Tuesday, 2022-05-17

T42<edp_17> @elros34 : Thanks.05:27
T42<edp_17> Here are the logs: dmesg: https://paste.ubuntu.com/p/qvVJG4Thtc/05:27
T42<edp_17> journalctl: https://paste.ubuntu.com/p/gBbJGK7Wxc/05:27
T42<edp_17> logcat: https://paste.ubuntu.com/p/fsf7P4kpT6/05:27
T42<edp_17> All of these are with unmodified droid-hal-init.service and dsme.service. (It means there are no commented lines in them.)05:29
T42<edp_17> Interesting. When I amended dsme.service to this, the device started to reboot quicker than I could get in: https://paste.ubuntu.com/p/kX4tftbWZM/06:21
T42<edp_17> Then I modified droid-hal-init.service to this but the same remained, too quick reboot: https://paste.ubuntu.com/p/XCtmR5THsw/06:22
T42<edp_17> No, I am masking droid-hal-init, to see what happens.06:22
T42<elros34> so nothing really crash in dmesg except dsme: drivers/net/loopback.c:205 but only once, in journal dsme timeouts after first crash. droid-hal-init finished. What happend after that ~30s in log? About dsme.service changes: I think you only need PrivateNetwork=false to prevent first crash. Did you enable namespace configs recommended by mer-kernel-check?07:24
T42<TheVancedGamer> USER_NS is broken on Samsung devices afail07:27
T42<TheVancedGamer> USER_NS is broken on Samsung devices afaik (edited)07:27
T42<edp_17> @elros34 : This is mer-kernel-check. What namespace configs were you referring to? https://paste.ubuntu.com/p/x3RpNHsm7j/07:30
T42<edp_17> I'll amend dsme.service as you suggested and try.07:30
T42<elros34> these at the ennd of mer_verify_kernel_config, maybe also this mentioned by @TheVancedGamer07:33
T42<elros34> to be clear: defined at the end of file not at the end of your output which suggest that you have them enabled assuming your are using up-to-date mer-kernel-check07:35
T42<edp_17> I am on this commit in hybris-17.1 branch (23 March 2022): fab7e5cd0b611c3c30fd3d182e77c09e40cd8aac07:38
T42<edp_17> I have a quite big journal log now starting at ay 17 07:26:43 and ending at May 17 08:34:11. Could you please let me know what time range you are interested? (I cannot upload all because it is too large.)07:40
T42<elros34> latest boot usually or if device reboots and you couldn't telnet then previous boot (-b -1). Previously I was talking about ~30s of dmesg. You should also really ask details about "USER_NS is broken on Samsung.."07:46
T42<edp_17> Here is the journal since last reboot. I'll get dmesg and check that flag in kernel. https://paste.ubuntu.com/p/f6VSHJVjXN/07:53
T42<edp_17> Oh, and I've amended dsme.service to this but since, there is no gui. (droid-hal-init is masked and started manually, line by line) https://paste.ubuntu.com/p/yZgNRVFv7v/07:55
T42<edp_17> The gui might be started, as now when I switched the device off, the home screen flashed up.07:56
T42<edp_17> Here are some dmesg outputs: https://paste.ubuntu.com/p/XC3cfjD86T/08:00
T42<elros34> you need to figure out from logs what cause device reboot so do not mask everything. First it was probably dsme but it doesn't fail now so search further. Next would be to figure out why many services fails (maybe namespaces) then qpa-hwcomper plugin errors to get gui working08:00
T42<edp_17> Hard to collect because the connection now interrupts very often.08:00
T42<TheVancedGamer> IMO, I think everything is failing because it relies on some namespaces broken in every Samsung kernel08:01
T42<TheVancedGamer> network namespaces are broken badly as well, along with user namespaces08:01
T42<edp_17> @TheVancedGamer : Since what sfos version this namespace required? sfos 4.2 works well on my Galaxy Note 4 and sfos 4.3 works well on my Galaxy S2.08:02
T42<TheVancedGamer> unless your device is a >2019 one08:02
T42<elros34> so the easiest test would be to disable USER_NS if that is even possible08:02
T42<TheVancedGamer> I think sailjail depends on it08:03
T42<TheVancedGamer> @elros34 : I think its not even enabled08:03
T42<TheVancedGamer> because then the kernel wouldn't compile08:03
T42<edp_17> @elros34 : With this dsme https://paste.ubuntu.com/p/kX4tftbWZM/ and masked droid-hal-init, I can get gui and many stuff working when I amnually start every line from /usr/bin/droid/droid-hal-startup.sh08:03
T42<elros34> @TheVancedGamer  ah. you can always get rid of sailjail and speed up you device.08:04
T42<edp_17> Yeah, USER_NS is not defined in defconfig.08:04
T42<TheVancedGamer> If everything manually works and makes droid-hal-init boots, I think its a race condition08:05
T42<elros34> @edp17 but you do not want to do it every time right? so let it reboot and figure out why08:05
T42<TheVancedGamer> @edp_17 : do make <name of defconfig>, and find USER_NS in .config08:05
T42<TheVancedGamer> if its disabled or =n, that's fine08:05
T42<elros34> at least 1 samsung device didn't even like opeing framebuffer in wrong time of boot sequence. If you will check dmesg/journal then you will see why it reboots08:06
T42<TheVancedGamer> Unmask droid-hal-init and stick in some sleep's08:07
T42<elros34> hostnamed also crash in same place in loopback.c:20508:21
T42<edp_17> No, especially because it requires me to telnet in to do it. :) (re @elros34: @edp17 but you do no...)08:29
T42<edp_17> @elros34 : Interesting. In defconfig CONFIG_USER_NS is not defined, but in .config it is set to Y.08:46
T42<edp_17> @TheVancedGamer : I was also thinking about putting some sleeps into droid-hal-init but am not sure how much. WIll try sleep 45 at the beginning.08:49
T42<TheVancedGamer> @edp_17 : sleep 5 should be enough in the script, not the systemd service08:52
T42<edp_17> Yeah, that's how I meant. In /usr/bin/droid/droid-hal-startup.sh08:53
T42<TheVancedGamer> alright08:54
T42<TheVancedGamer> Also, try adding some debugging in droid-hal08:54
T42<TheVancedGamer> Like `set -xe`08:54
T42<edp_17> How do you mean?08:55
T42<TheVancedGamer> append `set -xe` to the start of droid-hal-startup.sh09:00
T42<edp_17> Will do. With sleep 5 at the beginning of /usr/bin/droid/droid-hal-startup.sh and droid-hal-init.service unmasked, the devive keeps rebooting so quickly, I cannot telnet in. Try increase sleep to 30.09:01
T42<edp_17> @TheVancedGamer : Do you mean replace this "ExecStart=/bin/sh /usr/bin/droid/droid-hal-startup.sh" with this "ExecStart=/bin/sh -xe /usr/bin/droid/droid-hal-startup.sh" ?09:16
T42<TheVancedGamer> Yep09:17
T42<TheVancedGamer> Then unmask it09:17
T42<TheVancedGamer> and then reboot09:17
T42<TheVancedGamer> if you can get into telnet, check journal, it'll have way more logs09:18
T42<edp_17> @TheVancedGamer : My droid-hal-init.service looks like this now: https://paste.ubuntu.com/p/XCtmR5THsw/09:20
T42<edp_17> Should I uncomment the commented lines?09:21
T42<edp_17> I've uncommented all thise stuff but this way droid-hal-init crashes: https://paste.ubuntu.com/p/KQpRNd6HXR/10:51
T42<edp_17> *those10:51
T42<edp_17> Even when I restart droid-hal-inti.service, it fails.10:57
T42<TheVancedGamer> @edp_17 : lipstick fails a lot, can you mask it and then send journal again?11:11
T42<edp_17> I need to change stuff back as already started to experiment. Give me  a few minutes.11:16
T42<edp_17> @TheVancedGamer : Here it is. Lipstick masked: https://paste.ubuntu.com/p/XCmHJXwqcm/11:23
T42<edp_17> Ops, I forgot to add the -xe. Let me try again.11:24
T42<edp_17> Oh, and there is a sleep 30 in /usr/bin/droid/droid-hal-startup.sh11:24
T42<TheVancedGamer> yeah, I saw that11:24
T42<TheVancedGamer> add -xe back and get me journal again11:24
T42<edp_17> Should I keep the sleep 30? In log there is "droid-hal-init.service: Failed with result 'timeout'." Is this not because of that sleep 30?11:26
T42<TheVancedGamer> trim it down to sleep 511:26
T42<edp_17> Will do. Until here is another log with sleep 30 but -xe added: https://paste.ubuntu.com/p/nWJw4Yv5hK/11:29
T42<edp_17> Well, with sleep 5, the device keeps rebooting so quickly that I cannot telnet in.11:34
T42<TheVancedGamer> alright then11:50
T42<TheVancedGamer> @edp_17 : add back -xe, and then add `&> /var/log/tmp.log` to the Exec line for droid-hal11:51
T42<TheVancedGamer> and add back sleep 3011:51
T42<edp_17> @TheVancedGamer : Added -xe, sleep 30, lipstick masked : https://paste.ubuntu.com/p/kn4sWGDHdW/12:25
T42<edp_17> There is no /var/log/tmp.log file12:25
T42<edp_17> Have tried to add "&> /var/log/tmp.log" and then just "> /var/log/tmp.log" to the end of /usr/bin/droid/droid-hal-startup.sh but none of them created the tmp.log file.12:26
piggzanyone had an issue with call audio ... mic on phone is working, can hear in earpeice, but not being sent to other end ... incoming audio is ok16:28
T42<adampigg> @Verevka86 You had this?16:29
T42<adampigg> @b100dian you too? was on android 11?16:32
T42<b100dian> was on android 10, and it is some race when I am not disabling all android audio services. Make sure you actually have `override` in your disabled-services.rc, some may still start if they are parsed before.16:34
T42<adampigg> id better see whatts disabled ..this port is different, i havnt disabled any services yet! :D16:35
T42<adampigg> does /vendor/bin/hw/android.hardware.audio.service need disabled?16:36
T42<b100dian> `service audioserver /system/bin/audioserver` and `service vendor.audio-hal-2-0 /vendor/bin/hw/android.hardware.audio@2.0-service` need to be disabled16:37
T42<adampigg> i dont have them, but will try audio.service16:38
T42<b100dian> It's closest in naming, yeah16:40
T42<b100dian> is this the port with no .git folders:)?16:42
T42<adampigg> yes ... its using halium for the android side, with no DHD packages16:43
T42<adampigg> the exact same image also (nearly) worked with the volla2216:44
T42<adampigg> so, i quite like how easy it could be to bring up a device16:44
T42<eugenio_g7> piggz: update your halium gsi, see this: https://github.com/Halium/android_device_halium_halium_arm64/pull/1416:44
T42<eugenio_g7> welcome to the future ^^16:45
deathmistmal: what's holding up https://github.com/mer-hybris/android/pull/154 and the other related PRs?17:01
maldeathmist: good question, one of those is not approved17:04
deathmistmal: the same should be done for 18.1 as well, but I've not seen those repos change at all since last year17:05
malThaodan: ping, do you want to approve this https://github.com/mer-hybris/android/pull/15517:07
T42<adampigg> @eugenio_g7 you should also disable timedaemon, afaik its not needed, and very noisy in logs17:08
T42<Verevka86> https://github.com/SailfishOS-vayu/droid-config-vayu/commit/64da8821159aedef59c94f52c9faa3091596e2b017:13
T42<Verevka86> The microphone works when calling only in the bluetooth headset for me17:13
T42<Verevka86> ^ (re @adampigg: @Verevka86 You had t...)17:14
T42<Verevka86> I have a problem that not all sounds are played, the sound of the keyboard, music, video work, but the system sounds do not, the sound of the dialer, alarm clock and incoming ringtone does not work (although it works when choosing a melody in the settings), has anyone met this?17:25
malanything interesting looking in logs?17:35
malfrom pulseaudio or ohm17:35
T42<adampigg> @eugenio_g7 @NotKit thats fixed it :) #portcomplete17:37
T42<b100dian> @Verevka86 smells like ngfd crash. Do you have vibration?18:06
T42<Verevka86> No (re @b100dian: @Verevka86 smells li...)18:07
T42<b100dian> systemctl-user status ngfd18:07
T42<Verevka86> https://irc.thaodan.de/.imgstore/957d0277/_20220517_001.png18:13
T42<b100dian> Ok, so this is why you don't have audio for notifications, the non-graphical feedback daemon is not working. You need to debug that. It might be related to vibration (I had a crash at upload of FF_RUMBLE effect), but see what your crash is. You can run ngfd manually from defaultuser with -vvv18:16
malmaybe you are missing this change https://github.com/mer-hybris/droid-config-sony-seine/commit/1a806cead790fbdf607cbc3566cf3e876aeb68b118:21
mal@Verevka86 ^18:22
T42<b100dian> That's useful for Qt feedback based on ngfd, +1. But you still need to make ngfd start18:24
T42<A_T_R> Hi, Version is the latest sailfish os?18:24
T42<b100dian> or, if you wait a bit.. .18:26
T42<A_T_R> But hadk is ,why there is no hadk for the higher version18:27
malhadk version is not really critical that way18:28
T42<A_T_R> But hadk is exist on pinned,why there is no hadk for the higher version (edited)18:28
T42<Verevka86> https://paste.ubuntu.com/p/xfYYhz8DP3/ (re @b100dian: That's useful for Qt...)18:29
mal@Verevka86 do you see anything in journal about ngfd?18:29
maldo you use ffmemless?18:29
T42<Verevka86> Thanks, I'll try it (re @SailfishFreenodeIRCBridgeBot: <mal>maybe you are m...)18:29
T42<Verevka86> https://github.com/SailfishOS-vayu/droid-config-vayu/blob/master/patterns/patterns-sailfish-device-adaptation-vayu.inc#L34 (re @SailfishFreenodeIRCBridgeBot: <mal>do you use ffme...)18:30
T42<Verevka86> No18:30
T42<b100dian> > default effect load failed18:31
T42<b100dian> sounds familiar. Can you show kernel sources?18:31
T42<Verevka86> https://github.com/SailfishOS-vayu/android_kernel_xiaomi_sm8150 (re @b100dian: sounds familiar. Can...)18:32
T42<b100dian> @A_T_R you can build 4.4 with the instructions from HADK nonetheless18:32
T42<b100dian> You might have the vibrator that needs custom effects. You should try ngfd 1.4 first to see if it still crashes18:39
T42<b100dian> @Verevka86 ^^18:39
T42<Verevka86> Thank you, I will inform you about the result 🤝 (re @b100dian: You might have the v...)18:40
T42<b100dian> Also add what mal said to the patterns..18:43
malwait, so you have memless driver in kernel but you don't use it?19:06
T42<adampigg> also, if you are using ffmemless, dont you also need to patch the kernel to increase the number of possible effects?19:10
T42<b100dian> ngfd now has an option to 'upload before play' but unfortunately it still uploads all effects at the beginning. This could be patched e.g. if cache_effects is on, the effects uploaded at initialization should be deleted after the first upload. I think this could solve the number of effects problem.19:22
T42<Verevka86> [defaultuser@PocoX3Pro ~]$ sudo systemctl-user status ngfd19:28
T42<Verevka86> Password:19:28
T42<Verevka86> ● ngfd.service - Non-Graphic Feedback Daemon19:28
T42<Verevka86>    Loaded: loaded (/usr/lib/systemd/user/ngfd.service; disabled; vendor preset:19:28
T42<Verevka86>    Active: failed (Result: exit-code) since Tue 2022-05-17 22:21:35 MSK; 6min ag19:28
T42<Verevka86>   Process: 8950 ExecStart=/usr/bin/ngfd $NGFD_ARGS (code=exited, status=2)19:28
T42<Verevka86>  Main PID: 8950 (code=exited, status=2)19:28
T42<Verevka86> 19:28
T42<Verevka86> Warning: Journal has been rotated since unit was started. Log output is incomple (re @b100dian: Also add what mal sa...)19:28
T42<b100dian> What do the logs say?19:29
T42<Verevka86> [0.088] DEBUG: ffmemless: creating effect list for touch;touch_weak;touch_strong;release;release_weak;release_strong;drag_start;drag_fail;drag_boundary_drag_end;short;strong;long;notice;message;attention;alarm;ringtone19:29
T42<Verevka86> Vibra upload effect: Invalid argument19:29
T42<Verevka86> [0.089] DEBUG: ffmemless: default effect load failed19:29
T42<Verevka86> [0.089] ERROR: ffmemless: Could not load default fall-back effect19:29
T42<Verevka86> [0.089] ERROR: core: sink 'ffmemless' failed to initialize19:29
T42<Verevka86> [0.089] ERROR: daemon: Initialization failed. (re @b100dian: What do the logs say...)19:29
T42<b100dian> Yoi shpuld be using a paste service, that was probably horrible on IRC19:30
T42<b100dian> *should19:30
T42<Verevka86> I'm sorry19:31
T42<b100dian> did you update to ngfd 1.4?19:35
T42<Verevka86> I have completely rebuilt the entire sfos rom19:36
T42<b100dian> yes, but ngfd 1.4 is not delivered with any sfos for now. You need to git checkout  it and build it with build_packages.sh19:37
T42<b100dian> mal: how did you reckon there's a ffmemless driver? ff-memless.c presence? Am curious, have one, don't know how's without one19:39
T42<Verevka86> https://github.com/sailfishos/ngfd This? (re @b100dian: yes, but ngfd 1.4 is...)19:41
T42<b100dian> yes19:41
malthe earlier log "ffmemless: Successfully opened ff-memless event device"19:43
T42<b100dian> oh, got it - thanks19:45
malalso you can see it in output of "evdev_trace -i"19:48
T42<Verevka86> https://paste.ubuntu.com/p/bfZMVNGqWV/ (re @b100dian: yes)20:00
T42<Verevka86> ngfd 1.420:00
malafaik memless doesn't work without configs20:01
T42<Verevka86> add config- https://github.com/sailfishos-on-tucana/droid-config-tucana/blob/master/sparse/usr/share/ngfd/plugins.d/51-ffmemless.ini20:06
T42<Verevka86> 20:06
T42<Verevka86> System sound and vibro work 😍 thank you very much 🤝 (re @SailfishFreenodeIRCBridgeBot: <mal>afaik memless d...)20:06
malgood to hear20:06
T42<b100dian> > Using constant default effect20:07
T42<b100dian> Yes, that was the call sign to use that 51-ffmemless:)20:08

Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!