Sunday, 2021-03-14

T42<edp_17> Hi all, Can somebody take a look into these, please? Build on OBS failed. Can you help me how to fix these , please? (Is it something I need to change in the uploaded files or something else?)02:18
T42<edp_17> i9100:
T42<edp_17> treltexx:
T42<edp_17> Thanks in advance.02:18
mal@edp_17 too old libhybris in your project04:06
malupdate that to latest version in those projects04:06
T42<edp_17> mal: Thanks. I thought defined the latest in the _service file.07:39
mal@edp_17 you have defined but latest is
techgeekstermai: Goodmoring! So some of the ld.config.28.txt mounts do not exist. Should I delete the non-existant ones (like dir.vendor = /odm/bin/) and keep the ones that work?12:43
maltechgeekster: what? just copy the file from the link mentioned in faq and setup the mount service12:50
techgeeksterI did. The droid-hal-init service works fine now. Yesterday you told me to check if the paths of ld.conifg.28.txt apear in the system due to error  4023 W ServiceManagement: Waited for hwservicemanager.ready for a second, waiting another...12:52
T42<edp_17> mal: Oh, I did not realised that. Will review all four ports because probably this is a problem on all of them.12:53
T42<edp_17> Thanks!12:53
maltechgeekster: is ld.config.28.txt now should it output of mount12:57
techgeekstermal: Sorry did not understand. Could you rephrase?12:58
techgeekster This is the logcat. Should i put the library mentioned in line 2 in the ld file?13:01
T42<elros34> you should check whether your new ld.config is bind mounted. Type: findmnt or mount to figure out that13:03
maltechgeekster: I told you several times check the part about ld.config.28.txt13:04
maland like elros34 said after the changes and a reboot you need to verify that is shown in output of mount13:05
techgeeksterI did all the mentioned steps in the faq and rebuild the image. That is why I find it weird that it does not work. Let me try again13:09
maltechgeekster: did you check journal and output of mount13:09
malshow what changes you did, are you sure you symlinked it correctly13:10
malalso are you sure you copied the file correctly13:10
techgeeksterthe symlink should be ln -s .. ?13:12
techgeeksterLet me do the steps again (maybe I missed something), push the changes to github and send you the repo13:14
T42<zinstack625> I'm trying out hybris-17.1 and it's getting to initial setup, but with no display. During initial setup it seems to drop telnet, how can I disable that?13:14
maltechgeekster: we are still waiting for some logs13:16
maltechgeekster: if failed to do that service correctly maybe you should show exactly what you have on device related to those se we can see what is wrong13:17
mal@zinstack625 hybris-17.1 needs some manual changes to get it working13:17
T42<zinstack625> mal: is the "manualliness" like editing libhybris? I seems to get the android side up though..13:19
mal@zinstack625 how did you determine android side is up if you lose telnet?13:20
malvarious changes to configs and things like that are needed13:20
T42<zinstack625> @mal [@zinstack625 how did you determine android sid …], I lost telnet in hybris-16 during initial setup. It's reacting to power button presses, turning the display backlight on and off, which suggests that hwcomposer is likely fine13:22
T42<zinstack625> Oh, sorry, instinctive reply13:22
T42<zinstack625> I'm fine with changing configs, made a branch just for that13:23
malI'm confused now, you are talking about both 16.0 and 17.113:23
T42<zinstack625> Telnet also died with perfectly fine hybris 16. The same applies for 17.1, afaik it has nothing to do with libhybris13:25
T42<zinstack625> The common thing seems to be that it dies specifically at initial setup, and gets fine after that. That made my opinion that android side is up13:26
T42<elros34> @zinstack625 mask usb-moded. If that will not help then check whether some init*rc files imports *usb*.rc and comment it out.13:32
mal@zinstack625 what kind of things have do to config repo so far?13:35
T42<zinstack625> mal: shamelessly stolen ld.config.29.txt, some system/etc/init rc files to disable services and point some to existing binaries13:38
T42<zinstack625> Basic necessities to make android side happy13:38
malfrom where did you take those? do you have the needed selinux changes done?13:42
T42<zinstack625> mal: I just enabled selinux, what changes are needed?13:48
T42<zinstack625> was looking at this:
maldid you copy the selinux files, not sure if those are different for each device or not13:50
T42<zinstack625> mal: on it13:53
piggzmal: how do i inhibit the android charge screen?15:23
piggzi have both fighting eachother15:23
T42<F1a5H> I also noticed that ^ on ocean @zinstack62515:24
piggzi guess i delete the android side servive from init15:27
malpiggz: I thought that would be done by default15:28
piggzmal: well, its been missed somehoe on volle (mtk)15:29
piggzmal: ive identified the service, kpoc_charger15:37
piggzill just have to do the usual magic to inhibit it15:37
T42<zinstack625> Hmm, selinux is permissive on both sides, yet it disallows access to binder. logcat is spammed, but among the agressive waiting there is: For some reason, kernel cannot load vibrator firmware again, I would guess for the same reason (generally)18:18
T42<zinstack625> Here's a tiny slice of logcat I'm getting:
malshow also journal18:45
mal@zinstack625 did you copy the selinux files?18:45
malas mentioned in the link I gave earlier18:46
T42<zinstack625> Yep, but the script insisted on making those a symlink for some reason18:46
T42<zinstack625> Those symlinks are valid though18:46
malsymlinks do not work, that is why I said to copy those18:48
malwhat script do you mean?18:48
malif you mean when some issue when building the config repo then you have too old subdmodule18:49
T42<zinstack625> Ok, I'll try it18:51
T42<Verevka86> @zinstack625 [Ok, I'll try it],
piggzmal: got a ery strange error on 4.0.1 on the volla.....19:55
piggzshortly after UI loads, at some random time, UI lock up completely19:55
piggzlipstick prints19:55
piggzlipstick[1946]: [D] unknown:0 - prepare: validate failed for display 0: 619:56
piggza bunch of times19:56
piggzmal: hmmm20:00
piggzMar 14 19:59:22 VollaPhone droid-hal-init: Service 'vendor.hwcomposer-2-1' (pid 1056) received signal 1120:00
piggzMar 14 19:59:22 VollaPhone droid-hal-init: Sending signal 9 to service 'vendor.hwcomposer-2-1' (pid 1056) process group... HYBRIS: killing PID instead of process group.20:00
piggzMar 14 19:59:22 VollaPhone droid-hal-init: starting service 'vendor.hwcomposer-2-1'...20:00
piggzMar 14 19:59:22 VollaPhone lipstick[2013]: [D] unknown:0 - prepare: validate failed for display 0: 620:00
piggzso, somehow, 4.0.1 manages to crach hwcomposer20:00
piggzmal: and
piggzany debuging tips with this one?20:17
malpiggz: that didn't happen on 3.4.0?20:28
piggzmal: no it didnt20:29
piggzmal: i remember it happening in december on a cbeta version ... never got round to looking properly20:29
malwhich kernel did that have?20:29
piggzLinux VollaPhone 4.4.146+ #16 SMP PREEMPT Sun Nov 22 10:20:28 UTC 2020 aarch64 GNU/Linux20:30
maldoes your init.rc have something like this
malpiggz: but since your device has systemd mostly running not sure if that can be the reason20:31
maldo you use the legacy cgroup for systemd?20:31
piggzmal: i wondered that ... no, i havnt made that change....20:32
piggzesp as its 4.420:32
piggz    mkdir /dev/cg2_bpf20:32
piggz# Removed during droid-hal-device build :     mount cgroup2 cg2_bpf /dev/cg2_bpf nodev noexec nosuid20:32
piggz    chown root root /dev/cg2_bpf20:32
piggz    chmod 0600 /dev/cg2_bpf20:32
piggz# Removed during droid-hal-device build :     mount bpf bpf /sys/fs/bpf nodev noexec nosuid20:32
piggzso, its removed already20:32
malpiggz: remove all parts20:35
piggzmal: still crashes20:39
malpiggz: does test_hwcomposer work?20:46
malpiggz: oh, you said UI was up for a moment20:47
piggzmal: yes, ui works for a while20:48
piggzthen, after opening a few apps, locks up20:48
malpiggz: is it stable if you don't open apps?20:50
malpiggz: sound a bit like the old fence leak issue20:50
malbut since you didn't have it before that is not very likely20:50
piggzmal: well, earliler, it locked up when i was swiping up the menu a lot20:50
malpiggz: have you had any other graphics issues?20:53
piggzmal: none, ti was quite ok in < 420:53
malpiggz: like wrong order in rendering or something?20:53
piggzmal: clean boot, no issue yet...20:56
piggzbut some mali errors in logcat20:56
piggzmal: then, it failed when opening the weather20:57
T42<zinstack625> Updating submodules helped, now no firmware can load20:58
T42<zinstack625> Like I can read the file in /vendor/firmware/, selinux is permissive and doesn't seem to care, but after loading module, it says it couldn't read firmware21:00
mal@zinstack625 can you show some errors21:01
T42<zinstack625> mal: my lovely vibrator
malhave you tried if it still works+21:03
malmaybe focus on more critical issues first21:03
T42<zinstack625> mal: the same is for adreno and wlan, just can't demonstate that as vividly21:08
T42<zinstack625> Lipstick tries to start, but can't init hwcomposer, because it can't load firmware21:09
mal@zinstack625 check zcat /proc/config.gz | grep CONFIG_FW_LOADER_USER_HELPER21:12
malif you don't have that try enabling that in kernel21:12
T42<zinstack625> Ok, mer-kernel-check must have clouded my mind21:13
malthat needs some improvements21:20
T42<zinstack625> I've got gui on hybris-17.121:32
riniguspiggz: probably not related, but cg2_bpf in your case is still mounted by systemd unit. to get rid of it, you have to apply that patch and rebuild hal21:40
piggzrinigus:  i dont see a mount unut for that21:42
riniguspiggz: there should be something, otherwise you don't get that comment "Removed during droid-hal-device build" in init.rc21:43
rinigusmaybe not mount unit, but something is mounting it probably. df -ha ?21:44
piggz%define makefstab_skip_entries /dev/cg2_bpf /dev/cpuctl /dev/stune /sys/fs/bpf21:45
piggzi skip it21:45
riniguspiggz: haven't used that one. I hope it works :)21:46
piggzrinigus: u never skipped annoying entries?21:47
riniguspiggz: I am on simple device(s) and can mostly copy jolla's configs. so, maybe they are skipped somewhere that I don't realize21:48
rinigusgetting late over here, good luck with volla!21:49
piggzrinigus: well, late too21:49
piggzmal: i will try rebasing our custom qt5-qpa-hwcomposer-plugin21:51
malhow old is the version?21:56
piggzmal: modified
malpiggz: what changes do you have in it?21:58
piggzmal: nope, same crash22:07
T42<zinstack625> How does sparse-10 work? If it finds this folder, it ignores sparse, and if android_version_major is not 10, it ignores sparse-10?22:31
malno, it copies files from sparse and if android version major 10 is found then it copies over files from there22:37

Generated by 2.17.1 by Marius Gedminas - find it at!