Sunday, 2021-08-22

T42<Walid> Yes (re @SailfishFreenodeIRCBridgeBot: <mal>ok, did you get...)03:49
T42<Walid> also i got black backround home screen , theme not well applied04:36
*** asriel is now known as Guest504308:33
rinigusnote to future myself or someone interested - device timeout can be handled by creating .device unit with the correct name (such as dev-mapper-home_encrypted.device) and setting `JobTimeoutSec` for desired value10:18
T42<adampigg> Rinigus, hows the encryption going11:06
T42<b100dian> "LG bootanimation" may mean treble A/B activation?12:24
rinigus@adampigg: it is doing well. I can unlock loopback device with separate /home/encrypted folder on boot before hitting multi-user & lipstick12:44
rinigusso, GUI is there for unlocking.12:45
rinigusright now battling systemd boot scripts to get it to work with /home. if will not figure it out will start asking questions here. just few experiments still needed12:45
rinigusany idea which service starts welcome screen with "Ahoi" on new install?13:18
T42<Walid> No, treble is not activated (re @b100dian: "LG bootanimation" m...)13:21
T42<Walid> Why i have dark thème, i changed Many ambience but still background black13:23
T42<Walid> https://irc.thaodan.de/.imgstore/3ab2660a/file_2522.jpg13:23
T42<Walid> https://irc.thaodan.de/.imgstore/9e8c1877/file_2523.jpg13:24
T42<elros34> rinigus maybe jolla-startupwizard-pre-user-session.service13:25
rinigus@elros34: thanks, maybe13:25
riniguswill see if it helps to push mounting /home in front of that13:26
T42<elros34> Walid: search for errors in logcat maybe some gpu issue. Do you have adreno quirk added to droid-hal?13:27
T42<Walid> yes13:28
T42<Walid> %define android_config \13:28
T42<Walid> #define WANT_ADRENO_QUIRKS 1 \13:28
T42<Walid> #define QCOM_BSP 1 \13:28
T42<Walid> %{nil} (re @elros34: Walid: search for er...)13:29
T42<Walid> i will see logs13:29
T42<elros34> Walid what about T_OPENGL_NO_BGRA=113:31
T42<elros34> sry QT_OPENGL_NO_BGRA=1, iiRC this helps for black pictures in gallery13:31
rinigus@elros34: wizard is triggered after systemd-user-sessions.service which I already use to push mounts in front. will look for some other options ...13:32
T42<elros34> rinigus systemd-analyze plot could be useful to figure out what is started and when13:34
rinigus@elros34: indeed. sitting with it for 2nd day :)13:35
riniguswell, will take a break for a bit13:36
T42<Walid> yes works thx (re @elros34: sry QT_OPENGL_NO_BGR...)13:38
rinigusI have a problem with cleaning up after app was running in the same environment as lipstick during boot sequence: EGL_PLATFORM=hwcomposer QT_QPA_PLATFORM=hwcomposer17:03
riniguswhen using this UI for systemd ask password, I can boot without any issues and unlock an encrypted image file. this is if I boot into lipstick17:04
rinigusif I do the same and boot into jolla-startupwizard, my UI has already finished (image for /home decrypted and mounted), but wizard is somehow shown under my app's last screen17:05
rinigusas a result, I cannot interact with wizard, screen image even starts wiggling a bit.17:06
rinigusI am sure that the UI for systemd ask password has exited by that time - judging by journal logs17:07
rinigusif I look closely then during boot to lipstick I can see my app's last screen just for a sec before entering lock screen UI of lipstick.17:08
rinigusso, it looks like lipstick is able to cleanup after my app and boot properly while startup wizard is not17:08
rinigusany idea on how to cleanup hwcomposer properly on exit from Qt app?17:09
rinigusmy sources: https://github.com/rinigus/systemd-ask-password-gui17:09
rinigussystemd units not uploaded, should probably make new project for it17:10
rinigusversion of qt5-qpa-hwcomposer-plugin is 5.6.2.18 - should have a commit switching off display on app exit. can see that screen goes dark as well17:20
T42<eugenio_g7> rinigus: you should restart the composer service on the android side17:22
T42<Walid> For ril what conf should i set17:22
rinigus@eugenio_g7: how would I do that?17:22
T42<eugenio_g7> https://github.com/mer-hybris/droid-config-sony-seine/blob/1842f01de928f75aba859b0e1edf711a2225d7d0/sparse/usr/lib/systemd/system/yamuisplash.service.d/50-vendor-hwcomposer-2-3.conf17:24
rinigus@eugenio_g7: thanks! I will try to stop it after and start again after my app. will report back17:27
T42<eugenio_g7> you're welcome! I think that's needed only on aosp1017:29
rinigus@eugenio_g7: and it worked, thank you!17:37
rinigusI can now boot into /home encrypted with freely set password, as setup earlier with cryptsetup17:38
riniguspiggz: ^17:38
riniguswould need to cleanup, make new github repo and push the scripts there17:38
T42<eugenio_g7> great! :)17:38
rinigusafter that, small adjustment and I should have android keystore backed keys as well (using hwcrypt)17:39
T42<adampigg> Neat17:59
T42<adampigg> Rinigus: could use fingerprint as enc key?18:23
rinigus@adampigg: not sure. it would need some research... basically, you would need to have some secret stored behind fpd18:33
rinigusnow that's a long repo name (based on fpd scheme): sailfish-device-encryption-community18:33
rinigusif nothing better comes up in few minutes I will use that. can always rename later18:34
T42<adampigg> rinigus: seems possible https://stackoverflow.com/questions/44520343/android-how-to-get-unique-key-from-finger-print-authentication18:40
T42<adampigg> store key in see/tee, use fp to unlock/retrieve18:40
rinigus@adampigg: it boils down to whether `setUserAuthenticationRequired(true)` is protected from `root` or not.18:44
rinigusI'll start with the "normal" ways for encryption and let's see if we will deal with fpd later...18:45
T42<adampigg> Rinigus :yes, this talls about that too https://medium.com/beautycoder/android-security-and-fingerprint-ef0f6f34488818:50
rinigus@adampigg: I maybe reading to fast and have missed it, but I don't see anywhere in these docs that fp and keystore communicate directly. This is without leaving tee/se.18:58
rinigusAt https://developer.android.com/reference/android/security/keystore/KeyGenParameterSpec.Builder#setUserAuthenticationRequired(boolean) there is some talk about lockscreen suggesting that Android bits are in play. Which may make safety of the key in keystore questionable in this case19:00
rinigusSo far I have avoided user auth and use the key without it leaving tee/se. Which allows it to protect Luks password against root as well19:01
malrinigus: how does that keystore relate to keymaster?19:11
malrinigus: just trying to understand what things you code does with that19:13
rinigusmal  let me come back to you on this tomorrow morning. I wish suspect that I mixed up 'key*' terms and would have to look up correct used one (by hwcrypt)20:20
malrinigus: I just meant that since you seem to go the hal way and afaik key handling could be done also via binder using keymaster if I understand correctly what you want to do20:23
malunless I mix things a bit20:25
malrinigus: it would be nicer if we could do things via some binder hidl api and even make a generic build (unless we want to support older devices, android < 8)20:33
T42<b100dian> anyone has any clue how to debug missing incoming SMS messages? It used to work correctly for about one month on my port, but something changed over the weekend - even if I swap the SIM from slot 1 to slot 2, remove ~/.local/share/commhistory I can at most make it receive my own SMSes from another number, while authentication SMS's which usually don't come from a 'number' are never arriving20:41
T42<b100dian> adding -d to ofono.service doesn't reveal anything20:42
T42<b100dian> the trigger _might_ have been me sending an international SMS, but I removed that from the database and it didn't fix the issue20:42
T42<b100dian> is there any other service that needs to be made verbose?20:43
malcan you check permissions of /var/lib/ofono20:44
T42<b100dian> thanks mal - they're radio/radio all the way down20:45
malok, then that is ok20:46
maldo you see anything in journal log when you are supposed to get some sms, also check output of "/usr/libexec/droid-hybris/system/bin/logcat -b radio"20:47
T42<b100dian> ah, that is new, let me check both, I only looked at journalctl till now20:48
T42<b100dian> mal: logcat -b radio doesn't output anything on SMS (it does on call), and here's journalctl with the ofono -d parameter.. https://pastebin.ubuntu.com/p/fTwbxBhFS7/20:59
mal@b100dian have you done any sailfish os update or anything since it was working?21:02
T42<b100dian> hmm I might have updated packages from nemo:devel:hw:xiaomi:tucana which pull in nemo:devel:hw:common21:05
T42<b100dian> the only think that changed recently there is harbour-mmslog 14days ago21:06
T42<b100dian> but that21:07
T42<b100dian> is not installed21:08
malok21:08
T42<adampigg> https://irc.thaodan.de/.imgstore/c1029ea1/file_2525.jpg21:10
T42<b100dian> maybe I should install it:)21:10
T42<adampigg> using hidl for sensors on volla21:11
T42<b100dian> niice!21:12
T42<b100dian> what app is that21:12
mal@adampigg did you implement it properly?21:14
T42<adampigg> no (re @SailfishFreenodeIRCBridgeBot: <mal>@adampigg did y...)21:25
mal:)21:26
T42<b100dian> Ok, I don't know what happened. I was in a state where after restoring a backup (to get rid of recent smses), enabling or disabling airplane mode, sending smses from my other phone - then all the other 2fa smses started pouring in.21:37
T42<b100dian> mal you can make another round of finding unproperly stuff at my ngfd PR:)21:48
malok, need to have a look some time during next week22:03

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