Saturday, 2019-11-16

T42<翟豪淦 %lastname%> (Photo, 1255x679) "精准币圈菠菜棋牌粉" "非小号/mytoken惊爆价"1、币圈电报群/微信群炸群推广 2、twitter、facebook关注 3.国内外KYC身份材料出售  4、电报群/微信群/potato拉币圈活人 5、国内外交易所APP实名注册认证 6、电报群客服活聊天  7、批量注册国内外交易所实名账户,,出twitter、facebook、微信带粉03:21
T42<RK2019account> @翟豪淦 %lastname% [<reply to media>], AD04:22
rinigusmorning! for some reason, after update of sdk to, grep is obsoleted by patterns-sailfish-configuration-sdk-tooling and busybox-symlinks-grep. as a result, even libhybris compile fails with complains about grep -P06:54
malrinigus: local build or OBS?09:40
rinigusmal: local build09:40
rinigusobs seems to be fine09:40
malI know that local builds work with target, wondering what is wrong09:41
rinigusmal: but is it with PlatformSDK also updated using 'sudo ssu re; sudo zypper ref; sudo zypper dup'?09:42
rinigusmal: I have done that update09:42
malI will check09:43
T42<nitin_3> something else might be wrong09:43
T42<nitin_3> bcoz on local build I get this warning eveytime09:44
T42<nitin_3> Grep invalid option -- 'P'09:44
malrinigus: can you show whole libhybris build log09:44
T42<nitin_3> but the build goes on09:44
T42<nitin_3> doesn't fail bcoz of it09:44
T42<nitin_3> here is my log09:47
T42<nitin_3> At the beginning you can see that warning09:47
T42<nitin_3> but build goes on09:47
rinigusmal: sorry, no i cannot. It's deleted after rebuild, as I installed grep (and uninstalled couple of packages)09:47
malrinigus: :/09:48
riniguspatterns-sailfish-configuration-sdk-tooling and busybox-symlinks-grep09:48
malrinigus: you should remember to get logs always09:48
rinigusmal: I generally do. its just xz2 fails to boot now with 3.2.0 update and I am trying to figure out why. hence fresh rebuild. but I will bug you regarding it later, after I try with the fresh build09:50
rinigusas for failure of libhybris, I can try again a touch later with installed other packages and see if it fails for me. while the log above was similar to mine (it did continue to build), it had issues quite early in the build09:51
rinigusI am referring to by nitin_309:51
T42<dontatmepls> @elros34 the grep showed no results, so i did a find at the end, hope this helps10:14
rinigusmal: with image, I cannot boot xz2 anymore. this and 3.1. had `androidboot.selinux=permissive` and all worked fine in 3.1.sfos10:40
rinigusboot hangs on sony logo, no led. power button unresponsive10:41
riniguslooks like it fails rather early, as I have no logs either (I make snapshot from dmesg/journal and few other things as soon as a part of
rinigusxz2 should have all selinux patches applied, as its based on xperia 10 and I don't see anything special for these devices in latest mer commits10:44
rinigusI checked xa2 latest commits that were introduced as a path of 3.2.0. release, and xz2 does have those selinux policies there10:45
rinigusany tips?10:45
rinigusoooo, I have a journal if persistent is enabled...11:21
T42<adampigg> rinigus: if you had permissive in 3.1, thats a problem with 3.2, i had it with mido11:39
T42<adampigg> turn it off ... the new systemd explicitly turns on selinux if it is permissive, and breaks dhi11:39
rinigusadampigg: just managed to turn it off by setting SELINUX=disabled in /etc/selinux/config and it booted11:40
T42<adampigg> rinigus: cool ... i hadded selinux=0 to cmdnline11:40
rinigusadampigg: thing is, I had it in permissive in 3.1. as well. but we do get logs now with all avc's. so, it should be possible to fix it11:41
rinigusadampigg: according to manuals, permissive should let you get denials in logs, but it should still work11:42
rinigusadampigg: you mentioned 'dhi' as being broken. what's dhi?11:58
rinigusvknecht: that it probably does. I have log full of `Jan 30 23:06:55 Sailfish audit[1384]: AVC avc:  denied  { write } for  pid=1384 comm="systemd-udevd" name="pn54x" dev="devtmpfs" ino=564 scontext=u:r:kernel:s0 tcontext=u:object_r:nfc_device:s0 tclass=chr_file permissive=1`12:08
rinigusand similar12:08
rinigusbut I do wonder why would it disturb in 'permissive' mode. that mode was expected to log only, not to enforce it12:12
T42<elros34> @dontatmepls if "make modules" really doesn't contains fstab then find it in device/ and copy it somwhere to out/ so it will be found during droid-hal build12:26
T42<dontatmepls> aight :)12:29
malrinigus: note the permissive=1 in the errors, not errors but just log spam12:41
rinigusmal: not really spam in the sense that it would allow to make policy with those operations allowed :)12:42
rinigusbut something is different as soon as you disable it completely (as in
rinigususing that commit I can boot just fine12:43
malrinigus: if you don't disable selinux in android init in a certain way you can't have it enabled in systemd afaik12:46
rinigusmal: as far as I can see, I have the same patches and config on xz2 as xperia 10 does12:48
rinigusbut, there is still something which causes these diffences12:49
rinigusI guess, for now, I will keep it disabled and see how all works12:50
malrinigus: so you have the last system/core patch also?13:18
rinigusmal: the one with SYSTEMD_SELINUX define? that I do13:19
rinigus... let me just check again13:19
T42<dontatmepls> @elros34 as far as i can tell i see no difference, one thing ill say is im building the image using MIC fyi13:19
rinigusmal: the last applied patch in system category is 0039-hybris-Disable-SELinux-init.patch13:21
rinigusthat's the one that disables a mount of /sys/fs/selinux btw13:21
T42<elros34> @dontatmepls enable CONFIG_MEMCG_KMEM and CONFIG_AUTOFS4_FS (could have slightly different name in your kernel). Why you are building image if you didn't fix anything?13:25
T42<dontatmepls> i did build_packages first, but since it has issues building, someone told me to use mic after that instead13:25
T42<dontatmepls> and that works13:25
T42<elros34> but you still didn't copy fstab to out/target/product/$DEVICE/root/ because system vendor mount units are not created Also double check check-files in sdk target because I am pretty sure you made a typo if you don't have "installed but unpackaged files" error13:27
malrinigus: ok13:29
T42<dontatmepls> typo in where specifically?13:30
T42<elros34> here: remember to run —droid-hal after that13:31
T42<dontatmepls> done, im guessing you want the log again?13:35
T42<elros34> sure13:35
T42<dontatmepls> is this any better? i see some new lines appeared13:37
T42<elros34> CONFIG_MEMCG_KMEM is still not enabled..13:38
T42<dontatmepls> its enabled, and i checked dependencies13:39
T42<dontatmepls> oh maybe i should rebuild hybris boot first13:39
deathmistkrnlyng: what's the status of being merged to libhybris? it made my audio work OOB without needing a hacky symlink like
T42<elros34> of course if you change kernel you need to build hybris-boot13:40
T42<elros34> @dontatmepls you probably have create custom vendor.mount so copy it from tmp/units to your droid-config/sparse/lib/systemd/system/ and add  Before=vendor-firmware_mnt.mount vendor-dsp.mount vendor-bt_firmware.mount13:44
deathmist@elros34 I was about to say that! xD if the file isn't even created modify to match your device (may need the /system one too)13:45
T42<elros34> after that add /vendor to makefstab_skip_entries and rebuild droid-config and droid-hal13:45
deathmistin my case I didn't even have either one of those mounts in my fstab, this also allows proper control of mount order13:46
T42<elros34> @deathmist are symlinks to required?13:49
deathmist@elros34 pretty sure, this is how systemd works: loads stuff at specific boot stages from .wants/* files. I just copied what some other sony device iirc was doing and modified it slightly13:55
Thaodanrinigus: At which state is gpt-utils used?13:55
rinigusThaodan: when you finished the boot, to mark it successful.13:55
ThaodanAny idea how to fix a device where the device has black screen after booting to the vendor? It stays black for a min and than reboots.13:56
T42<elros34> @deathmist I ask because these simlinks are not created by default and mount units still works13:56
Thaodanrinigus: Is that  used even in the recovery?13:56
rinigusThaodan: for us, gpt-utils means that we have to use system and vendor compiled with that upstreamed patch13:56
rinigusThaodan: you have logs?13:57
ThaodanIf I had a way to access device at that state but sadly not.13:57
deathmist@elros34 huh, let me try it right now then on my device, guess those get loaded by something else or are mount units automatically ran? guess I need to do some more research too '=D13:57
rinigusThaodan: try to use this
rinigusafter failed boot, login into recovery and logs will be under /root of root partition13:58
Thaodaneven recovery doesn't work13:59
deathmist@elros34 nope, not booting to UI anymore after removing them from so they are definitely required14:01
T42<elros34> ok sorry so I was wrong. @dontatmepls  ^14:02
T42<dontatmepls> so what now14:02
T42<dontatmepls> was about to add it xD14:02
rinigusThaodan: in which respect it doesn't work? no rndis or reboots the same way?14:03
Thaodanit reboots14:03
T42<elros34> @dontatmepls  still same just don't forget to add symlink like here
ThaodanDo I need to flash dtbo.img and vbmeta.img when I flash boot?14:06
rinigusThaodan: usually, its ok without as long as you already flashed vbmeta before once with disabled verification flags14:21
rinigusThaodan: as for dtbo, I haven't seen any major effects with having it a bit out of sync14:22
rinigusI have done something stupid by removing a tag at github to changing release name pattern for
rinigusnow obs refuses to use that new release
rinigusremoving package and adding it didn't help either - I guess its caching it14:24
rinigusanyone knows how to reset obs state in tar_git?14:24
T42<elros34> change revision in _service file to HEAD?14:25
riniguselros34: nope, doesn't help. nor setting it to some other specific one14:27
rinigus(maybe for lower versions that are not interesting though)14:27
T42<elros34> maybe it doesn't like - in tag14:30
riniguselros34: I think it keeps its own cache of git repo at obs that is refreshed (from what I remember from reading tar_git code a while ago). and that doesn't play well with disappearance of a tag in between14:32
T42<elros34> rinigus: have you tried remove this 4.9.194-1.20191116  and adding something without - because that tag is still in your github?14:37
riniguselros34: no, that I haven't. let me try14:38
riniguselros34: thanks, that helped!14:40
T42<elros34> @monich looks like sms bug is gone in 3.2.0 but slow connection to provider it's still there. I send ofono logs right now. Do you think it could be improved? If I understand it right for now it waits 30s for failure. Maybe timeout could be decreased15:12
T42<dontatmepls> what have i done @elros3415:35
T42<elros34> you didn't add /vendor to makefstab_skip_entries like I told you and you copy system.mount which you shouldn't15:37
T42<dontatmepls> oh15:38
T42<dontatmepls> i did add /vendor btw, inside rpm/droid-hal-$DEVICE.spec15:40
T42<elros34> so you didn;t rebuild droid-hal15:43
T42<dontatmepls> already did15:44
T42<dontatmepls> still errors out15:45
T42<dontatmepls> @elros34 heres all logs after applying your changes, will send the droid-hal-$DEVICE.log in a sec16:45
T42<dontatmepls> maybe something changed now16:45
T42<dontatmepls> and heres the log of that16:46
deathmist@dontatmepls "/plat_property_contexts: No such file or directory" & "droid-hal-init: Failed to initialize property" is the issue you need to fix now, do you have any *contexts files in out/.../root/?16:52
deathmistlooks like the mount units for /system and /vendor function16:52
T42<elros34> c'mon CONFIG_MEMCG_KMEM  is still not enabled. Dependency failed for Droid mount for /system: did you add anything to system.mount? check it content16:52
T42<dontatmepls> CONFIG_MEMCG_KMEM is enabled in defconfig and all requirements are too16:53
T42<dontatmepls> also deathmist no, i dont16:54
deathmisthmm, you should have "/system/etc/selinux/plat_property_contexts" on the device then, right? make sure /system and /vendor are mounted for sure too from "mount" command output16:55
T42<elros34> no point to check property context if system mount fails, check content of /system_root16:56
T42<Lukapanio> hwservicemanager is braking16:57
T42<Lukapanio> @hacker12455  now usb isok16:57
T42<Lukapanio> SELinux : SELinux: Loaded service_contexts from:16:59
T42<Lukapanio> 07-25 15:04:14.510  7127  7127 I SELinux :     /vendor/etc/selinux/vndservice_contexts16:59
T42<Lukapanio> 07-25 15:04:14.510  7127  7127 E VendorServiceManager: SELinux: Failed to acquire service_manager context. Aborting.16:59
T42<Lukapanio> 07-25 15:04:14.523  7129  7129 E cutils-trace: Error opening trace file: Permission denied (13)16:59
T42<Lukapanio> 07-25 15:04:14.525  7126  7126 F         : Failed to acquire hwservicemanager context.16:59
T42<dontatmepls> i rebooted my device and now i cant telnet into it, however its recognized as the actual device name now17:01
T42<Lukapanio> same here17:01
T42<Lukapanio> but i disabled17:01
T42<Lukapanio> that17:02
T42<Lukapanio> just by making usb init empty17:02
T42<Lukapanio> so @elros34 what can i do with that?17:02
T42<elros34> with what? provide full journalctl, dmesg and logcat not just part of it. Didn't you have same issue with not starting or waiting for something hwservicemanager earlier?17:04
T42<nitin_3> @elros34 in case of weebgames systemd running in system mode(+PAM +AUDIT +SELINUX17:05
T42<nitin_3> shouldn't it be running as -SELINUX17:05
T42<elros34> I think selinux was enabled recently in many places in sailfish core components17:06
T42<Lukapanio> @hacker12455  recompile kernel with big buffer17:07
T42<Lukapanio> so i will be able to provide dmesg17:07
T42<elros34> @dontatmepls there should be  /vendor  which you added to skip_entires17:07
T42<dontatmepls> but whats about the device being recognized as actual device name now?17:08
T42<dontatmepls> i didnt even change anything17:08
T42<dontatmepls> and now my OS just says Xiaomi Mi 9t Pro and telnet doesnt work17:09
T42<elros34> try to flash kernel again17:09
T42<Lukapanio> @dontatmepls [and now my OS just says Xiaomi Mi 9t Pro and t …], first finde netd init rc and delete it17:10
T42<Lukapanio> and delete usb init17:10
T42<dontatmepls> @elros34 just making sure, so i dont edit files aimlessly - youre speaking of the rpm/droid-config-$DEVICE.spec file, right?17:11
T42<dontatmepls> because there it is17:11
T42<elros34> and you confirmed that droid-hal-$device*rpm (in droid-local-repo/) doesn't contains vendor.mount but droid-config*rpm does? How about you upload your droid repos17:13
T42<Lukapanio> logcat
T42<dontatmepls> @elros34 im afraid i dont understand17:15
T42<elros34> just unpack these rpms and check which one contains vendor.mount or on device: rpm -qf /lib/system/system/vendor.mount17:17
T42<Lukapanio> journal17:18
T42<Lukapanio> @elros3417:18
T42<elros34> these logs are useless, you are waiting to long to grab them also you should use /usr/libexec/droid-hybris/system/bin/logcat17:20
T42<Lukapanio> @elros34  like that?17:27
T42<dontatmepls> @elros34 you wanna look at my droid-hal-$DEVICE repo?17:28
rinigusnitin_3: systemd is starting from 3.2 compiled with selinux support. As a result, you see those +SELINUX options. To disable it in the system, either boot option can be used or setting in /etc/selinux17:38
T42<nitin_3> good to know17:41
T42<nitin_3> Thank you rinigus17:41
T42<dontatmepls> @elros34 "sh-3.2# rpm -qf /lib/systemd/system/vendor.mount17:50
T42<dontatmepls> droid-hal-raphael-0.0.6-201911161603.armv7hl17:50
T42<dontatmepls> droid-config-raphael-1-1.armv7hl17:50
T42<dontatmepls> "17:50
T42<dontatmepls> i think vendor is mounted in there since i can access it in sfos17:51
T42<elros34> but as you can see it's provided by droid-hal but it shouldn't be17:52
T42<dontatmepls> seems like the devices vendor, but lemme double check17:53
T42<dontatmepls> ill add a file to the real vendor in twrp and list it17:53
T42<elros34> I have no idea what twrp has to it, it's rpm package17:57
T42<dontatmepls> it has mounted the real vendor under /vendor with no problems17:57
deathmist@dontatmepls well clearly 2 packages are providing a single file on your device, no clue how that even works at all as it should've errored already early during mic / droid-hal build18:02
T42<dontatmepls> my pc is cursed xd18:02
T42<dontatmepls> so what to do now18:03
T42<Lukapanio> @Lukapanio [], @elros34  like this18:05
T42<Lukapanio> ?18:05
deathmist@dontatmepls that needs to be resolved and only provided by droid-config, but since it somehow works, give output of "mount". afterwards you can probably try "ln -s /system/etc/selinux/plat_property_contexts /plat_property_contexts" and reboot18:09
T42<elros34> @Lukapanio how about full journal, dmesg?18:09
T42<dontatmepls> deathmist i found the solution already, will do what you said rn, one sec18:10
T42<dontatmepls> deathmist heres mount18:15
T42<dontatmepls> also that ln command and reboot worked, what now18:18
deathmist@dontatmepls what do you mean "worked", do you get GUI now or what? give journal again18:19
T42<dontatmepls> no i dont get gui, but it didnt prevent boot, thats what i mean :p18:19
T42<dontatmepls> also will do18:19
T42<dontatmepls> here ya go18:24
deathmist"/plat_property_contexts: Too many symbolic links encountered" at least error is different, run "readlink -f /system/etc/selinux/plat_property_contexts" so we see where the file actually is18:28
deathmist@dontatmepls oh right, your /system is actually mounted on /system_root, which means "ln -sf /system_root/etc/selinux/plat_property_contexts /plat_property_contexts" should do it18:33
T42<elros34> not really, there are 2 mount units: system.mount and system_root.mount and system.mount fails thats why I asked him to check content of them and /system_root directory18:36
T42<dontatmepls> OHHHH @elros34 i misunderstood you sorry18:38
T42<dontatmepls> it has to be /system because /system_root is just a clone of root when on android roms iirc18:39
T42<dontatmepls> yeah sorry doing it now18:39
T42<dontatmepls> after doing the system.mount system_root disappears19:13
T42<dontatmepls> and still no system19:13
T42<dontatmepls> @elros3419:14
T42<dontatmepls> nvm system_root suddenly appeared19:15
piggzr0kk3rz: still use a BT keyboard? wanna try the new taskswitcher?19:44
deathmistI assume "hwclock --show" works on most ported devices? I just get "No usable clock interface found." when using "--verbose" :/  (ran as root or not, same output)20:12
deathmistI can get rid of the constant RTC / dsme failure call trace spam from logs when reverting some RTC sources back to 15.1 ones, but it doesn't help to make it work any better really20:15
T42<elros34> no, I guess because rtc can be opened only by one process and it's already used by dsme20:15
deathmistoh, so it should be working with the modified sources? I only get 2 lines after boot with them: "registered qpnp_rtc as rtc0" & "setting system clock to ..." (some 1972 date)20:17
T42<dontatmepls> Well I'll stop my stuff for today20:19
r0kk3rzpiggz sure!22:19
T42<adampigg> R0kk3rz,.on openrepos22:23
r0kk3rzwhats new?22:27
T42<adampigg> Settings page, built into jolla-settings22:28
T42<adampigg> Allows u to select device,.and orientation lock settings, aswell as service management22:29
r0kk3rzoh nice22:34

Generated by 2.17.1 by Marius Gedminas - find it at!