Saturday, 2019-04-20

r0kk3rzsomeone fixed it the other day, but never said what they did01:46
r0kk3rzfairly sure its a permissions issue01:46
osum4estit looks like initialize_properties looks for a property_contexts file in the vendor partition. the vendor partition isn't mounted for me. could this be the issue? i added vendor to fstab.common but it didn't seem to want to mount03:08
r0kk3rzsystemd does the mounting03:17
r0kk3rzbut yes id expect vendor to be mounted03:18
osum4estthen should i add a vendor systemd mount unit?03:21
osum4estsystem is also not mounted btw03:32
r0kk3rzthat definitely needs to be mounted03:32
r0kk3rzi wonder why they're missing?03:32
osum4estno idea. what is supposed to mount them? systemd? init.rc? hybris-boot?03:33
osum4esti don't see any .mount files for them03:34
r0kk3rzbut it should mount everything in the fstab03:34
osum4estthere's multiple fstabs. are you talking about /etc/fstab? because that one has no entries for system or vendor03:34
r0kk3rzah perhaps thats why03:35
r0kk3rzolder devices have like, 1 fstab03:35
r0kk3rzbut try adding them to /etc/fstab03:35
osum4estwhat type are they? ext4?03:39
osum4estkeep getting "dev-block-platform-soc-624000.ufshc-by\x2dname-system_a.device: Job dev-block-platform-soc-624000.ufshc-by\x2dname-system_a.device/start timed out." although i might be using the fstab wrong03:43
osum4estadded "/dev/block/platform/soc/624000.ufshc/by-name/vendor_a      /vendor   ext4    defaults         0 0" to it03:44
r0kk3rzdont use the alias, point it direct to the mmc device03:45
osum4estalright, thanks. those are mounted now. still getting the initialize property area error, but it got a little further. now it created a /dev/__properties__/properties_serial file03:51
T4<Hound77> <mal> What should I do after generating the .zip file ? And fails to flash using twrp ?04:03
osum4estr0kk3rz: wait isn't system the same partition that's mount at root? becuase default.prop is a symlink to itself, therefore inaccessible.04:07
osum4estas well as ld.config.txt since /system/etc/ is also a symlink to itself. not sure if i mounted system correctly04:15
osum4estit seems like i want system/system to be mounted at /system, but i'm not sure how to mount a specific folder of a partition04:16
r0kk3rzso you mounted it to /system?04:20
osum4esti mounted the system partition to /system, yes04:20
osum4estjust tried mounting system to /system_raw and making a symlink at /system to /system_raw/system but now i'm bootlooping04:21
r0kk3rzyeah i dont think you want to do that04:21
r0kk3rzwhat device is this?04:22
osum4estthen how do i deal with the recursive symlinks? its marlin04:22
r0kk3rzwhich symlinks?04:27
osum4estif i mount system to /system, then there is a symlink at /system/etc that points to /system/etc04:29
r0kk3rzhmm, seems like the partitions on this device are a bit different to usual04:32
osum4esthere a recent journalctl: it's unable to read ld.config.txt and unable to access a property (in default.prop i assume)04:33
r0kk3rzok, yeah dont mount system like that04:37
r0kk3rzbut make sure vendor is mounted04:37
osum4esti think i'm correct in symlinking /system to the systempartition/system. i turned on logs across reboots and it bootlooped because of selinux. which makes sense because i haven't figured out how to disable it04:42
osum4esti think it may have gotten further that time, but i could also be completely wrong...04:43
r0kk3rzsystem should already be mounted as root04:43
osum4estmounted where?04:44
T4<MarcoDS_bit> Does anyone know where the unit.$DEVICE.rc is located?07:09
T4<MarcoDS_bit> init*07:14
T4<NotKit> it used to be in / before Treble, now it probably got moved to /vendor/etc or somewhere in /vendor07:41
T4<MarcoDS_bit> Found thank you08:06
AlexKrysh_Hallow. I porting Sailfish OS on Mi MIX 2S. GUI do not start. I have this error: "Mar 13 03:16:11 Sailfish jolla-startupwizard-pre-user-session[3715]: library "" not found", but found in /vendor/lib/egl. Journal:  Logcat:
T4<adampigg> mal: @eugenio_g7: any tips for building a mainline kernel?i _think_ it boots and then panics/reboots ... im building a kernel bzImage, and using the sailfish ramdisk ...... when I boot it in QEMU i get
T4<adampigg> hang on, maybe that was the wrong ram disk....09:26
T4<adampigg> i'll keep playing to siggestions!09:27
T4<eugenio_g7> Those messages remind me of the Android init09:44
T4<Hound77> @elros34 Any suggestion on what to do next. I failed to flash the .zip as mentioned on 8.5. Then I moved to 11.1 and extracted root fs and what after that ?10:39
T4<Hound77> I am stuck here for days.10:40
T4<austenite> You flash hybris-boot and get logs thats it10:48
T4<Hound77> How can I take the logs ? telnet ?10:50
T4<austenite> yes10:50
T4<Hound77> Or does logs can be retrived from the device ? Does it store the log ?10:52
T4<elros34> Charlie: how did you flash kernel?10:53
T4<Hound77> fastboot on bootloader. sudo fastboot flash boot hybris-boot.img10:54
T4<elros34> now on your pc: dmesg -w. Turn on device and connect it to usb10:55
T4<elros34> does device reboot? try also sudo lsusb command from hadk11:02
T4<Hound77> It bootloops.11:06
T4<Hound77> somethimes boots back to bootloader11:06
T4<elros34> so try flash hybris-recovery to boot partition and get all logs before it reboots. How fast it reboots?11:09
T4<Hound77> Its slow. I do get the waning about "You device cant be checked for errors..." everytime. Its common after I unlocked the bootloader.11:11
T4<Hound77> This is new, "Bootloader logs" appeared on my phone.11:15
T4<Hound77> It says ""11:15
T4<Hound77> "No bootable A/B slot  … Failed to boot Linux, falling back to fastboot … Fastboot reason: Fall-through from normal boot mode".11:18
T4<Hound77> But lsusb -v | grep iSerial doesnt show anything required.11:18
T4<MarcoDS_bit> So, we've got a connection over port 2323, still no UI but that seems much better anyway, what do you think? Journalctl: dmesg:
r0kk3rzmal: Jan 02 01:01:31 Sailfish droid-hal-init: Failed to initialize property area11:42
r0kk3rzits everywhere11:42
T4<elros34> Charlie: maybe activate different a/b slot or you need to clear some flag because it fails to boot too many times11:43
T4<Hound77> @elros34 flash every thing ?11:47
T4<elros34> I don't know you need to figure it out. After quick search of error (no bootable slot) I find out it's common issue.11:51
T4<Hound77> Okay11:53
steeeveHey r0kk3rz13:50
T4<austenite> are overlayfs patches needed for kernel 3.18.31?15:10
T4<Hound77> @austenite What happens if the the zip obtained after mic create flashes successfully ? Do it boot to Sailfish or display some error.15:13
T4leserf50 (MISSING @USERNAME! was added by: leserf50 (MISSING @USERNAME!
T4<austenite> @Hound77 it displayed one error but in last line it said successfully flashed15:15
T4<austenite> definitely not error 715:15
T4<Hound77> @austenite See the problem is that if I use latest TWRP I can't flash Sailfish zip but it flashes Lineage but if I use old one its vice versa. What do you think I should do ?15:20
T4<austenite> flash lineage with new twrp and ( fastboot boot  oldtwrp.img) and flash sailfish :)15:21
T4<austenite> and i dont know about oreo, i have done adaptation for noughat15:23
T4<elros34> you should not focus on that right now. First try to get hybris-boot (it's just a modified android  kernel with init script) working with telnet then you can focus and runnnig sailfish:
T4<Hound77> @elros34 I ment for the zip that I got from mic create at 8.515:26
T4<austenite> @Hound77 dont worry about zip, errors during zip creation /before sucessful flashing dont appear to hamper booting process ..15:27
T4<austenite> problem is more with kernel modification, once sailfish boots up, you can use the same kernel for halium ;)15:29
T4<austenite> *same kernel sources15:30
T4<Hound77> I flashed the zip file now. But it showed me one failure to mount, … ```mount: Failed to mount /dev/block/bootdevice/by-name/userdata at /data: Device or resource busy``` … but rest of it ended well.15:33
T4<austenite> its normal message you can check the folder in /data/.stowaways/sailfishos if you are worried15:37
T4<austenite> also you can solve problems by pasting the error line in google followed by 'merproject logs' it was very helpful15:41
T4<Hound77> Okay15:42
T4<Hound77> @austenite What should I do next ?15:51
T4<Hound77> @austenite I rebooted and it boots to bootloader. Should I continue to 12.1 ? Do I have to do 9.3 (Logs across reboots) ?16:01
T4<austenite> no, commands under DEVICE $  is to be performed on the device16:20
T4<austenite> *are16:21
T4<Hound77> adb shell ?16:21
T4<austenite> no, in the terminal / telnet16:22
T4<Hound77> How do I get to telnet ? 9.2 ?16:22
T4<austenite> actually you can get some kernel logs so someone can help..
T4<adampigg> @eug16:31
T4<adampigg> @eugenio_g7 with the correct initram, i get in qemu16:31
T4<eugenio_g7> @adampigg: which rootfs are trying it with? Sailfish SDK's?16:42
T4<adampigg> @eugenio_g7 thats the initramfs from hybris-boot16:43
T4<adampigg> im gonna try that on the tablet now16:43
T4<eugenio_g7> @adampigg: have you attached any disk with Sailfish OS on it in qemu? Otherwise it does seem ok that it hangs there16:48
T4<Hound77> @austenite It says to have few things on the kernel configs. So I should rerun make hybris with added configs ? By the way the logs should be stored under sailfishos rootfs right ?16:48
T4<eugenio_g7> Probably SDK's emulator won't work either as latte's initramfs expect a different partiton table16:49
T4<eugenio_g7> SDK's virtual disk*16:49
T4<austenite> @Hound77 no you can retrieve from twrp in next boot16:52
osum4est@MarcoDS_bit let me know if you figure out "failed to initialize property area" i'm stuck on the same thing17:28
*** osum4est_ is now known as osum4est17:36
T4<NotKit> is it hybris-15.1 branch? did mal update it?17:43
malI updated it, maybe there is something wrong with it17:43
osum4esti tried an old rootfs i had before you updated it and it had the same problem17:43
TheKitthat is usually due to SELinux code in init not getting neutralized17:45
TheKitcould you give full log?17:45
osum4esti was never  able to disable selinux so i wouldn't be suprised, one sec17:45
T4<adampigg> @eugenio_g7 booting that initramfs on real device just reboots after  a while...17:49
T4<adampigg> i wonder if i can get an frame buffer console?17:49
TheKit - I wonder why was it never merged17:51
osum4estso just ignore the error? not sure if it's a problem but i don't have any files in the /dev/__properties/ folder, which is that it looked to me like what that method was doing17:53
TheKitosum4est, you need to modify Android init to ignore the error17:53
T4<eugenio_g7> @adampigg: have you touched /init_enter_debug2 in the rootfs?17:53
TheKitwith either that PR or -  *fsetxattr_failed = false;17:54
osum4estright, right. just wondering if there were supposed to be files in the properties folder. but yea i'll do that17:54
T4<adampigg> @eugenio_g7 i havnt...17:54
TheKitbut I can't remember which is the "canonical" solution to that issue or find it in Xperia XA2 branches17:54
T4<eugenio_g7> @adanpigg: try.. perhaps (hopefully!) it's further up the boot process17:55
T4<eugenio_g7> if you get telnet up at port 2323 I'd take that as a good sign!17:55
T4<MarcoDS_bit> Could the problem here be my hardware adaptation as I have an opengl adaptation not found error?
T4<MarcoDS_bit> I mean, idk what I should change anyway18:00
malTheKit: if people update libhybris submodule manually there is debugging support for android side libs18:02
TheKitmal, but why, droid-hal-init is pure Android binary18:03
malwell does it give reasonable output when run via gdb?18:03
malalso strace might give some hint to what is happening18:04
malTheKit: hybris-15.1 should match xa2 repos18:04
malunless I messed up something18:05
mal@elros34 in case some device needs to mark boot as successful then it needs the thing that xa2 uses, the symptom of that issue is that every 5th boot or so kernel needs to be reflashed18:28
T4<MarcoDS_bit> Every time we install sailfish we have to write this within the first ~20 seconds "systemctl mask usb-moded.service" how should we fix/include that?19:31
malwhat does it do otherwise?19:32
T4<MarcoDS_bit> It freezes19:34
maleverything freezes or just usb?19:43
maldoes that device boot to UI?19:43
malalso which android base?19:45
T4<MarcoDS_bit> Telnet stops working, UI doesn't work but we have an "opengl es adaptation not found" so I guess I have done something wrong in the hardware adaptation. Also I haven't tested the latest build but apart from the errors about usb and lipstick one tester had charging led working and sounds while pressing physical buttons19:48
T4<NotKit> @MarcoDS_bit when you run systemctl mask, you can see the symlink it creates19:57
T4<NotKit> you can put same in sparse folder of hybris/droid-configs/sparse/ until you get UI fixed19:58
T4<NotKit> however, according to, you have droid-hal-init crashing which is first thing to fix19:58
mal@MarcoDS_bit in $ANDROID_ROOT/hybris/droid-configs run ln -s /dev/null sparse/etc/systemd/system/usb-moded.service19:58
malthen rebuild droid-config package20:03
T4<MarcoDS_bit> Ok thank you20:06
T4<MarcoDS_bit> Also, what would cause droid-hal-init to crash usually?20:06
TheKitJan 02 01:01:31 Sailfish droid-hal-init: Failed to initialize property area20:07
TheKitsee what we talked with osum4est before20:07
T4<MarcoDS_bit> Ok fine20:07
osum4estTheKit, i modified it, but it's still not working. this is because the initialize_properties call fails, because it doesn't have access to /system, since its not mounted. /system and /vendor are not mounted by default for me. i added /vendor to my fstab, but not sure about system. mounting /system causes there to be recursive symlinks (lke /etc/system points to /etc/system)20:08
osum4estthere is a folder in the system partition named "system" do i need to mount system somewhere else and symlink /system to the systempartition/system? the system subdirectory has the correct path to the plat_property_contexts file, and no symlink recursion20:09
TheKitprobably your device is intiramfs as roto device, so you need to mount folder named "system" over /system20:09
TheKit*as root20:09
osum4estok. can i add system to my fstab, and have it mounted to "/system_raw" and make a symlink at "/system" that points to "/system_raw/system"? i tried that earlier but i results in a bootloop (which i believe was due to selinux)20:10
osum4esti'm not sure how to directly mount the system folder to /system20:11
osum4estok here's the log when i properly mount /system. it looks like it gets further now, but the phone freezes and i can't telnet in (dones't bootloop this time for some reason). i had to turn on logs across reboots to get it. looks like now theres a ton of "failed to set context" and SELinux stuff now. where should i go from here?20:44
malso you didn't have /system mounted?20:45
osum4esti had system mounted to /system_raw and a symlink /system that pointed to /system_raw/systme20:45
malor did you have incorretly mounted, some 15.1 device could have the new system partition content which need bind mounting20:45
malosum4est: can you show your fstab20:46
osum4esti added in mkdir's to the area_init method and that method completely finishes with the current setup. yes, one sec20:47
osum4esthere's my current /etc/fstab:
malosum4est: not that, the fstab from android20:49
mali.e. in your device repo20:49
osum4estmal, i have multiple. heres fstab.common:
malthe first mountpoint is / which means you need to add to droid-hal spec this: %define makefstab_skip_entries /20:52
osum4estto rpm/droid-hal-marlin.spec?20:53
malthat should make the symlinking of system partition automatic20:55
malfor reference20:56
osum4estthanks, rebuilding now. i still haven't been able to set the kernal config for the selinux bootparam with selinux=0. is that not going to be a problem?20:58
malwhy is it not working?20:59
osum4estsetting the bootparam=y and adding selinux=0 prevents me from even reaching the init script20:59
T4<eugenio_g7> piggz: cool stuff! What happens when you let the boot continue?21:03
T4<elros34> osum4est: have you tried CONFIG_AUDIT=n instead?21:04
osum4estelros34, instead of the bootparam or with the bootparam?21:04
T4<elros34> audit=n should disable selinux also so it doesnt matter21:05
malsome device don't like that21:05
osum4esti'll give it a shot, thanks21:06
T4<adampigg> start, but no graphics yet21:26
T4<adampigg> Im.also getting a reboot21:26
T4<adampigg> Might be watchdog21:26
T4<adampigg> But, atleast it boots!21:26
T4<NotKit> @adampigg is it PinePhone?21:30
osum4estmal, added that to my spec file and rebuilt but system isn't mounted21:31
mal[nemo@Sailfish /]$ uname -a21:32
malLinux Sailfish 5.1.0-rc4-147279-g5735d3dc2c76-dirty #11 SMP PREEMPT Sat Apr 20 21:55:18 EEST 2019 armv7l armv7l armv7l GNU/Linux21:32
malosum4est: hmm, odd21:32
malosum4est: sorry, that lol wasn't meant for you21:33
T4<MarcoDS_bit> conflicts :021:33
osum4estmal, no problem haha. /etc/fstab looks the same. shouldn't symlinking it manually have worked though? (as a quick workaround)21:34
mal@MarcoDS_bit sb2 -t $VENDOR-$DEVICE-$PORT_ARCH -m sdk-install -R zypper rm hybris-libsensorfw-qt521:34
T4<adampigg> @notkit, no, xiaomi latte...21:35
T4<adampigg> Pinephone not arrived yet21:35
malosum4est: sure it would work but there is the existing method which should work21:35
T4<MarcoDS_bit> mal: thank you, btw we discovered ui boots for some seconds and then just a black screen, so yeah now I'm building with latest changes hoping for it to boot just fine21:36
osum4estmal, elros34, also setting CONFIG_AUDIT=n also prevents me from reaching the init-script21:37
malI have telnet to a qcom device running sfos on a patched 5.1 rc mainline kernel21:39
malnothing really works though21:39
T4Mic_a was added by: Mic_a21:44
osum4estso do you happen to know why i couldn't telnet in when i symlinked it manually? my notification light turned white like usual, just can't telnet in21:47
osum4estjournalctl when i symlink it manually:
T4<elros34> some init*usb*rc might overrides android0 usb configuration22:00
T4<eugenio_g7> @adampigg: if no watchdog pings are in the dmesg it could be indeed the culprit of the reboots22:02
osum4estelros34, there was an init.usb.rc in /vendor/etc/init, but removing it didn't seem to help :/22:12
T4<elros34> you may also try to mask usb-moded.service22:14
malosum4est: what android base was that again?22:21
osum4estlineage 15.1 for marlin22:21
malosum4est: did you do the symlinks mentioned in faq for 15.1 porting?22:21
malat least your journal looks like you didn't22:22
osum4estno, i didn't. where is that? in the pdf?22:23
osum4estelros34, masking that also prevents from telnetting in22:28
malin topic22:35
osum4estthe only stuff i see for 15.1 is the libhybris clone and the rc files that need to be copied to sparse/init. dont see any symlink stuff?22:42
malosum4est: sorry, I mean copying the files23:04
malosum4est: it was 14.1 that needed symlinks, 15.1 needs copying files23:04
osum4estah, ok. just did that, ill rebuild and give it another go23:04
malosum4est: just making sure you don't build useless stuff, -c is enough after that, and image build of course23:06
mal-c means config packages23:06
osum4estgot it, thanks!23:06
malyou can check --help (or was it -h) to see what it supports23:07
osum4estmal, alright, added those files. here's the new log same thing happens though, still can't telnet in23:27
malosum4est: are you sure you added those?23:32
malcheck line 2092 of your pastebin23:33
malhow did you get that log?23:33
osum4estmal, pretty sure. i've got 11 .rc files in hadk/hybris/droid-configs/sparse/init, then did -c, rebuilt the zip file, and reflashed it. to get the log i added system and vendor to my fstab, created the /system symlink, turned on logs across reboots, booted into sailfish for a bit, and booted into twrp and pulled it23:37
malosum4est: NOOO23:37
malosum4est: hmm, sorry, faq has an error, not sure when that happened, it used to be ok23:38
osum4estwait, did i do it wrong then?23:38
malvery odd23:38
malfixed it23:39
malthe link had the correct path23:39
osum4estalright, i'll try it again23:39
malbut telnet issue could be unrelated anyway, but that fix is needed anyway if you want to have UI working23:40
osum4estmal, awesome! that fixed the telnet issue! and droid-hal-init is no longer failing!23:56

Generated by 2.17.1 by Marius Gedminas - find it at!