Monday, 2019-04-29

osum4esthello everyone! still trying to get the android services to stop crashing. i looked for missing rc files, but the only ones i could find where the init.zygote ones, which i copied over. there is not an odm mount in android. lineage's init.rc also looks for init.marlin.rc, but that doesn't exist in android. here's a list of all my rc files, while booted into android: system and vendor are00:57
osum4estmounted in sailfish with those rc files. here's the journalctl:
r0kk3rzwhich services are crashing?01:00
osum4estservicemanager, vndservicemanager, hwservicemanager, hwcomposer, boot-hal, and bunch of others01:02
T4southern_dust was added by: southern_dust01:10
malosum4est_: have you tried the selinux_stubs patch I linked there a few days ago08:51
malosum4est_: apply this patch to $ANDROID_ROOT/external/selinux_stubs and then rebuild hybris-hal and of course repackage it with " -d"08:52
ghosalmartinosum4est_, i still have my pixel with the sfos rootfs on it, ill fire it up and see what i did with the rcs09:24
ghosalmartinalthough mac and telnet seems to be shite09:25
malthe issue is probably not .rc files09:26
ghosalmartini was looking at the services he had failing and I think successfully started09:28
ghosalmartinalthough am not 100% sure, its been a whiel09:28
malghosalmartin: most likely that selinux stubs issue, at least one other use had the same issue and that fixed it09:51
ghosalmartinmal: hopefully it does, cause am on a shitty mac i cant seem to do anything useful. ill have to wait till i get home to see what state this rootfs is in09:55
malthe error is quite obvious from backtrace in logs10:02
ghosalmartinahh, hows has the new bluebinder been running? ive been out of the loop for a whileee10:05
ghosalmartinalso hows anbox, was anymore progress made or isnt it just there as a poc10:06
malghosalmartin: bluebinder works quite ok10:38
Mister_Magistermal: so i should update android_system_core in older adaptations? something else?11:03
Mister_Magistermal: well android_system_core didn't change since ages11:05
malMister_Magister: ?11:08
Mister_Magistermal: ealier you said something about updating old adaptation (rebuilding dhd) and i'm wondering which repos i should update11:13
Mister_Magisternothing changed so no need to rebuild hal11:21
malMister_Magister: no idea what you are talking about11:29
Mister_Magistermal: nevermind then11:34
Mister_Magistermal: btw have you seen wlan log? anything caught your eye?11:37
malnothing obvious reason, need to check more some other time11:38
Mister_Magistermal: latest gstreamer is supposed to fail to build11:45
malMister_Magister: no, can you give a link11:49
Mister_Magisterone second11:50
malmaybe reinstall target11:52
Mister_Magisteri just installed it11:52
Mister_Magisterit's fresh target11:52
malMister_Magister: did you also update sdk?11:52
malplatform sdk11:52
Mister_Magisteri thought it's up to date enough11:52
malno, it should be also updated occasionally11:53
malsdk-assistant update or something like that11:53
Mister_Magisterbut it's for updating targets it seems11:53
malother option is to run just sudo zypper ref and sudo zypper dup in platform sdk11:55
Mister_Magistermy sdk is
Mister_Magistermal: ye that's what wiki says11:55
Mister_Magistermal: nope updating didn't help12:09
malnot sure what is wrong, how did you update your target? manually or sdk-assistant?12:20
maldid tooling get updated properly?12:20
Mister_Magisteri removed tooling and target and reinstalled12:24
Mister_Magisterjust for sake of eliminating update errors12:25
Mister_Magistermal: you said i need lineage 15 for roma bt?13:53
ghosalmartinMister_Magister, yeah guessing its to use the bluebinder instead of attempt to init it directly13:55
Mister_Magisterghosalmartin: okay, mission: port lineage 15 using lineage1413:55
ghosalmartinMister_Magister, ah dayumm, what device?13:55
Mister_Magisterghosalmartin: moto x force my beloved one13:56
* Mister_Magister alternative mission: port mainline (doable)13:56
ghosalmartinyeah I need to do bullhead next, thankfully that already has lineageos 15.113:56
malMister_Magister: 15.1 is easier for rome chips, nothing really prevents it from working in older based but requires quite much work13:57
malMister_Magister: I have bluetooth working in mainline kernel for a qcom device13:57
Mister_Magisterghosalmartin: im kinda fed up with making ports for no people (without reward u loose motivation) but i love my x force13:58
Mister_Magistermal: there is mainline for some nokia phone that one guy is porting to oneplus 2 that i could adapt to x force (screen should be working and stuff)13:59
Mister_Magistermal: but mainline will come when i buy more x forces. Wanna buy a lot of them and live with them forever13:59
maloneplus one i.e. bacon has quite ok-ish mainline support13:59
malso does nexus 514:00
ghosalmartinmal: do you have rome chipsets working with mainline?14:00
malbut mainline does have the driver for those14:00
Mister_Magistermal: my target is to have some os running on mainline on my x force <314:01
malthe issue is more in bluez than kernel14:01
malMister_Magister: prepare to spend quite many days or weeks to get it running reasonably14:01
malalthough now it's easier because there are example for few devices14:01
malMister_Magister: which platform does that device have?14:02
mallike msm8974 or something14:02
Mister_Magistermal: im prepared to wait months to get it working14:04
Mister_Magisterbut as i said when i get multiple devices14:04
Mister_Magistermal: 899414:04
malMister_Magister: at leats msm8994 has some support in mainline14:06
Mister_Magistermal: ye from this project looks like screen works so14:07
riniguskimmoli & mal: I am rebuilding OTA with libhybris .25. If it works, I'll walk along the versions to determine which broke the boot15:30
T4<adampigg> Rinigus,.are _better_ maps available in osm scout server?15:35
rinigusadampigg: what in particular better?15:35
T4<adampigg> Rinigus, well, for fell walking/mountain biking,  the tiles from opencyclemap are far far better, with height maps and footpaths etc....15:39
rinigusadampigg: while you specify: these maps are combination of several parts: import scripts, styles to show them, and libraries used for showing/searching/routing. while libs for routing and showing are done by others, search, import and styling is done by me. routing needs to be updated, same with mapbox gl - this is planned in short term. import/style - as feedback from users comes, I'll try to adjust. search is15:39
rinigusplanned to15:39
T4<adampigg> The osm maps are only good for roads15:39
rinigusadampigg: opencyclemap is using osm, as far as I know. so, question is then in terms of import and style15:40
rinigusadampigg: height map is separate dataset, but can be added15:40
T4<adampigg> Its the detail such as footpaths / bridleways etc in the country id really like, when you are most likely offline too with no signal15:41
rinigusadampigg: to proceed with it, let's bring up an issue at osm scout server. its actually there
rinigusadampigg: so, please provide with some examples. ideally, using url, so its easy to examine15:42
T4<adampigg> Rinigus,  screenshots of different detail level?15:43
rinigusthen I can find missing data and target it15:43
rinigusadampigg: you could do screenshot, but please also add corresponding location url at osm. this you could get as "share location" in pure maps15:44
rinigusadampigg: to avoid confusion, please open new issue at server's project15:45
T4<adampigg> Ok15:46
rinigusadampigg: in general, I wanted to improve it, just have too many other aspects to look after15:47
T4<adampigg> Rinigus, as.we all!16:32
riniguskimmoli & mal: plan failed - I cannot boot with .25 either :( . looks similar to .31, except there is no green blinking. couldn't get any connection to it either...17:00
malrinigus: ssh didn't work?17:06
rinigusmal: nope - linux didn't managed to make any device with it, as far as I could see. as a result, no net for it.17:12
rinigusbut never used it like that, restored phone to 3.0.1 and will look inyo how to get ssh into it via usb17:13
T4<adampigg> Mal, what is droid system device?17:13
mal@adampigg that is used when packaging system partition in rpm17:14
rinigusmal: how I am expected to ssh into device? is there something' that I should do on PC?17:14
malrinigus: so nothing in host dmesg?17:15
rinigus^mal - on linux pc17:15
malrinigus: yes, do you see anything in the dmesg of you computer about usb device17:16
rinigusmal: not much good. this is after insertion of 3.0.1 (same for 3.0.2):
rinigusmal: that's testing from OBS, if it matters.17:25
rinigusmal: also its possible that my linux pc kernel is missing some drivers. its gentoo and kernel options are selected on install17:27
malrinigus: only difference between devel and testing is telnet17:27
malrinigus: could be17:27
rinigusmal: let me check with ubuntu vm17:28
rinigusmal: gotcha! since I have one of these magnetic charger connectors, I can insert it 2 ways. as expected, I did it wrong. now it detects it as a storage. how do I get it to some net mode?17:35
T4<DSstill01> does 4.4 base work with sailfishos 3.0?17:40
malalthough occasionally kernel might need some small patching if it's very old kernel17:54
malwhich device?17:54
malrinigus: are you sure the device didn't reboot to recovery or something17:55
T4<DSstill01> no, i was joking ahah18:00
rinigusmal: here is pc dmesg when I boot onyx - . I can use it as normal on SFOS. on pc, that mer boot device is not seen via lsusb, only google device18:00
T4<DSstill01> i dont have time to port android+sailfishos18:00
rinigusmal: so, its not recovery. this is on 3.0.118:00
malrinigus: are you sure, the dmesg had a disconnect and then the product name changed18:02
malit might not be recovery but some other mode if the device failed to boot18:02
rinigusmal: appart from that disconnect, I don't see any indication that I am in recovery. its "normal" mode where we are on onyx, at least on my device. is there a way I can distinguish? in recovery, I would expect twrp (that I get if I boot into it), not sfos18:09
T4<DSstill01> since all versions of aliendalvik are for 64 bit devices...anyone has a 32bit device with aliendalvik and an android base higher than 6.0 or equal, that know :D18:28
wdehoogmal: I updated my libhbris from .25 to .31 and after reboot no UI anymore. see log
wdehoogmal: any idea what I can do next to investigate this?19:10
malcheck also logcat, and is lipstick running?19:12
wdehoogat 384 Apr 29 21:02:11 Sailfish lipstick[1758]: [W] unknown:0 - EGLImage allocation error19:13
malwdehoog: and that is android 5 base (hybris-12.1)?19:15
wdehoogmal: yes it is19:15
T4<adampigg> s/ah/fuck i know whats wrong19:16
mal@adampigg where?19:16
T4<adampigg> mal: no, it was a joke19:16
malwdehoog: does test_hwcomposer also fail?19:18
malor minimer?19:18
wdehoogmal: lipstick runs, should I kill it before trying test_hwcomposer?19:19
malsystemctl-user stop lipstick19:20
wdehoogtest_hwcomposer starts and stops. no frightning messages. don't know what to expect19:21
wdehoogwhere can I find minimer?19:25
rinigusmal: tested update to 3.0.2 with libhybris .25 again and it worked for some reason. don't think anything was significantly different. I just enabled wifi to be able to login on local net and used shutdown instead of restart after OTA. go figure...19:33
typeofHello everybody, does the old porting way can be used instead of halium?19:39
malwdehoog: did you run test_hwcomposer correctly with EGL_PLATFORM=hwcomposer test_hwcomposer?19:42
malrinigus: worked completely or what?19:42
maltypeof: what do you mean? as far as I know nobody has used halium with sailfish19:43
wdehoogmal: yes, found minimer, seems to run ok, nothing on display19:43
rinigusmal: yes, completely with gui and everything. now I am rebuilding .31 and will update to that. I wonder if I'll see the same lipstic error as wdehoog does.19:43
rinigusmal: ... assuming that wifi holds19:44
malwdehoog: anything in log?19:44
typeofAh wrong group LOL sorry19:44
typeofI though...nvm. Anyway sailfish is great19:44
malwdehoog: also check logcat after you run that19:45
wdehoogrinigus: wifi remains working for me19:45
wdehoogmal: no traces of minimer in journal19:46
wdehoogmal: logcat is full of those repeating qmi errors:
malwdehoog: but nothin related to hwcomposer in logcat19:48
malso the issue has to be in lower level in android side19:49
malwdehoog: check also dmesg19:53
T4<elros34> iirc test_hwcomposer worked for  only minimer/lipstick fails for libhybris >0.25.19:54
wdehoogmal: I think dmesg is full of stuff due to missing sim card:
T4<elros34> best is to mask user@100000.service otherwise lipstick will try to restart constantly19:55
malwdehoog: missing card doesn't cause those19:58
malwdehoog: that mean modem is restarting all the time because android side in bad state19:59
mal@elros34 that sounds odd20:00
T4<elros34> I will try again later to make sure20:01
mal@elros34 and none of the versions between .25 and .31 worked?20:01
T4<elros34> I just tried 5 or 6 version between .25 and master and nothing really worked20:02
malso then it has to be something in .26?20:03
rinigusmal: I can reproduce issues with .31. boots without gui / ssh is possible from wifi. from journal -
T4<elros34> like I said few days ago, this one cause it
piggzrinigus: what further detail would be useful for ?20:05
rinigusmal: logcat with lipstic sig=11:
mal@elros34 yes, but have you checked how much changes submodule had in that, not easy to find the cause20:07
malrinigus: run it in gdb if you can20:07
T4<elros34> no, I will try20:08
rinigusmal: how? its complaining 'bout "XDG_RUNTIME_DIR not set in the environment"20:11
malrinigus: run as nemo20:11
rinigusmal: as nemo - Failed to create display (Connection refused)20:12
riniguspiggz: looks good, will see what can I do20:12
malshow the whole command you use20:12
rinigusmal: /usr/bin/lipstick20:13
rinigus(same with 'gdb /usr/bin/lipstick') as nemo20:13
malcopy the whole command from service file20:15
maland the env vars etc20:16
wdehoogmal: if I do that strace gives: connect(3, {sa_family=AF_UNIX, sun_path="/run/user/100000/../../display/wayland-0"}, 43) = -1 ENOENT (No such file or directory)20:20
malwdehoog: how did you run it?20:20
wdehoogstrace -f /usr/bin/lipstick -plugin evdevtouch:/dev/input/event0 -plugin evdevkeyboard:keymap=/usr/share/qt5/keymaps/droid.qmap --systemd > ls.log 2>&120:21
wdehoogmal: but it exits immediately while when starting as the service it kept running20:21
rinigusmal: backtrace at
rinigusI have to go now, sorry. will be able to continue tomorrow night20:22
rinigusmal: thank you for the help!20:23
osum4est_mal, thanks! that worked, the servicemanagers are no longer crashing. there are a few more processes getting killed, but much, much less than before. still no ui, though. here's the new log:
malosum4est_: show output of "mount" on the device, also output of "systemctl"20:37
malosum4est_: the log looks like it doesn't find some firmware files20:38
osum4est_mount: systemctl:
malosum4est_: do you have a folder /vendor/firmware20:40
osum4est_mal, yes, and it has a bunch of stuff in it20:41
maltry running "ln -s /vendor/firmware /lib/firmware" and then reboot20:41
osum4est_mal, didn't seem to help. new log:
malosum4est_: well you can see that the firmware loading errors are gone20:47
malopen you droid-hal devel rpm in droid-local-repo and check the android-config.h file in there20:48
osum4est_mal, how do i do that? i tried tar xvf ~/hadk/droid-local-repo/marlin/droid-hal-marlin/droid-hal-marlin-devel-0.0.6-201904292005.armv7hl.rpm but i get "This does not look like a tar archive"20:57
malif you don't want to a normal gui way for uncompressing that (in ubuntu it's simple) then you can use rpm2cpio some.rpm | cpio -idmv20:59
osum4est_mal, thanks. there's not much in the file. just some comments and an include
*** pketo_ is now known as pketo21:11
malosum4est_: that is the problem, assuming your device is qualcomm, check faq lines 886-893 (linked in topic)21:12
malosum4est_: before running I suggest you add the symlink I mentioned earlier to your config repo21:16
osum4est_mal, how do i add it to my config repo? just add it directly to rpm/droid-hal-marlin.spec or somewhere else?21:23
wdehoogmal: just tried .26 this one also does not work for me, also with EGLImage allocation error21:26
malosum4est_: go to hybris/droid-configs folder and mkdir -p sparse/lib and then cd sparse/lib/ and then ln -s /vendor/firmware firmware21:33
malosum4est_: never confuse droid-hal and droid-config repos21:34
malosum4est_: so basically in config repo anything you add to sparse folder will end up on device21:34
osum4est_got it. that makes sense, thanks!21:35
maldid you understand the other thing you need to do based on faq?21:35
osum4est_yep i added the android_config line to my .spec file in the rpm folder21:36
malosum4est_: hopefully it knows to rebuild enough after adding that define21:43
osum4est_male, can i just delete the droid-local-repo and rebuild the packages to make sure?21:44
malosum4est_: no, deleting that won't help, if you want to be sure, remove folders libhybris and qt5-qpa-hwcomposer-plugin from hybris/mw/21:45
malafter that run build_packages.sh21:45
T4<elros34> sorry I was wrong test_hwcomposer with master libhybris also cause device reboot, continue testing21:49
osum4est_mal, i don't have a libhybris folder in hybris/mw. i do have the qt5 one and a bunch of others though. is this a part of the problem?22:01
malosum4est_: ah, then it's in external/22:15
malosum4est_: so remove external/libhybris and then run build_packages.sh22:16
osum4est_mal, sorry, took awhile to rebuild and flash and everything. here's the new log after adding those lines to my spec file:
malok, needs some more debugging but it's quite late here so some other time22:57
osum4est_alright, thanks for all your help. is it going to do with not being found or am i not to that stage yet22:58
osum4est_also btw checked the rpm and yes those two defines got added22:58
T4<elros34> mal: this commit: cause kmap2qmap in sdk segfault and test_hwcomposer segfault in device (upstream cause device reboot)23:00
T4<elros34> osum4est: you could at least try test_hwcomposer (faq). I don't think is so important23:01
osum4est_elros34, segfaults, and says that libraries,, and are not found23:02
r0kk3rzservices look happier now, i see progress23:03
T4<elros34> mask user@100000.service so it will lipstick will not restart and you can strace/gdb test_hwcomposer and so on23:04
r0kk3rzdoes it use hwc2?23:05
T4<elros34> krnlyng: upstream libhybris cause lipstick fail and device reboots for cm11 base,
T4<elros34> do you have any ideas?23:07
osum4est_r0kk3rz, not sure how to tell23:08
T4<elros34> mal, krnlyng: HYBRIS_LINKER_DEBUG test_hwcomposer
malosum4est_: probably has nothing to do with your issue23:17
osum4est_here's the hwcomposer strace doesn't seem to get the trace_marker thing23:32
T4<elros34> even getprop segfault. Anyway here are the HYBRIS_TRACE and HYBRIS_LINKER_DEBUG:

Generated by 2.17.1 by Marius Gedminas - find it at!