Sunday, 2023-03-26

T42<k1gen> every time after last 10-15 reboots phone freezes 2-4 seconds after lockscreen appears, no ssh over usb or wlan. I can turn the screen off, but not back on07:32
T42<k1gen> ~20 reboots with and without usb attached and it booted fine07:45
T42<k1gen> mal: pulseaudio stuff:
T42<k1gen> I think I found the error: E: [pulseaudio] core-util.c: Assertion 's' failed at ../src/pulsecore/core-util.c:2186, function pa_atoi(). Aborting.07:48
T42<k1gen> should I comment out both service audioserver audioserver_HYBRIS_DISABLED and service /vendor/bin/hw/ in /usr/libexec/droid-hybris/system/etc/init/disabled_services.rc for sound to work?10:26
*** poetaster is now known as Guest891311:07
T42<k1gen> well, I left sound for now, but I almost got modem working. I have LTE connectivity, APN is set as it was on android excluding default,supl in APN type line (there's no such line in sailfish), and everything seems to be working fine - but there's no internet access, I can't load websites11:51
T42<k1gen> still, device is becoming more usable every day - all I have to get working is sound, finish modem (I'm sure there's just some little thing I missed), camera (which is working, but only once per session, and no video), and vibro - and I will be able to use it everyday!12:04
mal@k1gen please show all pulseaudio message from journal12:39
T42<k1gen> mal: devel-su journalctl -b --no-pager | grep pulseaudio?12:44
T42<k1gen> I'm asking because pulseaudio is a user service, but journalctl without devel-su returns nothing12:49
T42<elros34> it matters only when you restart service for journal logs you always must be root12:52
T42<k1gen> thanks12:53
T42<k1gen> mal:
T42<k1gen> mal: is not enough?14:01
T42<k1gen> *is it14:04
T42<elros34> how do you think? it's basically one line of log14:05
T42<k1gen> it's all I have, I didn't remove some parts of log or something14:06
T42<elros34> stop service, check what commandit use to start pulseaudio and start it manually with additional -v (or more vvv argument)14:06
T42<k1gen> ok14:06
T42<elros34> '| grep' for sure removes parts of log14:07
T42<elros34> no, this is not how process start normally, see pulseaudio.service and all arguments14:58
T42<k1gen> thanks, I'll get back in an hour14:58
T42<elros34> systemctl-user cat pulseaudio.service  then you see it's mysterious $CONFIG which is hidden in EnvironmentFile=15:00
T42<k1gen> thanks for saving me 10 minutes of googling :)15:00
T42<k1gen> I'm back:
T42<k1gen> I'm sorry, forgot the -vvv:
T42<elros34> show /vendor/etc/audio_policy_configuration.xml and add more vvv16:16
T42<k1gen> thanks for saying exactly what to do, I appreciate it16:18
T42<elros34> copy that xml end remove maxActiveCount prop, then bind mount edited file over /vendor one and see if that helps16:21
T42<elros34> oh I see common copy-paste mistake in pulseaudio code: pa_atoi(max_open_count instead pa_atoi(max_active_count16:24
T42<elros34> jusa, mal^16:24
T42<k1gen> @elros34 how do I bind mount a file? mount -o ro,bind /vendor/etc/audio_policy_configuration.xml /odm/etc/audio_policy_configuration.xml ?16:26
T42<elros34> no, you cant change anything in vendor so you must copy it to writable directory like /etc for example and bind mount that to /vendor one. odm has nothing to it16:28
T42<elros34> unless you want to just put modified file in odm. Maybe that will work if parser will stop after first file but I am not sure, you must test16:29
T42<k1gen> so just remove /odm from my command?16:30
T42<elros34> do you understand mount bind command?16:31
T42<elros34> copy /vendor/etc/audio/audio_policy_configuration.xml to /odm/etc/audio_policy_configuration.xml, edit /odm/etc/audio_policy_configuration.xml and than try again if that will not work then will will try bind mount16:32
T42<k1gen> thank you. here's the output:
T42<elros34> ok so it will work but it must be bind mounted. Remove it from /odm then copy and edit file to /etc, then: mount -o bind /etc/audio_policy_configuration.xml /vendor/etc/audio_policy_configuration.xml16:37
T42<k1gen> thank you very much for providing commands! here's the output after following instructions:
mal@elros34 interesting bug, I can fix that tomorrow17:35
mal@k1gen do you have gdb installed on your device, would be good to get backtrace of the crash, gdb --args pulseaudio and then the parameters17:36
T42<k1gen> sure, let me install it17:37
T42<k1gen> it's already installed, and I've ran gdb --args pulseaudio --daemonize=no -n --file=/etc/pulse/ -vvvvvvvvvv17:38
T42<k1gen> should I install pulseaudio-debuginfo and pulseaudio-debugsource as gdb suggests?17:39
T42<k1gen> I'm in gdb17:42
maltype "bt"17:43
malwait first "run"17:43
malthen once it crashes type bt17:43
T42<k1gen> got it, wait for log17:44
T42<k1gen> mal:
malah, use this and get new backtrace: HYBRIS_USE_VENDOR_NAMESPACE=yes gdb --args pulseaudio --daemonize=no -n --file=/etc/pulse/ -vvvvvvvvvv17:49
T42<k1gen> sorry forgot to export this after ressh17:50
T42<k1gen> how do you guys read these?17:56
malnot sure yet why that one variable is null which then causes the crash18:28
T42<elros34> about your non gdb log: try: 'find / -iname *primary*so'. Do you see 64bit version?18:28
mal@elros34 that gdb log shows Loaded hw module audio.primary (flame)18:29
T42<elros34> ok18:34
T42<b100dian> what does it mean when init-debug lights up an orange led about 42 seconds in? (and how does it do that?)18:36
T42<elros34> init script doesnt touch led at all but that could be usefull feature sometimes18:38
T42<k1gen> @elros34 on *primary*so: yes, four 64-bit ones and four 32-bit18:53
T42<k1gen> mal: so forget about sound for now? any ideas on that lte problem I have?18:56
T42<elros34> I wonder if pulseaudio-modules-droid-jb2q could be used in 18 base18:57
T42<k1gen> I followed hadk-hot on that this time :D18:58
T42<k1gen> by the way, I still haven't enabled, as in today's message: should I comment out both service audioserver audioserver_HYBRIS_DISABLED and service /vendor/bin/hw/ in /usr/libexec/droid-hybris/system/etc/init/disabled_services.rc for sound to work?18:59
T42<elros34> it's disabled on purpose so why would enable it19:00
T42<elros34> didn't you just said you have 64bit primary audio library?19:02
T42<k1gen> ah, I thought it's related to pulseaudio-modules-droid >= android 11 line and not 32-bit stuff19:04
mal@k1gen what lte issue?19:35
T42<k1gen> let me copypaste the message: well, I left sound for now, but I almost got modem working. I have LTE connectivity, APN is set as it was on android excluding default,supl in APN type line (there's no such line in sailfish), and everything seems to be working fine - but there's no internet access, I can't load websites19:36
T42<k1gen> you weren't online 8 hours ago I guess19:37
T42<k1gen> I'd be happy to provide some logs, just tell me what to run :)19:39
malcheck journal log for messages from ofono19:54
mal@elros34 I wonder if that "Ignore profile with no supported channels" is bad thing in pulseaudio output20:00
T42<k1gen> mal: full journalctl after reboot:
T42<k1gen> 68 mentions of ofono20:10
T42<k1gen> mal: calls in and out work, though there's no sound20:16
T42<k1gen> damn, even Wi-Fi hotspot works - I'm sending it from device, connected to pixel20:19
T42<k1gen> no ideas?20:24
T42<k1gen> there's a lot of20:29
T42<k1gen> Activating context: 120:29
T42<k1gen> Unexpected data call status 410020:29
T42<k1gen> [ofono] ERROR! GDBus.Error:org.ofono.Error.Failed: Operation failed20:29
T42<k1gen> in journalctl20:29
maldo you have dummy_netd installed and running?20:29
T42<k1gen> no, I don't20:31
T42<k1gen> now I do, but cellular network still won't work. do I need to reboot?20:35
T42<k1gen> mal: thanks, after reboot everything works!20:38
malgood, remember to add that dummy_netd to patterns20:47
malwhat other things are broken than audio?20:48
T42<k1gen> camera works only once - after boot, then crash and black screen, flashlight also works only once per boot session, vibration doesn't work, as well as autobrightness20:51
malis sensorfw running?20:51
T42<k1gen> yeah, /usr/sbin/sensorfwd -c=/etc/sensorfw/primaryuse.conf --systemd --log-level=warning --no-magnetometer-bg-calibration20:52
maldo the sensors work in csd?20:54
T42<k1gen> csd? the Components gui app?20:54
T42<elros34> no, sensorfw starts but doesn't really work, there are no sensors detected20:55
malwell maybe it's easier if you install messwerk app20:55
T42<k1gen> I lanched csd from fingerterm, it won't work over ssh obviously, that's why I asked20:56
malthere is a way to access it from ui also, it's just hidden20:56
T42<elros34> why it wouldn't work from ssh?20:56
T42<k1gen> @elros34 error: XDG_RUNTIME_DIR not set in the environment.20:57
T42<k1gen> [C] unknown:0 - Failed to create display (No such file or directory)20:57
malmaybe you were as root?20:57
T42<elros34> so don't start apps as root20:57
T42<elros34> it's not X20:57
T42<k1gen> got it, sorry20:57
T42<k1gen> sensors don't work in csd btw20:57
T42<elros34> but there is no point, clearly sensorfw doesn't work20:57
malsomething is wrong, the messages "Waiting for service 'vold' on '/dev/binder'..." look bad20:59
T42<k1gen> it's a bummer mal is online and active when I'm going to rest :(20:59
malit's midnight here so I probably won't be active much longer21:00
T42<k1gen> it's midnight for me too actually21:01
T42<k1gen> goodnight then, hope we'll continue in the morning/day21:02
mal@k1gen there are two mount services still failing which you probably should fix just in case, metadata.mount and mnt-vendor-persist.mount21:08
T42<k1gen> thanks for noticing, I'll get to them in the morning21:08
malalso show the defconfig you use, just in case the ohmd failure is because of some missing config21:09
T42<elros34> probably missing namespace (NS) options from mer-kernel-check21:12
malyes, that is my guess21:12
T42<k1gen> mal:
T42<elros34> .config or zcat /proc/config.gz not defconfig file21:15
T42<elros34> but clearly you are missing all
T42<k1gen> oh, sorry, a little sleepy. a moment21:16
malipc and pid are there but the other two are not21:17
T42<elros34> no it's just me who prefer always .config because people think adding something to defconfig always enables it21:17
T42<elros34> see how it's different all of them are enabled anyway except CONFIG_USER_NS21:22
T42<k1gen> but USER_NS is not specified here
T42<elros34> no, so not sure this will fix it21:26
malyeah, could be some other issue21:26
T42<k1gen> well, gotta try anyway. I'll send journal from kernel with USER_NS later21:27
T42<k1gen> thank you for suggestions21:28
malabout vibrator issue, check if you have any paths like /sys/class/leds/vibrator* on your device21:29
T42<k1gen> I do, but I don't know how to select and copy in fingerterm21:31
malah, your device appears to have ffmemless, that requires some changes, can you install mce-tools and show output of "evdev_trace -i"21:31
malbut if there is some path like I mentioned we should investigate which one is really used21:32
T42<k1gen> I'm sorry, is it fine if I'll screenshot the output? I really can't find a way to copy text and don't have access to my laptop right now21:33
T42<elros34> it's in menu21:34
T42<k1gen> mal: I have qpnp_pon ob event0, gpio-keys on event1 and fts on event221:37

Generated by 2.17.1 by Marius Gedminas - find it at!