Sunday, 2019-11-17

T42<birdzhang> Can you use gitlab ci build  Error <creator>[11/16 06:02:36] : Command 'modprobe' is not available.02:49
T42<birdzhang> ^^
T42<birdzhang> local build is okay02:49
mal@adampigg I removed some useless packages, which have same versions in 3.2.0 already, from the main dontbeevil OBS project06:51
*** birdzhang1 is now known as birdzhang08:01
deathmistmorning, should I switch to memless haptics on 16.0? droid vibrator worked fine on 15.1, but after rebase it only works properly on some boots, and on others it doesn't work at all in pulley menus or home screen (but e.g. keyboard taps still vibrate), very strange behavior08:21
r0kk3rzuse native plugin08:22
deathmistr0kk3rz: I think I'm actually using it, probably got the names mixed up
r0kk3rzyeah i dont think we really use the hybris one anymore08:24
r0kk3rzsounds like you need a cuatom config though08:24
rinigusmorning! as obs is setup and can be used to make working images, I am ready to discuss naming and organization for repos for xz2 and tama platform08:28
riniguscurrently, hadk conventions and same naming as for xperia 10 are used. as a result, kickstart files are generated correctly.08:28
rinigusin addition to h8216 project, there is a sub-project containing system and vendor. that is separate allowing to have repos according to android base08:28
rinigusas a recap, tama platform is for xz2, xz2c, xz3. I presume all of them have 1 and 2 sim variants. total - 6 device codes08:28
rinigusfrom looking into packages in h8216, there aren't many which would not depend on libhybris/droid-hal. so, I am not sure I can organize it somehow differently without using separate projects for each of the devices08:28
riniguswe can have tama project and move there boot image helpers and have tama:extras with system and vendor images.08:28
rinigusI can also make main packages in tama and have each device mirror it (somehow?), but build separately.08:29
rinigusmal ^08:29
r0kk3rztypically the ones we build at all dependent on hybris08:34
maldeathmist: looks like your droidmedia is a bit behind, you should rebuild it, at least based on OBS
deathmistmal: ah yeah, they aren't even updated to 16.0 yet since I want it to at least function properly with cellular data etc so it's usable :p  I just do local builds with latest upstream versions myself08:49
malwhat was the problem with cellular data?08:49
maldeathmist: that link to your logs in there has expired08:51
deathmistalright, I'll grab a fresh journal at least after toggling cellular data again in a sec08:52
maldeathmist: also get logcat and logcat -b radio08:54
deathmistthe libexec logcat or one from /system?08:55
deathmistmal: uh, I get absolutely nothing with "locat -b radio" even with /system/bin/logcat :/  here's journal and full libexec logcat
T42<adampigg> Mal, thx...also, take a look at the 2 bt packages on latte, i thibk they could be common candidates09:15
mal@adampigg any reason you haven't done a pull request of that broadcom-bluetooth change to mer-hybris? although we might need to consider having two spec files, that is used on tablet09:20
T42<adampigg> Mal, forgetfulness :)09:20
malso just make a PR with a new spec with name broadcom-bluetooth-bluez5.spec and it should be fine09:21
malthat won't break tablet09:21
malafter that we can add it to common probably09:22
malalthough those are rarely needed on other device probably09:22
malmost new devices use bluebinder09:23
rinigusin context of, what's a difference between haptics and vibra?10:38
malrinigus: I think maybe like vkbd vibra vs notification stuff etc, those come via different middleware11:10
entilhiyall! anyone know if there's work on fairphone3?11:19
entilmy fp2 is severely breaking down, the modem's gone missing again, and I think I'm done with that hardware and it's time to consider something else11:20
entilethically I'd love sticking to fairphone but if it doesn't exist it can't be helped11:20
rinigusmal: thanks. Looks like the both are working for me11:23
malrinigus: usually both of those are working if one is11:24
maljust for reference those are ngfd-plugin-native-vibrator and qt5-feedback-haptics-native-vibrator11:25
rinigusmal: thanks, good to know.11:28
rinigusAny thoughts regarding sony tama repos?11:29
malrinigus: well there are options how to handle those, how similar are those? those could either be all in tama repo as those still use same android base and probably would be so similar libhybris and things would work on all11:34
malor have a common repo for some stuff11:35
rinigusmal: I don't know how similar are android bases. I expect they are, but there are still separate options you choose for `launch`11:35
rinigussurely some common stuff can be put together.11:36
rinigusalso probably the same versions of packages will be used for all. so, maybe I should branch packages at obs11:36
malrinigus: of course you need separate droid-* packages for those but those packages can all be in same repo if needed11:37
rinigusbut apart from that, it looks like separate builds still. I wonder how xperia 10 or xa2 is organized in jolla's internal obs. I presume this is something similar11:37
rinigusmal: droid-hal indeed. not droidmedia, at least without changing the names11:38
malrinigus: droidmedia only cares about the API stuff11:39
malwhich is most likely the same as those use same hybris branch11:40
deathmistmal: is libhybris/porters in merproject wiki still even used? I could add OP5/5T there as I finally also added them to the adaptations index today too :p11:40
maldeathmist: not all people add things there, you can add if you want, it's always nice to have info there11:40
rinigusmal: from development speed pov, it still makes sense to have droid-hal- packages separate. as it will trigger rebuilds only for the ones that are dependent. if I change only one package11:41
rinigusre droidmedia: good to know, then I can move that to tama11:42
rinigusI presume that audioflingerglue is the same11:42
rinigusmal: so, in my understanding: all that is independent from droid-hal- can go to unified repo - :sony:tama, this includes droidmedia and audioflingerglue and packages depending on them, if any11:45
rinigusthe rest will go into device specific ones. android system and vendor would be under :sony:tama:extra-images in testing11:45
malmaybe it's easier to have separate project for each set of devices which can use the same droid-hal packages11:47
rinigusmal: what would be set then? All 6?11:48
malrinigus: well actually you could have all in same project and just use one (via project config) for building libhybris and stuff11:49
malbecause the base is so similar that way libhybris would work for all even if you just build it against one11:50
rinigusmal: let me think about it...11:52
malthat's how it's done for nile and ganges11:57
deathmistmal: so I was looking thru my journal for HWC version to fill in, I assume it's 2.1? also about RIL, version should be latest mentioned on the comment in ril.h (15 in my case) here or should I find it on the running device somehow?12:05
maldeathmist: that is usually available in logcat -b radio which for some reason didn't work for you12:11
malI mean RIL version12:11
malhwc version is seen in journal log12:12
maldeathmist: hmm, I need to remember how to parse that hwc version12:12
malprobably 2.1 is fine12:13
deathmistweirdly enough in my reference libexec logcat on 15.1 there also is 0 mention of "radio" in the entire file12:14
maldeathmist: normal logcat (without parameters) doesn't have radio related things "/usr/libexec/droid-hybris/system/bin/logcat -b radio" should have12:16
rinigusmal: sounds like a great plan! assumption is that libhybris will work and all differences are in config/version packages and system/vendor images.12:29
rinigusmal: as for libhybris preference (which hal to use), I guess it will go into config as in
rinigusmal: question regarding this organization - how do you tell obs to build all included SPEC files?12:30
malyes, just add Prefer: droid-hal-something-devel12:30
rinigusas if I have multiple specs in
malrinigus: you add separate packages for each12:31
malI think it doesn't know how to handle multiple specs otherwise except to have package names after each spec12:32
rinigusmal: OK, but how do I tell which SPEC it will use in each package? as when should it choose droid-config-h8216.spec and when something else from the same source tree.12:32
malrinigus: it chooses spec based on package name12:33
rinigusmal: nice! will try.12:33
malrinigus: so if you add a package with name droid-config-h8216 it swill use a spec with that name, in case there is no such spec then it search for any spec, if there are more spec files then it might fail, you can test multiple spec build by renaming some package with multiple specs to something that doesn't match any of those12:34
rinigusmal: as it works for ganges and nile, should work here as well12:35
malafaik it should only know how to build one spec for each package so just add the separate packages12:35
malwhich all then use the same git repo12:35
rinigusmal: now what about generated kickstart file in . I suspect that url will be generated with the device name (as h8216) and not tama in it.12:35
malrinigus: check how it's done for ganges, it has stuff like board mappings12:37
mali.e. variants12:38
malrinigus: which means we probably should rename the OBS project12:40
rinigusmal: yes, I am sure we will need to rename, probably to tama. from those mappings I cannot figure out what will be url in kickstart12:41
malrinigus: probably related to adaptation-repos in 05-sony-ganges.ini which is then used by ssu things12:42
rinigusmal: I guess I will have to experiment and see what url will be composed12:43
rinigusmal: or do you have somewhere nearby ganges ks?12:43
malthere is hopefully it will work, if not then we need to fix the ssu things12:44
rinigus.. but I can also experiment with it12:44
maljust check how the kickstarts look if you add such configs12:44
rinigusmal: ok, will do. So, I will check how kickstart will change and then, if all is fine, we will rename project accordingly12:45
rinigusmal: unfortunately, it looks to be ignored. commit
rinigusks -
rinigusmal: missed that, sorry!13:15
rinigusmal: still seems to be ignored for ssu ks generation13:30
malrinigus: there is the new repo in there though13:31
malbut not completely correct though13:31
malrinigus: maybe community adaptation messes that up13:32
malin your spec13:32
* vknecht sharpens his Loire Pie builds13:34
rinigusmal: not sure what you mean regarding incorrect url, looks to be ok - also when tested via browser13:40
rinigusas for community adaptation, maybe. no idea13:40
riniguswill test without it13:41
vknechtiiuc setting community_adaptation = 1 sets some sort of default when there's no ssu board mappings, so it's sort of exclusive13:42
vknechtMal, curious to know what POV snapshot PR gave with another device, had you have a chance to check and did it work better ?13:49
rinigusmal, vknecht : that worked. not what we want, but at least some change -
rinigusI guess I should base it on
malrinigus: I meant that your config has sony-tama:/extra-images but the url .ks has h8216:/extra-images14:11
malrinigus: I think we might need to do some changes somewhere, not sure if there is even point having the extra-images project14:12
malrinigus: we need community-adaptations anyway it seems14:13
rinigusmal: I agree. except extra images project is good to have to avoid many copies of the same images for multiple sfos releases. right now its one image per android base14:14
rinigusmal: one way would be to fork community-adaptations and make changes in its urls14:14
malrinigus: that will need some thinking because for example how would that work in testing, the url would have to have some custom target name14:17
malvknecht: I need to test that again on some devices14:19
rinigusmal: devel/testing is done via spec name trick in community-adaptations and declaring project either -devel or -testing in name. for extra images, we will have the same url for devel and testing. no need to keep 2 copies14:19
malrinigus: I mean that community-adaptation forces the target name to be versioned in testing14:19
rinigusmal: if we talk about extra-images, this doesn't matter, as far as I can see. we just include it as 'external' repo. or do I miss something?14:21
malrinigus: where did the url get that target name?14:22
rinigusmal: I need more specific question, sorry. which url and what's target in it?14:24
malrepo --name=adaptation-community-h8216-extra-images-@RELEASE@ --baseurl=
malwhere did that aosp_9.0.0._r46 come to that, where did you define it14:24
malthat is the name of the OBS target14:25
malyou must have defined it in some config14:25
rinigusmal: sure. its defined in
rinigusas for url, I define it using a hack in
deathmistalright just tested calling and seems to work nicely on hybris-16.0 :)  just no cellular data for some reason, SMS already works too15:08
deathmistwhat I did notice tho is that my BT headset still doesn't function during calls (ringtone from phone speaker & different tone from headphones, no mic unless switched to phone speakers, no call audio from headphones) when media audio works normally. done the following already am I missing something?15:12
maldeathmist: what BT configs do you have enabled in kernel?15:25
malin defconfig15:25
deathmistmal: at least these ones I've moved down with the other hybris changes
malok, those should be enough15:30
maldeathmist: remove droid-config-cheeseburger-bluez515:33
maldeathmist: sorry, don't remove15:33
malmisread some configs15:33
maldeathmist: do you see anything in journalctl when you try to use BT headset during calls?15:36
deathmistmal: there is "Refusing unexpected connect" from bluetoothd and "audio_route_sink(): unknown audio sink device = 'bthfpforcall'" from mce at least, I first placed a call normally and then received one after already connecting the BT headset
maldeathmist: check what "pactl list" tells as nemo user16:07
T42<Lukapanio> @Lukapanio [], here is logcat16:07
mal@Lukapanio which android base?16:09
T42<Lukapanio> @mal 1516:13
T42<Lukapanio> journal16:13
deathmist@Lukapanio you've clearly missed entirely on reading second part of I don't see overrides for hwservicemanager and vndservicemanager being used...16:24
maldeathmist: that was my guess also and why I asked which base it was :)16:29
*** electro is now known as Guest7671716:29
T42<nitin_3> @mal I am getting access call barring error from one service provider16:40
T42<nitin_3> when on sfos ,someone else also tried with same service provider and their out going calls are also getting barred16:40
T42<Lukapanio> @deathmist thanks done tha on device16:40
T42<Lukapanio> got boot animation of los16:40
deathmist@Lukapanio isn't that worse? SFOS has no bootanim or did you run surfaceflinger manually? I don't understand16:41
maldeathmist: your ril config looks odd, you have both socket and transport in there16:45
maldeathmist: compare to this
deathmistmal: oh yeah I was meant to clean that file up a bit after rebase, those are indeed useless but also don't help with cellular data in anyway (just tested after removing them and rebooting, same errors still)16:49
deathmistand by "those" I meant socket lines16:50
maldeathmist: how did you select the other custom changes you have there?16:50
T42<Lukapanio> I HwcPassthrough: HwcHal::createClient: Client already exists. Waiting for it to be destroyed.16:50
T42<Lukapanio> hmmmmm how to know what client, and how to destroy it16:51
deathmistmal: I looked thru the reference ril_subscription config upstream and picked options after I made syre they didn't cause any regressions back on my 15.1 port, just carried the file over to 16.0 as well. tried Xperia 10 and XZ2 (which worked fine for a start in 15.1) configs too, but still no cellular data with all same errors16:56
maldeathmist: did you have the same android side services disabled in 15.1 or did you disable something more in 16.0 (other than the standard ones like in ganges)17:06
malthe changes you had in config repo looked quite scary17:06
deathmistlmao, yes I did disable more stuff, maybe that's worth yet another look, I'll start re-enabling stuff until SFOS won't boot or stuff starts working and take notes :p17:14
deathmistmal: btw the "patched" vendor files I have there mainly just changed paths to /dev/cpuset so that worked, but after further research there is already another cpuset cgroup mounted: /sys/fs/cgroup/cpuset so maybe I'll do udev magic and make a symlink to /dev/cpuset etc for droid stuff compatability. I've already disabled mount units for those too to17:17
deathmisteliminate them causing issues17:17
maldeathmist: those vendor things in libexec look very odd17:21
deathmistthey're mostly useless, already disabled their respective bind mount units so they can't affect anything17:22
maldeathmist: you mean those vendor.dataqti and rmt_storage is very imprtant (it's for mobile data or modem in general)17:23
maland so on17:23
malI'm quite sure those are required17:24
malalso android 9 has override option for services in which case there no need to edit the original files anyway, the whole thing is just very wrong17:25
malmany of those look like something that shouldn't be removed17:25
deathmistmal: my libexec vendor init files? the things I removed just spam logs and die continuously or (most of them) the actual binaries the services reference don't exist. either way I removed them from libexec now on the device at least as well as their mount units, re-enabled qti stuff in init files, no cellular data stil after rebootl. I'll keep going17:29
malwithout actually seeing logs I can't really be sure17:32
malif a service dies there is usually some reason for that, it's not common for that to happen17:33
deathmistmal: when I use instead, still no luck. constantly failing services with that file include time_daemon (safe to disable), vendor.dataqti (looking thru init files it's some logging service, seems to be equivalent to my previous 'qti' service on 15.1 which I didn't need for cellular data) and keystore17:38
deathmistI'll send journal in a bit17:40
deathmistmal: journal after ~2 min of waiting also "vendor.dataqti" is marked as "disabled" in my /vendor/etc/init/hw/init.qcom.rc and it's binary is indeed just "qti"17:51
T42<Lukapanio> @Lukapanio [ …], hmmm @birdzhang  how have you fixed?17:52
T42<Lukapanio> that 2 commits dosnt help17:52
deathmist@Lukapanio which issue? I can try to help, but telegram replies cut off so the URL is broken in IRC17:56
rinigusis there any reason to not include jolla-settings-accounts-extensions-3rd-party-all? is it included on official devices by default?17:57
deathmist@Lukapanio I assume you're in an early porting stage, common cause is /system & /vendor not being mounted, so maybe give output of "mount"18:00
deathmistrinigus: idk what you mean, pretty sure that's been default in patterns for a while even on community devices
rinigusdeathmist: then I managed to miss it somehow in my config.18:02
T42<Lukapanio> @deathmist18:02
deathmist@Lukapanio I already noticed, but seems both are mounted. what is the device?18:03
T42<Lukapanio> p18:03
T42<Lukapanio> moto g518:03
T42<Lukapanio> cedric18:03
T42<Lukapanio> test_lights18:04
T42<Lukapanio> is ok18:04
T42<Lukapanio> test_sensors is ok18:04
deathmist@Lukapanio after adding init files from FAQ, did you rebuild droid-configs and make new image or somehow place them on the device rootfs? if the ladder, how? I had issues manually placing files to where they should be from TWRP (it was init files specifically too), so maybe make a new image with the changes and see if anything is different18:08
T42<Lukapanio> changed inrootfs18:09
deathmistthen I suggest you doing the changes to your droid-config repo, run build_packages -c and build_packages -i after18:11
deathmist@Lukapanio you've made your droid-hal spec incorrectly, you never terminate straggler_files with a "%{nil}", also doesn't SD430 have 64-bit support?18:16
T42<Lukapanio> it have18:16
deathmistwell why don't you have "%define droid_target_aarch64 1" in your spec file then too?18:17
T42<Lukapanio> yep18:17
deathmistso you don't keep your github sources in sync with your local ones? I can't really help you much if I don't know what your repos actually look like when you're asking a question18:18
T42<Lukapanio> ok will upstream to new branch18:22
T42<Lukapanio> %define droid_target_aarch64 118:22
T42<Lukapanio> i have it18:22
T42彭汝掎 %lastname% was added by: 彭汝掎 %lastname%18:36
deathmistmal: what do you think about ? :)  I'll prob move over some steps from my GitHub guides later, but would it be ok to merge the two 5/5T models together? the hardware is pretty much identical and only difference is droid-* packages which I can deal19:47
deathmistwith myself19:47
vknechtdeathmist, nice pic :-)19:51
vknechtproblem with wlan hotspot/connection sharing ?19:56
deathmistthanks :) yeah well cellular data is broken so I can't share hotspot, it probably works fine but I can't test it :p19:57
vknechtwasn't fan of upping base version, but I guess no much other choice now... hope it'll play well20:09

Generated by 2.17.1 by Marius Gedminas - find it at!