Saturday, 2023-04-01

T42<k1gen> mal:
*** amccarthy is now known as Guest957407:48
*** amccarthy_ is now known as amccarthy07:48
voidanix[m]is there anything required to make work on 4.5.0?11:34
voidanix[m]other than updating droid-configs-device, ofc11:34
voidanix[m]device boots to a black screen otherwise11:35
voidanix[m]nvm it's not on the upgrade-4.5.0 branch, guess it's for a later version...11:56
malit needs newer mce if you use latest master branch of the submodule13:22
mal@k1gen ok, then no useful modules in there for audio13:24
malwhich just means all of those are builtin13:25
T42<k1gen> is it bad?13:25
malno, I just wanted to be sure there was nothing there13:27
T42<k1gen> alright13:28
T42<b100dian> I have trouble starting gui. I am masking user@10000 and test_hwcomposer crashes and qmlscene minimer does not (but it doesn't do anything either)15:45
T42<b100dian> I notice that one of the outputs of qmlscene is that libandroidicu is not found. Linking that from /odm/lib64 gives me the "is not accessible for the namespace "(default)" message15:46
T42<b100dian> and then grepping the logs I found that this might be /linkerconfig related. I do have a systemd service that mounts What=/linkerconfig/bootstrap Where=/linkerconfig15:47
T42<b100dian> Should I change that to default?15:48
T42<elros34> not sure if you can change that easily without breaking some boot services, IIRC switching from bootstrap to default is commented out by hybris patches. Does surfaceflinger work?16:20
T42<b100dian> I think I just broke it, but it used to vibrate but do nothing16:38
T42<b100dian> Ok, back to previous state - vibrates, but doesn't display anything16:42
T42<b100dian> (I broke it by reverting "in spirit", adding `#ExecStartPre`, maybe that's incorrect)16:42
T42<b100dian> In theory those files are too new, I should have just deleted them16:43
T42<elros34> yeah to new submodule16:46
T42<elros34> @b100dian I think I used something like this for linkerconfig (and not relevant for you apexes) but I am not sure:
T42<elros34> and i think without commenting out "mount none /linkerconfig/bootstrap /linkerconfig bind rec" in init.rc17:21
T42<b100dian> The result seems to be that there is no /linkerconfig/default or bootstrap subfolder.  I'm not sure how come this is not visible in the mount | grep linkerconfig but I assume it worked18:51
T42<b100dian> I don't see a difference in my error: library "" needed or dlopened by "/usr/libexec/droid-hybris/system/lib64/" is not accessible for the namespace "(default)"18:52
T42<b100dian> Maybe the ld.config.txt files need to be changed. I remember there was a such change before, was it in our etc?18:52
T42<k1gen> @b100dian I had that error before, search for libandroidicu.so18:53
T42<b100dian> Yeah. How did you solve it, by bind mount instead of linking?18:53
mal@b100dian check strace to see if it really fails to open the lib, sometimes it falls back to some other path for loading but still logs an error18:56
malwhich android base is that?18:57
T42<b100dian> True that. It might be a red herring in the case of qmlscene/minimer.18:57
T42<b100dian> Lineage 18.118:57
malhow does logcat ad journal look on your device, just wondering if some important services fail18:58
T42<elros34> @b100dian about patch: I assume you disabled systemd mount and enabled bootstrap mount in init.rc?18:59
malwasn't there that one patch for bootstrap stuff, I was supposed to push that to hybris-patches but probably forgot19:00
malfor devices which don't have updatable apex19:00
malnot sure which type that device is19:01
T42<b100dian> @elros34 no, I assumed they would be equivalent..19:04
T42<b100dian> mal: probably the last patch Elros linked is the one19:04
T42<b100dian> @k1gen how did you get past that "default" namespace error then?19:04
malyes that one19:05
T42<elros34> not sure about init.rc anymore, it was quiet a time ago but I think systemd mount should be disabled19:05
T42<b100dian> Init has > # Removed during droid-hal-device build :     mount none /linkerconfig/bootstrap /linkerconfig bind rec19:06
T42<k1gen> @b100dian sorry, fixed couple other things after that. check in chat history, I should've said how I fixed that (if I did)19:06
T42<b100dian> Systemd mount is [Mount] What=/linkerconfig/bootstrap \n Where=/linkerconfig \n Type=none \n Options=bind,rec19:06
T42<elros34> I doubt it, he probably still have it or talking about different error about missing file not namespace19:07
T42<b100dian> Yes, this was my impression too19:08
T42<b100dian> but as mal says, qmlscene is not blocked by this. It actually finds /apex/ :)19:11
T42<b100dian> So I would like suggestions what to do about the no-GUI19:12
T42<elros34> so have you tried starting surfaceflinger?19:12
T42<b100dian> Yes, it doesn't crash but doesn't display anything19:12
T42<elros34> together with bootanim?19:13
T42<elros34> ation19:13
T42<b100dian> Interesting one. When I start surfaceflinger, bootanimation says "ANDROID_ROOT not set"19:14
T42<b100dian> and aborts19:14
T42<elros34> you need oto set that env: hadk-faq19:14
T42<elros34> or hot19:15
T42<b100dian> ANDROID_ROOT="/system" /system/bin/bootanimation yes, I found that in my reverse-serach19:16
T42<b100dian> "successfully turns off the screen"19:17
T42<b100dian> (I mean, ASUS logo is gone)19:17
mal@b100dian how does lipstick fail?19:18
T42<elros34> no errors in dmesg/logcat?19:18
T42<b100dian> Logcat around that surfaceflinger start is
T42<b100dian> This is suspicious > HWComposer: isConnected failed for display 19261202339590786: Invalid display19:39
malyou still haven't shown full logs, or I missed those19:42
T42<b100dian> And journalctl about the same time (should include dmesg)
T42<b100dian> mal: no, I haven't. So you want to unmask user@10000 and show boot -> lipstick? logcat and journal?19:45
malwell you are only showing small parts of logs so something important could be hidden in the earlier part of the logs19:45
malat least show full journal19:46
mallogcat is also helpful probably19:46
T42<elros34> disable that vendor.time_daemon , it only spam your logs19:54
T42<b100dian> journal
T42<b100dian> logcat
T42<b100dian> Elros: I did try to _HYBRIS_DISABLE it before booting. I think it needs override & disable too19:56
T42<elros34> "dummy_compositor: unrecognized option '--hwc-restart", are you sure you use correct submodules?19:56
T42<b100dian> no, I am probably not. That's why I tried to revert on device earlier. I should just rm the files19:58
mal@b100dian have you checked if any mounts fail?19:59
malthat dummy_compositor error means you have too new submodule20:00
T42<b100dian> Removed thos files, bootanim now works20:05
T42<b100dian> *those20:05
T42<b100dian> They submodule is reset locally to the corrrect commit, I just haven't correctly done the same changes on the device, nor reinstalled20:05
T42<elros34> so no longer isConnected failed error?20:06
T42<elros34> do you have /mnt/vendor/persist? 'findmnt'20:11
T42<b100dian> I still see that isConnected error from surfaceflinger20:12
T42<b100dian> Yes, I have /mnt/vendor/persist20:13
T42<b100dian> To mal's previous question: the only mounts I see failing are /vendor/asusfw and /vendor/vm-system which I checked that they don't have any corresponding img in the asus firmware folder and no contents (0) on the device, they were just listed in fstab.default for some reason I don't know20:16
T42<elros34> did you enable for some reason, I see it in log?20:17
T42<b100dian> no20:18
T42<b100dian> but my hunch is that some things are parsed in some other order20:18
T42<b100dian> and it also needs override + disabled20:18
T42<b100dian> I remember to have lost some time with this with the previous device. I think this exact service20:18
T42<b100dian> (altough that device rebooted if audio-hal was enabled and audioserver was not)20:20
T42<b100dian> Ok, disabled audio-hal + override. I notice lipstick crashes because of a library on startup but starts later20:51
T42<b100dian> Adreno  : IsValidNativeBuffer: Buffer has a NULL handle20:51
T42<b100dian> Adreno  : DequeueBuffer: Dequeued Buffer is not valid20:51
T42<b100dian> While /vendor/bin/hw/vendor.qti.hardware.display.composer-service says > DisplayBase::SetVSyncState: Can't enable vsync when display 54-0 is powered off or SecureDisplay/TUI in progress20:52
T42<b100dian> This is from
T42<b100dian> fresh logcat with 5988 being composer service and starting qmlscene (11457):

Generated by 2.17.1 by Marius Gedminas - find it at!