to test indeed | 14:23 |
bencoh | I usually call my "real" day-to-day cellphone | 14:26 |
Wizzup | yeah but for testing whether there is echo cancellation and not that's painful/difficult | 14:27 |
Wizzup | but I also have done that | 14:27 |
bencoh | oh, that. yeah | 14:27 |
Wizzup | I would also call my landline and pick up using sip, but at some point I worried the neighbours might hear it and get annoyed | 14:27 |
Wizzup | (since I forgot to disconnect the old actual phone) | 14:27 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!