Sunday, 2019-04-21

osum4estnew log:
osum4estwhat do i need to fix next? autologin, start-user-session, or user@10000.service?00:03
mala lot of android side services fail, which will prevent anything from working00:06
osum4esti get [TP]: [ERR]:fwu_start_reflash: Firmware image synaptics_bl71.img not available so i think i need the proprietary vendor drivers00:08
osum4estaccording to
malfocus on the android side issue, that is preventing UI and everything00:10
osum4estoh, i thought maybe the drivers were the cause of those issues. where in the log are they? and how do i fix them?00:11
mallook for "killed by signal 9" in the logs00:11
malosum4est: didn't you say you had issues with selinux disabling?00:12
malmaybe try it again00:12
osum4esttry disabling it again?00:12
malit won't work if selinux is enabled00:12
osum4estright. just to double check: i don't have to rebuild the whole zip and reflash for that right? i can just "make hybris-boot" and boot the image?00:13
osum4estmal, still nope... defconfig: device/google/marlin/marlin/
malso what does it do now?00:25
malflashing kernel should be enough00:25
osum4esti made my led turn blue in my init script - which works when selinux is enabled. with selinux disabled the light never comes on, shows the vendors screen for about 15-20 seconds, then reboots00:27
malis it really possible that booloader would prevent disabling selinux?00:28
osum4estif it helps at all my bootloader screen says: "Secure Boot: PRODUCTION"00:29
malno idea what that means00:31
r0kk3rzi would've thought those two things are unrelated00:32
malit's way too late already here so I'll go offline now00:32
osum4estalright, see ya. thanks for your help. hopefully i can get selinux disabled00:32
r0kk3rzosum4est: do you know if you can disable secure boot?00:37
osum4estr0kk3rz, haven't found a way, all web pages just talk about unlocking the bootloader, which i've already done of course. this seems like its some separate setting00:39
r0kk3rzhmm ok, ive only heard about 'secureboot' with UEFI bootloaders00:40
r0kk3rzi assume this is something else00:40
osum4estyea. i have to turn that off on my desktop/laptop for booting linux, no idea what it means on an android device though00:40
r0kk3rzin that case it means 'only boot signed images'00:41
r0kk3rzbut obviously it does boot sailfish, so who knows00:41
osum4estright? is there a way to disable selinux later in the boot process instead?00:42
r0kk3rzi dont think so00:42
osum4estmaybe it will only boot selinux disabled kernels if they're signed? but will boot any selinux enabled kernel? no idea how to sign it though00:43
r0kk3rzthis is too far into guesswork land00:44
r0kk3rzits simpler to expect that something in the kernel doesnt like selinux being disabled00:45
osum4esta few things metion a "ALLOW_DISABLE_SELINUX" flag, if you've heard of it00:46
r0kk3rzthere is also this:
osum4esti've had a few of those verity options a "n" but ill do the rest as well00:52
osum4estthat defconfig doesn't have the selinux bootparam enabled00:55
r0kk3rztry that then00:55
r0kk3rz!seen ghosalmartin00:55
merbotr0kk3rz: ghosalmartin was last seen in #sailfishos-porters 4 weeks, 4 days, 13 hours, 3 minutes, and 18 seconds ago: <ghosalmartin> mal: hows yours holding up :)00:55
osum4estchange the kernel command line to his, commented out all the verity stuff and didin't enable the bootparam. i booted, but didn't seem to help
osum4estenabling the bootparam causes the bootloop and doesnt reach init script01:01
r0kk3rzim not sure how far martin got with this device01:07
r0kk3rzlooks like not far01:07
T4chuvilin was added by: chuvilin05:54
T4<MarcoDS_bit> Just to know, is there any form of encryption in sailfish?08:09
vknechthello... a bit OT since it's in LOS but... with a new(er) kernel I'm getting fence timeouts related to kgsl and mdss_fb
vknechtany tip/insight about what to look for ?08:11
r0kk3rz@MarcoDS_bit: depends what you mean by encryption09:23
T4<MarcoDS_bit> r0kk3rz: user's data encryption09:41
r0kk3rzyou can do the sdcard...09:42
T4<MarcoDS_bit> Oh lol10:08
T4<adampigg> rinigus: is is possible to use the OpenCycleMap in offline mode in puremaps?  Theyre _soooo_ much better for walking/biking than the OSM ones11:39
*** OhYash1 is now known as ohyash11:50
vknechtso, what's the dhd submodule revision to use ? 9398ac1 ?13:54
malvknecht: why not latest?13:55
vknechttried to mimic revs/versions as in fp2, still getting unresolvable dhv due to  nothing providing libc++/libc/etc. for dhc13:56
malshow whole build log of dhv13:56
malvknecht: fp2 uses latest, I just forgot to push it to git13:57
Its_Mike__Hey sailors :)14:00
malvknecht: for some reason that is depending on 64-bit lib14:01
Its_Mike__Tried to build sf recently, but no gui at all. what's the problem? anything's changed with the new version?14:01
malIts_Mike__: it used to work?14:03
malyou need to get some logs14:04
Its_Mike__Yes, everything was ok with the previous versions14:04
malcan you telnet to the device14:04
malvknecht: there is no need to downgrade submodules14:04
Its_Mike__I guess so, should flash the broken zip file again...14:04
vknechtI'll try again with latest dhv14:05
Its_Mike__Ok got telnet. let's grab some logs14:12
malfirst grab journalctl -b14:14
Its_Mike__Doesn't sounds good...14:18
Its_Mike__Let me send them for you14:20
Its_Mike__and logcat:
Its_Mike__Can iptables and related things caused this problem?14:23
malIts_Mike__: did you make any other changes than building
malIts_Mike__: what was the previously working release?14:24
Its_Mike__Hmm... no? just updated submodules14:24
Its_Mike__It was
maloh, that old14:24
maldo you mean
Its_Mike__Uh sorry14:24
Its_Mike__yes :D14:24
malIts_Mike__: you are sure hadn't done any local changes on the device in previous version?14:26
Its_Mike__Yes I'm sure. even re-sync everything with my github repo14:27
malwhat android base?14:28
Its_Mike__cm 12.1 , 5.1.114:28
Its_Mike__Do you remember the firewall problem, mal?14:28
Its_Mike__Make some changes to kernel for fixing that14:29
Its_Mike__it works well on
malbased on those logs the android side display service dies for some reason which causes graphics to fail14:29
malget dmesg also14:30
Its_Mike__Uh that sounds bad :((14:30
Its_Mike__ Here you are14:31
malIts_Mike__: can you pastabin you droid-hal spec file and also check in droid-local-repo what the droid-hal devel rpm has in android-config.h file (the rpm is just a compressed package so you and open it with some extraction tool)14:34
Its_Mike__Sure, let me start ubuntu..14:35
malthe android-config.h /usr/lib/droid-devel/droid-headers/ in the rpm14:36
mal+is in14:36
malIts_Mike__: did you sync android source again or something? (don't if you didn't)14:37
Its_Mike__You mean repo sync?14:45
malvknecht: I know the reason for your issue but not the fix yet14:45
Its_Mike__after second fail, yes :(14:45
malIts_Mike__: did it actually sync anything?14:46
vknechtmal, ah, so I can take a break for now, thanks :-)14:46
Its_Mike__I don't think so, just revert some of my useless changes...14:47
vknechtbut it could be I forgot to commit the dhc submodule update14:48
Its_Mike__ droid-hal-*.spec14:50
malvknecht: the issue is the you have in that repo, there did you get that, we need to somehow exclude the dependencies from that, or maybe just %define __provides_exclude something14:51
mal*where14:52 android-config.h from rpm package14:53
Its_Mike__Uh sorry : ^^14:53
malvknecht: maybe ?14:54
malIts_Mike__: ok, that looks ok14:54
Its_Mike__The repo sync I've made you mean?14:55
malIts_Mike__: what changes did the sync revert?14:55
malIts_Mike__: I mean the android-config.h looks good14:55
Its_Mike__Uh Ok :)14:55
malif the defines were missing it could have explained the issue14:55
vknechtgot it from device iirc14:55
Its_Mike__Can't remember exactly!14:55
T4<MarcoDS_bit> Failed to initialize property area again :/
malIts_Mike__: one idea, downgrade libhybris to revision 20bcd5be11fcd82aa9e2e995c4bb37af9c94cbad14:57
Its_Mike__sure! let me try?14:57
malIts_Mike__: do you know how to do that?14:58
Its_Mike__Yes :)14:58
malit's just very common that some people have limited knowledge of git and how to build the a downgraded package with build_packages.sh14:59
mal@MarcoDS_bit have you ever had a build that didn't have that error?15:01
T4<MarcoDS_bit> mal: nope actually, I have applied all the suggested fixes from the faq but I still have this15:05
mal@MarcoDS_bit can you install strace on the device and try to run droid-hal-init with that15:06
T4<MarcoDS_bit> mal: mhh how to install/use it?15:07
Its_Mike__mal : I understand :) no problem15:07
mal@MarcoDS_bit at least it's possible to include it in the image and just run strace droid-hal-init15:08
maland the env stuff that is done in the droid-hal-startup script15:08
T4<MarcoDS_bit> ok I'm trying it15:08
Its_Mike__Got some problems while installing that version of libhybris, try to solve them...15:10
malIts_Mike__: what?15:10
Its_Mike__Let me send the log15:11
vknechtmal, adding "%define __provides_exclude_from ^%{_libexecdir}/droid-hybris/system/lib64/.*$" in droid-config-idol3.spec ?15:13
malIts_Mike__: sb2 -t $VENDOR-$DEVICE-$PORT_ARCH -m sdk-install -R zypper rm libhybris-tests-upstream-devel libhybris-libsf libhybris-libsf-devel15:14
malvknecht: something like that15:14
T4<MarcoDS_bit> mal: "strace droid-hal-init" makes my phone reboot to download/fastboot mode (?), and I get this
maldid you run the stuff from the script before, also you have to mask droid-hal before trying to run droid-hal-init manually15:17
Its_Mike__Done :)15:20
malsame behavior?15:20
Its_Mike__About to test it mal... will notice you15:21
T4<MarcoDS_bit> mal: what stuff from what script? I'm sorry for asking but could you guide me please? also droid-hal masked now15:25
malcheck droid-hal-init service for the script it runs and run everything in it15:30
T4<MarcoDS_bit> I ran everything, the result from "strace droid-hal-init" is the same, and that's strace /sbin/droid-hal-init
Its_Mike__Still have error while installation15:48
Its_Mike__The same log15:48
malIts_Mike__: really, exactly the same error?15:50
vknechthmm, dhv still unresolvable ; maybe rather use __requires_exclude ?15:53
malvknecht: that was my other idea15:54
malvknecht: some combination of those or something15:55
Its_Mike__Yep :(16:00
Its_Mike__Kind of annoying problem :D16:00
malso did the command I gave do anything?16:01
Its_Mike__Just removed those packages...16:02
Its_Mike__Let me try something...16:02
vknechtdhv built with only the __require_exclude, tho I see this line in dcd build log:16:04
vknecht[  129s] Provides: bluetooth-rfkill-event-configs droid-config droid-config-idol3 = 0.0.37-1.43.1.jolla droid-config-idol3(armv7hl-32) = 0.0.37-1.43.1.jolla obexd-configs16:05
malIts_Mike__: how did you build the downgraded libhybris16:05
Its_Mike__cd into libhybris directory16:05
malso you didn't build_packages?16:05
Its_Mike__then do git --reset-hard ...16:05
malonly part I'm interesting in is the exact command you used16:06
Its_Mike__and at the last used the libhybris spec with the build_package command to build it16:06
malexact command16:07 -b hybris/mw/libhybris -s rpm/libhybris.spec16:07
malnot sure what you are doing wrong because removing the the conflicting packages should fix the issue16:07
malso you need to give whole libhybris.log16:08
Its_Mike__Sure! let me try again...16:08
Its_Mike__to build...16:08
malvknecht: just a comment, you will get a very ugly commit log if you always tag all test builds instead of manually triggering the builds on OBS until it works and then making just one clean commit, it's probably just me but I like to have a clean commit history16:09
vknechtah right will keep that in ming ; seemed to me that touching service file wasn't always sufficient16:12
Its_Mike__Here you are mal, it's the whole log :
vknechtseems ok now libdiag doesn't appear in either log ; thanks a lot mal :-)16:13
malIts_Mike__: that claims you still have libhybris-tests-upstream-devel installed or it tries to install it, check droid-local-repo for that rpm16:17
malIts_Mike__: also check sb2 -t $VENDOR-$DEVICE-$PORT_ARCH -m sdk-install -R zypper if libhybris-tests-upstream-devel16:17
Its_Mike__On it mal, Thanks!16:17
Its_Mike__Removed it from local repo, rebuilding right now, hope it works16:21
Its_Mike__Same log I guess...16:26
Its_Mike__Got a random idea, maybe reinstalling platform sdk16:27
malno need16:27
Its_Mike__or installing libhybris manually16:28
malso you now made sure it's not installed and not in local repo?16:28
T4<elros34> just install it manually via sb216:28
Its_Mike__if you mean libhybris-tests-upstream-devel package, yes :) removed from everywhere16:29
malthe latest log has different error16:29
malI suggest you remove libhybris sources and reclone it and then checkout the correct revision and try again16:29
Its_Mike__@elros34 : maybe...16:30
malit really shouldn't be this difficult to build a simple package16:30
Its_Mike__:D Yep16:30
Its_Mike__Highly agreed16:30
T4<elros34> I just did the same yesterday, build_packages will not downgrade it for you do it manually16:30
malyou might have some remnants from old build in your libhybris source tree16:31
T4<elros34> sb2 -t $VENDOR-$DEVICE-armv7hl -m sdk-install -R zypper rm libhybris-libsf libhybris-libsf-devel libhybris-tests-upstream-devel16:31
T4<elros34> and then something like sb2 -t $VENDOR-$DEVICE-armv7hl -m sdk-install -R rpm -U —force libhybris-0.0*rpm libhybris-devel*rpm libhybris-libEGL*rpm libhybris-libEGL-devel*rpm libhybris-libGLESv2*rpm libhybris-libGLESv2-devel*rpm libhybris-libhardware*rpm libhybris-libhardware-devel*rpm libhybris-libsync*rpm libhybris-libsync-devel*rpm libhy16:31
T4bris-libwayland-egl*rpm libhybris-libwayland-egl-devel*rpm16:31
mal@elros34 looks at the latest log he gave, different error16:31
Its_Mike__Confusing, let me do that16:32
Its_Mike__mal it actually builds the package, it only can't install it on platform target, I guess...16:33
Its_Mike__what package provides ?16:41
malIts_Mike__: did you do what I said?16:46
Its_Mike__recloning the source?16:47
malnote that you need to use --recurse-submodules when cloning16:47
malIts_Mike__: I really don't know anymore what you are doing, it's impossible to get the same error in that case16:48
Its_Mike__Just get a bit confused mal16:48
maldid you remember to downgrade it after recloning? and also removing any stuff from droid-local-repo16:50
mal*any libhybris stuff16:50
Its_Mike__May I ask a noob question?16:51
Its_Mike__Sure , I've removed libhybris packages from local repo16:52
Its_Mike__I should run something like this : git clone --recurse-submodules "url" libhybris ?16:53
malyes, I assumed you did that already16:53
Its_Mike__and then reset to the right reversion ?16:53
malat least you said you did16:53
Its_Mike__Done it without --recurse-submodules tag16:53
Its_Mike__ok reclone it like what I said16:55
Its_Mike__now I should just run git reset --hard "tag" under libhybris?16:56
Its_Mike__" Just want to be sure I'm doing everything right :D "16:56
malyes, I told you already16:57
Its_Mike__Sorry for noob questions :)))16:57
Its_Mike__Rebuilding ... I'll kill myself if it won't works this time xD17:00
Its_Mike__O_O still can't install packages17:02
malwhat error now17:06
Its_Mike__let me pastebin them17:08
malare you sure the submodule was downgraded correctly?17:10
T4<elros34> Its_Mike: git checkout; git submodule update17:12
malwondering why the version of that package is not but also has some git hash in it17:12
Its_Mike__Let me see guys17:21
Its_Mike__Finally :DD17:38
Its_Mike__Reinstalled the platform sdk17:39
malthat is almost never needed17:39
Its_Mike__I know... but todays adventure with sailfish was too odd for me17:42
Its_Mike__Ok let me build rootfs and see if it comes to gui17:43
osum4estmal, still stuck on selinux... any other ideas? tried disabling audit, disabling selinux via bootparam, and disabling verity. also is it possible to use the halium kernel to boot sailfish (since halium supposedly supports marlin, maybe it will be able to disable selinux)17:56
malno idea if halium kernel is compatible17:59
malhave a look at halium kernel to see what they have done17:59
osum4estactually looks like  halium doesnt support lineage 15.1, only 14.1. no idea how marlin is supposed to work since lineage 14.1 isn't available. yea i'll take a look at their boot params and kernel config18:00
Its_Mike__Thanks a lot mal, gui boots finally !18:41
T4<austenite> hello, For anbox : Failed to mount Android data directory .Has anyone faced this?
osum4estmal, is there a way to disable selinux later in the boot process?19:05
vknecht@austenite, maybe give some context : which overlayfs patch (unless 3.18 which should have it already ?), which repo used to install anbox, lxc check result, ...19:07
vknechtosum4est, maybe setenforce 0, if this command is available ?19:09
osum4estafter booting into rootfs?19:09
vknechton workstations and servers yes, not sure how that applies to these embedded pixies19:12
osum4estyea just tried it, that command doesn't exist19:12
vknechthmm, lemme check something19:12
vknechtmaybe it's somewhere in there :
vknechtin libselinux-utils, if you can get to install it19:17
malosum4est: afaik no19:17
osum4estactually, how can i check if selinux is disabled? i may have gotten it to boot with the selinux bootparam set to yes19:17
vknechtgetenforce :-) but there's probably another way, maybe in sysfs or proc19:18
malosum4est: if droid-hal-init and android side services work then it's fine, check dmesg for failing things19:18
osum4estmal, boot-hal, hwcomposer, configstore-hal, and gralloc are still getting killed by signal 9, and still have a bunch of "SELinux could not set context" messages. dont see any errors in dmesg19:21
malare servicemanagers running, there should be 3 hw, vnd and normal19:22
osum4estlooks like vnd and hw are getting killed by process 6. journal:
malthat is very bad19:28
osum4estis selinux still not disabled or is it something else?19:28
malprobably not disabled19:29
maldo you have selinux=0 in kernel commandline?19:29
osum4estyes, the kernel parameters are also at the top of that journal log19:30
malosum4est: what did you change to make it boot now?19:30
osum4estnot sure really. i have verity disabled, bootparam=y and audit=y, and rebuild and reflash the whole image, rather than just the hybris-boot.img19:31
T4<elros34> make sure you check .config generated when you make kernel not your defconfig19:34
osum4estelro34, yea that has the bootparam enabled. it also has "CONFIG_SECURITY_SELINUX_BOOTPARAM_VALUE=1" but shouldn't that get overriden by the kernel param?20:05
malshow the whole dmesg20:07
malpreferrably after reboot so it has everything20:07
osum4estsure, one sec20:07
T4<elros34> dont know, anyway you should really try to get logs from that failing kernel either via ramoops (flash working kernel and boot failing so it will reboot to working ?) or serial console (jack)20:09
osum4estmal, dmesg: elros34: sure i could try that20:11
malosum4est: obviously from device, why would I want host dmesg20:11
osum4estoh... didn't even think about that sorry :/20:12
osum4estdevice dmesg:
osum4estsee anything in there? for some reason if i boot hybris-boot.img directly it doesn't work, but flashing the whole zip does. i could just be doing something completely wrong though20:36
malwell it looks like it's selinux issue but not sure why20:37
malcheck /proc/cmdline just in case20:37
maland zcat /proc/config.gz | grep SELINUX20:37
osum4esthuh, bootparam is not set for some reason20:38
osum4estyeah for some reason the hybris-boot.img included in the zip has the selinux bootparam unset20:43
malosum4est: you do know that you have to run -d to included new kernel20:46
osum4estnope, that would explain why. still back at square one though. i'll  grab the ramoops after the kernel reboots. would there be any point in disabling selinux in the base lineage install?20:53
malbase android has no effect on that20:56
osum4estmal, here's the ramoops: i booted the broken kernel using fastboot and let it reboot into the working flashed boot20:58
malwhere are the kernel sources for that device?20:59
osum4estit shows that selinux is disabled in there, but i have no idea why it's panicing21:27
malbased on backtrace it could be here but no idea why
malosum4est: quite near the end Fatal Error: NON_SECURE_WDT21:30
osum4estyea i saw that. there's literally only 5 wepages that mention it, and 4 of them are other logs haha21:31
osum4esti can try turning off CONFIG_SCSI_UFS_QCOM_ICE=y21:38
malno idea what does, worth a try21:41
osum4estbuild errors... ufs-qcom.c no longer has access to the ice functions21:48
T4<DSstill01> @sorbite thats because chinese guys like to change names of things21:53
T4<elros34> if pfk_kc_deinit is called anyway and it led to crash  so maybe don't init it at all (pfk_kc_init) or just disable CONFIG_PFK :
T4<NotKit> @MarcoDS_bit you need 64-bit Android strace21:57
steeevevknecht: did you manage to run anbox on your device or you have a crash in android rootfs or other (not specified)?22:01
osum4estelros34: interesting. Disabled CONFIG_PFK and i can now reach the init script with selinux disabled. i cannot reach rootfs though, because it cannot mount /data. because /dev/sda33 doesn't exist? there are no /dev/sda* files?22:03
osum4estthere a bunch of other stuff /dev/ (fb*, msm*, ram*, etc.) but none of my sda partitions22:04
osum4estnevermind, i had another setting disabled that shouldn't have been. i got into rootfs now!!! with selinux disabled!! thank you elros34!!!22:13
T4<elros34> awesome now try config_vt22:14
T4<elros34> also post what setting was disabled so others can benefits22:15
osum4estCONFIG_PFK=n was the fix, we should add that to the faq22:15
osum4estelros34, config_vt still does not work22:19
malosum4est: nice22:23
T4<elros34> afaik you will need it to get ui22:23
osum4estoh, i thought it wasn't actually required?22:23
T4<elros34> I mean config_vt22:23
malosum4est: what do you mean sda33 doesn't exist, it did before22:23
osum4estmal, ignore that part, i had another config setting disabled that i shouldn't have disabled. it works fine now with just CONFIG_PFK=n22:24
osum4estso i need CONFIG_VT=y for ui to work?22:25
T4<elros34> I think so, you can fix it :)22:27
malwhat is the problem with VT?22:27
osum4esti can't reach the init script with CONFIG_VT=y also, hw&vndservicemanagers are still getting killed: but no selinux stuff now22:29
malthe killing of those is usually selinux related22:30
osum4estwhat else could cause it? because selinux is definitely disabled now.22:32
malnot sure right now22:33
osum4esthere's the ramoops with CONFIG_VT enabled. this time i get "Fatal Error: RPM_ERR"22:39
*** Piece_Maker is now known as Acou_Bass22:40
*** Piece_Maker is now known as Acou_Bass23:34

Generated by 2.17.1 by Marius Gedminas - find it at!