Thursday, 2023-07-20

CarbontatedCaffeineIt's still on00:00
CarbontatedCaffeineWe're in00:00
malok, then get output of "journalctl -b --no-pager"00:00
CarbontatedCaffeineI reconnected and the network now says "Pixel XL"00:00
CarbontatedCaffeineheres the log00:01
mallooks like it doesn't show the whole log, what does "systemctl" output00:02
malwhich service are failing00:02
CarbontatedCaffeinedev-binderfs.mount, droid-bootctl.service, droid-late-start.service, systemd-tmpfiles-setup.service, wait_for_keymaster.service (all report failing)00:03
malwhat do you see in output of "mount"00:04
malok, mounts look ok, probably00:06
malthen we need to figure out why droid-hal-init is causing the device to reboot00:08
CarbontatedCaffeineHow can I do this when it will stop the phone from booting correctly00:09
malthat just means you need to run droid-hal-init manually (with correct env set) and monitor logs in another telnet terminal00:12
malso you run droid-hal-init in one telnet and use "dmesg -w" in another and maybe "journalctl -fa" in third one00:13
malkernel logs00:13
CarbontatedCaffeinewhich one is that00:13
maldmesg is kernel logs00:13
malmost of the time those appear is journal also00:13
maldepending on how soon it reboot you could also run /usr/libexec/droid-hybris/system/bin/logcat in one telnet right after starting droid-hal-init00:14
CarbontatedCaffeineI don't have that command00:15
CarbontatedCaffeineI get a file not found00:16
malI mean it can only be run after droid-hal-init is running00:16
malanyway, you can see how droid-hal-init is started in /usr/bin/droid/droid-hal-startup.sh00:17
malwhen you start it run the touch command in there and that export also, then /sbin/droid-hal-init00:17
CarbontatedCaffeineI didn't have enough time to start logcat00:18
maljust to be sure, did you correctly configure selinux?00:22
CarbontatedCaffeineI put my selinux stuff in there yes00:23
CarbontatedCaffeineI don't know if I set it to 100:23
CarbontatedCaffeinecause I don't remember how to do that00:23
malshow "cat /proc/cmdline"00:24
malok, found something wrong in the logs at least, it starts ueventd which should not be started, you need to override that service in /usr/libexec/droid-hybris/system/etc/init/ check if that ueventd service is defined in /vendor/ueventd.rc00:34
malit's possible you might also need one patch to system/core00:35
mal I needed that one 18.1 base, can't remember if that is always needed or not00:36
CarbontatedCaffeineI cannot find anything to do with the service being defined00:37
CarbontatedCaffeineI will paste the output anyway00:37
CarbontatedCaffeinecause I can't be trusted obviously00:37
CarbontatedCaffeineI did search ".service" and "uevent" but nothing came up except for the command I typed in00:38
malnot .service00:39
malI mean android side services00:39
CarbontatedCaffeineI will patch the file00:41
CarbontatedCaffeineOkay, done.00:46
CarbontatedCaffeineCan I "make -j$(nproc --all) hybris-hal droidmedia"00:46
malno need to build droidmedia, hybris-hal is enough00:48
CarbontatedCaffeineall good00:49
CarbontatedCaffeine> it starts ueventd which should not be started01:08
CarbontatedCaffeineI just realized I never got to this01:09
CarbontatedCaffeineI'll have to take a look in a moment01:09
CarbontatedCaffeineAyy, great progress01:10
CarbontatedCaffeineit doesn't reboot anymore01:10
CarbontatedCaffeinedroid-hal-init works now01:11
CarbontatedCaffeinethanks for the patch01:11
malany luck with graphics?01:12
CarbontatedCaffeineNo graphics01:12
CarbontatedCaffeinedev-binderfs.mount, droid-bootctl.service, systemd-tmpfiles-setup.service, wait_for_keymaster.service (are still failing)01:13
maljournalctl and logcat might be useful now01:19
CarbontatedCaffeinelogcat does so much01:24
CarbontatedCaffeineI can't read it all01:24
CarbontatedCaffeinehopefully this is enough ^ logcat01:25
maltry to kill vndserbicemanager process and see if anything happens after that01:30
CarbontatedCaffeineI can't find it in top01:32
malps aux | grep vnd01:33
CarbontatedCaffeineI klled it01:33
CarbontatedCaffeinenow the notification light is blinking01:34
malis the hwcomposer service in logcat still starting and stopping01:35
malhmm, needs some more debugging then, but it's late now so I need some sleep01:44
CarbontatedCaffeineAll good, thanks for your help01:46
*** Danct12 is now known as Guest648701:55
T42<smatkovi> what can i do about this: Repository 'dir:/home/sebastian/hadk/droid-local-repo/onyx' is invalid.07:02
T42<smatkovi> [~plus-repo-1|dir:/home/sebastian/hadk/droid-local-repo/onyx] Valid metadata not found at specified URL07:02
T42<smatkovi>  - Signature verification failed for repomd.xml07:02
T42<smatkovi> removing this directory seems to have worked07:17
T42<smatkovi> now it complains about missing dependencies for the droid-hal-version-onyx package07:26
T42<smatkovi> okay i tried to set this to 1 now ./hybris/droid-hal-version-onyx.log:222:warning: Macro expanded in comment on line 9: %define have_vibrator 007:36
T42<elros34> see second to last line in your paste07:42
T42<smatkovi> you mean i should look into the log07:44
T42<smatkovi> there i find File /etc/ofono/binder.conf  from install of droid-config-onyx-1-202307200732.armv7hl (dir:/home/sebastian/hadk/hybris/droid-hal-version-onyx/.mb2/filtered-output-dir) conflicts with file from package   ofono-configs-binder-1.1.11-1.5.1.jolla.armv7hl (@System)07:45
T42<smatkovi> i have to put this in obsoletes somewhere07:45
T42<smatkovi> but i forgot where07:45
T42<smatkovi> maybe here? /home/sebastian/hadk/hybris/droid-configs/rpm/droid-config-onyx.spec07:47
T42<elros34> see hadk-hot, it shows where and what to put07:48
T42<smatkovi> droid-config-onyx-bluez5 conflicts with bluez5-configs-mer08:11
T42<smatkovi> faq says something about uninstalling the package, but it isn't installed08:11
T42<smatkovi> hot i mean08:11
T42<smatkovi> i wrote provides droid-config-onyx-bluez5 and now it seems to work09:16
T42<elros34> basically this is already there,  droid-config-onyx-bluez5 should be created09:28
T42<smatkovi> i got a zip file :)09:31
T42<smatkovi> now i got this in dmesg on telnet
T42<smatkovi> journal is not available09:42
T42<elros34> looks like missing some changes in old 3.4 kernel. I think there should something in hadk-hot about it10:04
T42<smatkovi> what should i search for in the hadk-hot?10:09
T42<smatkovi> maybe i need these vfs patches10:16
T42<elros34> yeah this one10:18
T42<smatkovi> should i apply every diff from these two commits in the hammerhead repo? vfs: make O_PATH file descriptors usable for fstat()10:20
T42<smatkovi> vfs: Fix /proc/<tid>/fdinfo/<fd> file handling10:20
T42<elros34> some devices had already partial changes some none, you need to  check. I do not remember exactly10:33
T42<smatkovi> okay, thanks. i thought building for onyx would be easier because there is already an older port and i didn't expect patching the kernel. i will try, because i'll learn something along the way, but maybe i should consider also taking a more recent lineageos? with that i would have to find a manifest then that suits though...10:42
T42<elros34> I do not see a point in using newer lineage unless it uses newer kernel/android blobs or some hardware driver was fixed compared to older version11:20
T42<smatkovi> I only know about a newer bootloader, but that probably doesn't have to do with sailfish OS then11:58
CarbontatedCaffeineHello, I have completed some debugging12:47
CarbontatedCaffeinerunning "EGL_PLATFORM=hwcomposer test_hwcomposer" gives me an error message of "library "" not found" that's not right :P12:48
T42<Mister_Magister> it litererally doesn't do what you say12:49
T42<Mister_Magister> "library "" not found" is not an error12:50
T42<Mister_Magister> test_hwcomposer: test_common.cpp:370: HWComposer* create_hwcomposer_window(): Assertion `err == 0' failed. this is an error12:50
CarbontatedCaffeineI am stuck with the same issue another person was12:51
CarbontatedCaffeinewhere "ServiceManager: Waiting for service 'display.qservice' on '/dev/vndbinder'..."12:51
CarbontatedCaffeineI'm getting this constantly in logcat12:51
CarbontatedCaffeine"Could not restart 'surfaceflinger': Cannot find 'surfaceflinger_HYBRIS_DISABLED': No such file or directory"12:52
T42<Mister_Magister> thats clearly intentional it says disabled12:53
CarbontatedCaffeineIt's not clear to me12:53
CarbontatedCaffeineremember, I haven't been around12:53
T42<Mister_Magister> it lterally says disabled12:53
CarbontatedCaffeineI haven't got a clue what the actual issue is12:55
CarbontatedCaffeinethere's so much on the screen12:55
CarbontatedCaffeine"ServiceManager: Service display.qservice didn't start. Returning NULL"12:58
malthis somehow reminds me of the issue I had on one device, the problem being vndservicemanager13:02
CarbontatedCaffeineWhy was it behaving that way?13:03
CarbontatedCaffeineand what did you do to fix it13:04
malI haven't figure out a fix for the issue13:08
malI had to kill some process during boot and then start the rest13:09
CarbonatedCaffeineDoesn't sound nice13:09
CarbonatedCaffeineWell that's underwhelming, though I don't really want to give up considering how much time I've spent on it13:11
malbut we need to first test if that is the case for you13:11
CarbonatedCaffeineWhat do I test13:12
malcheck what binder-list command shows for the different /dev/*binder* devices13:18
malthat is in libgbinder-tools package13:18
CarbonatedCaffeineI wasn't able to get the internet routing working this time13:31
CarbonatedCaffeinenevermind, got it!13:38
CarbonatedCaffeine"Package 'libgbinder-tools' not found."13:40
CarbonatedCaffeine"Error message: SSL certificate problem: certificate is not yet valid"13:41
malset correct date on your devie13:58
T42<b100dian> mal: droidmedia pull 111 got a bit fatter:) but currently no more crashes on change res or buffer leaks (hopefully)14:25
T42<elros34> @CarbonatedCaffeine: read 'generating logs' section in hadk-hot. It's easy to miss some important information with partial logcat you provided. Upload your kernel/droid changes to github14:43
mal@b100dian interesting17:29

Generated by 2.17.1 by Marius Gedminas - find it at!