Saturday, 2019-10-26

malI was reviewing and testing some stuff00:00
Thaodanchanges in middlewhere like those that were merged to droidmedia?00:01
malthat also, and some of those droid-hal-device changes also00:02
malthose will make local builds a bit easier00:03
Thaodanwhich ones do you mean?00:03
Thaodanah you mean the one from sledges ?00:04
malyes, in hybris-boot and droid-hal-device00:06
Thaodanah great makes sense.00:09
ThaodanI've updated the android pr with the last errors that I've got. I'm still guessing which has changes brought the error00:43
rinigusI am trying to setup rndis connection during a boot (device sony/xz2, aosp9 based). right now its all limited to hybris-boot. as stock android has rndis support which works, I am trying to reproduce its configfs configuration.11:32
rinigushere is a summary of changed init script, dmesg in hybris with failure to start rndis, and section of dmesg that's missing in hybris, but available in aosp9:
malcheck channel logs for RNDIS11:33
rinigusmal: where are those?11:34
rinigus(mal sorry, figured out)11:34
rinigusmal: I did look into channel logs, but it doesn't seem to work for me. I probably haven't tested all possible forms over here, but quite a few.11:35
rinigus... so, I was thinking that as it works on aosp9, it should be possible to reproduce its config11:36
rinigusmal: essentially, I followed your advice as in (found the way its initialized in aosp9, corresponding function name for configfs, but still no success)11:45
malrinigus: at which point of boot are you now?11:58
rinigusmal: same as before - failing on the initial one. I am now having aosp9 installed and working on setting up rndis using aosp9/userdata for dumping the logs. idea is to setup network during boot, so I can properly see what's missing when I boot sfos12:00
rinigusin this respect - at
maldid you do the kernel usb changes mentioned in logs?12:02
rinigusmal: those I did before, now I started to retest them again.12:02
rinigusjust added12:03
malthat has a bit different ones12:05
rinigusmal: I'll try those. but then its not really clear what's going to be used for starting rndis via configfs12:08
rinigusmal: test with options that I mentioned above didn't work either12:09
rinigus... fails with the same "id0:gsi_bind: ipa ready timeout \n configfs-gadget a600000.dwc3: failed to start g1: -110"12:09
T42<elros34> looks like ipastart_sh triggers loading ipa firmware which will be not available in hybris-boot so gsi will not work12:14
malrinigus: that is needed only for early development, after you have sfos running then you can revert that and configure usb-moded correctly12:16
riniguselros34: indeed, looks like that way.12:16
rinigusmal: its true. maybe then I will see if I can setup logging of init troubles somewhere else, like sdcard.12:17
malrinigus: so look in logs for RNDIS and surrounding lines and see if you find something12:17
rinigusmal: if I use recommended settings, I get
rinigusat the same time, mkdir: can't create directory '/config/usb_gadget/g1/functions/rndis.usb0':12:19
rinigusrecommended are the ones which I mentioned above (CONFIG_USB_CONFIGFS_QCRNDIS=y CONFIG_USB_CONFIGFS_RMNET_BAM=y CONFIG_IPA=y)12:19
rinigushaven't tried with your ones12:19
malso try with the ones I gave, those must have been in logs for a reason12:20
mallike my testing on some devices12:20
rinigusmal: that fails with compilation of kernel - .12:23
malfix it12:25
rinigusmal: its actually fails all over the place with rndis.o -
riniguslooks like some kernel option is missing ...13:00
T42<elros34> have you disabled GSI in defconfig?13:01
T42<elros34> at least these USB*_GSI13:03
riniguselros34: no, will ltry13:03
riniguselros34: #CONFIG_USB_CONFIGFS_F_GSI=y helped to compile the kernel. let's see if that helps with rndis13:07
rinigusmal & elrso34: and I have rndis!!!13:09
rinigus... and I can even login. thank you!13:10
malI will try to add the new hadk faq online soon so we can add such things there13:12
T42<elros34> mal: can't we use
mal@elros34 we decided to add a faq repo to mer-hybris github which can have nice formatting also, in addition to that more people have github account than mer account13:14
T42<elros34> ah great13:15
T42<DylanVanAssche> mal: Can I have access to native-common on OBS? I want to trigger a build for `geoclue-provider-gpsd3`..13:15
rinigusmal: great idea regarding github13:15
mal@DylanVanAssche done13:16
T42<DylanVanAssche> mal: Thanks! And good idea for the FAQ repo, it will be more visible too :D13:16
malI still have some parts of faq to cleanup before pushing it there13:16
malalso git log is nice for seeing what has been changed13:16
T42<DylanVanAssche> Indeed!13:17
malalso we can even review changes nicely13:17
ThaodanIs there a good why to get where a file was recursively included in c++ headers?13:45
ThaodanI think camera headers now include stagefright13:45
T42SauravDharwadkar was added by: SauravDharwadkar14:42
ThaodanI found out ICameraService.h pulls in stagefright14:48
malrinigus: @elros34 @DylanVanAssche comments welcome
deathmistwould the new dhd submodule updates allow me to define pre-downgraded packages such as ofono-ril-binder-plugin & nfcd-binder-plugin in my manifest files (e.g. ".repo/local_manifests/middleware.xml") and when are they built anyway? I'm assuming at the end of build_packages15:43
vknecht👍 looks beautiful and well organized :-)15:43
deathmistcurrently I have a rather crappy solution and modify the entire script with the ones I need :p15:44
maldeathmist: the new dhd submodule doesn't pull repo updates automatically unless you add --pull (or -p) switch to, so if you do git reset to older version in some package and don't use --pull then you can just use and no need to skip anything15:45
malvknecht: there may be parts that could do some more formatting but that was just a quick formatting I did15:45
deathmistmal: looks very neat, but I saw my sources linked there for disabled_services.rc xD I need to clean that up as I import some vendor init.rc script after disabling qti & time_daemon (thinking of utilizing the brand new service override-disable method introduced in Pie instead)15:47
maldeathmist: oh, ok15:47
rinigusmal: looking good!15:48
rinigus^ hadk-faq15:48
rinigusmal, hadk-faq: does hybris-16 require source code patches as well? that seem to be missing in 3.315:49
vknechtiirc one of the latest addition to faq not in backups was a spreadtrum/mediatek entry, any way to recover that ?15:50
* vknecht dug all his caches for some trace of faq, but failed15:50
malrinigus: 16.0 part needs fixing, it's also missing ldconfig fix15:53
malvknecht: too bad there was no newer version available15:53
malpull requests welcome15:54
rinigusmal: now we just have to point url at room topic to it15:57
malrinigus: not sure who has permissions for that16:02
rinigusmal: probably room admins, but don't know how it works for irc16:14
malon IRC ops can do that16:14
malvknecht: I tried your cam snapshot PR, it doesn't work, I get "FrameGrabber::emitVideoFrameProbed:57 - Unable to save" and then a path that looks ok16:55
T42<DylanVanAssche> mal: Looks nice! Much more structured than the previous version!16:56
vknechtoh ?17:01
vknechtwill check17:01
malvknecht: also, if I have video player page open in cam app and then swipe the app to background and bring back to front video cannot be played anymore until I close the gallery part of the app and go back from camera view to that17:03
malthat happens at least if I had already played the video17:04
vknechtthis probably has to do with
vknechtcamera.start()/cam running breaks mediaplayer, for some reason17:06
malseems to fix the issue17:08
malvknecht: in the future could you capitalize the first letter of the commit message (after the tag obviously), I like good looking commit messages, in that PR you could add the clarification another line (in git message the main commit message is in first line, then after an empty line you can add more lines if needed for additional info)17:10
vknechtno problemo17:12
malvknecht: in that message you can see that the first line is a bit long and is clipped in the github view17:13
vknechtyes, d*mn github doesn't respect 79 char column convention :-P17:15
malvknecht: some sites suggest using 50/72 rule for commit message, first line max 50 chars and others 7217:19
vknechtok, working on one at 48 ([videoplayer] Fix playing after regaining focus)17:21
mallooks ok, I think it should not be a strict rule but as a guideline, the parenthesis you had we not needed on first line anyway17:23
vknechtthanks :-)17:28
T42<DylanVanAssche> mal: Indeed, compass apps don't work... Needs a virtual sensor that combines the 3 hardware sensors17:28
mal@DylanVanAssche I have partial code for that but it wasn't finished17:32
T42<DylanVanAssche> mal: Neochapay told me that he's handeling the modem stuff. So I can focus on other things.17:33
T42<DylanVanAssche> The V4L camera stuff or the compass. I guess the V4L stuff is more important. However, if you have the code, feel free to push them somewhere :)17:33
mal@DylanVanAssche v4l is more important for you now17:37
piggzmal: resolution switch can be merged?17:48
malpiggz: were you supposed to fix the enums?17:49
piggzmal: i did, u just didnt notice it was 30 seconds ago ;)17:49
malpiggz: heh, I checked a few minutes ago :)17:49
malpiggz: looks ok, please squash the fix commits away17:53
malbefore merging17:53
piggzi intended to17:53
piggzvknecht: SD card is still WIP?17:55
piggzwhat about video snapshots?17:56
vknechtcurrently adding a model, and moving to a dockedview, hopefuly will help with refreshing the list17:56
malpiggz: video snapshots didn't work for me anymore after the latest changes17:56
vknechtfor snapshot, just checked out the branch to investigate17:57
vknechtnot sure I'll do much tonight, if that's your question ;-)17:57
piggznote, when you commit in github, the summary line becomes the first line of the commit message, so no need to duplicate it it. I just force pushed to cleanup the last message17:57
piggzvknecht: no, im just wondering if i should do an interim release, as the current fixes are worthwhile17:58
piggzi think i will17:58
vknechtyes, sensed you'd want to release now :-)17:58
piggzi wonder if i h ca time to fix the 'dont sleep while recording'17:59
malpiggz: that is easy now with Nemo keepalive17:59
malpiggz: I can do it now18:00
piggzso can i, but feel free as im also cooking ;)18:01
malpiggz: do we need it also to video player?18:04
piggzprobably yes18:04
deathmistdoes "- audiosystem-passthrough" need to be added to patterns now after pulseaudio-modules-droid-hidl got split into 2? device is on hybris-16.0 and a QCOM one as well18:10
deathmistI assume I'll have to add it to dcd patterns as it won't get "packaged in" the hidl module or is it only required during build-time? I'm not very experienced in RPM packaging :/18:14
piggzits in the common repo, so it should get pulled in automatically18:15
maldeathmist: no need to edit patterns18:16
T42<DylanVanAssche> mal: `===` :)18:16
maloops, copy paste mistake18:17
T42<DylanVanAssche> mal: Awesome!18:17
T42<Akatsu %lastname%> if i have init file and interpretation will fail on one line will init stop interpreting it in place of crash or not?18:18
malpiggz: that new import needs some rather new sfos version, not sure if it was 3.1.0 or something like that18:20
rinigusas I have now access to logs, I'd like to give it a go and continue with xperia10/aosp9 based approach applied to xperia xz2. after fixing find-mmc-bypartlabel (xz2 has sdaXX disks not mmc...), I have the first boot with journal log.18:20
riniguslog at
rinigusit doesnt boot to gui. initialized led18:21
riniguswhen pressing power button after some time, shutdown was done (red led as well)18:21
rinigusas there are many errors in the logs, what is the recommended one to start with?18:22
rinigusbtw, should I enable back kernel settings that got disabled for rndis (now its CONFIG_USB_CONFIGFS_F_GSI=n) ?18:23
piggzmal: well, rpm validatr like it, so its fine18:23
piggzwe need 3.1 anyway for the modes18:23
malpiggz: just add the correct version on store if there was some way to prevent using on older versions18:25
piggzmal: i dont think 3.1 is an option in the store, it seems to be lagging behind. i already have latest available selected18:27
malrinigus: remove this file
malthat prevents lipstick from starting18:28
rinigusmal: thank you!18:28
rinigus... will test18:28
malpiggz: ok, need to tell about that internally18:29
piggzopenrepos done....18:29
malpiggz: I was just about to rebase my cleanup18:30
vknechtproductive saturday evening :-)18:31
piggzmal: as soon as any cleanups are done, i will update :)18:38
vknechtstrange, snapshots work here...18:42
piggzoh crap, did we break something18:45
piggzyes :(18:46
piggzi need to pull that18:46
piggzits getting stuck in a loop18:46
piggzwhen selecting manual focus mode18:48
rinigusmal: removed - now lipstick killed, status=6/ABRT18:54
vknechtmal, you mean it's with this cleanup integrated that snapshots break ?18:59
malvknecht: no, completely unrelated, that was something I did earlier19:00
malvknecht: I tested your clean branch, need to test on some other device19:00
rinigusmal: should I drop as well?19:01
T42<DylanVanAssche> Oh search is broken on OBS :(19:01
malnot sure, there might be some things in repos that require newer middleware19:01
rinigusmal: I guess I should go once more and delete as much as possible from droid-config and later build it up again, as needed. (like I can drop /opt mounting, for example). hopefully, will not remove too much :)19:13
piggzmal: vknecht: 0.6.6 on openrepos with small fix, feel free to post review!19:27
piggzmal: big question .... you removed all eol ; .... i thought they wre preferred in JS?  i know they are implied, but i thought it was better style to have them19:43
malI wad pretty sure I read somewhere that it's not but need to recheck now19:45
malpiggz: no idea where I read it but it seems I was wrong, fixing19:48
piggzmal: even that page is inconsistent!19:49
piggzthe mousaarea onCLicked at the bottom uses ;19:50
piggzother places dont19:50
malpiggz: it's the qtcreator auto formatter that removes those19:51
piggzmal: ok, but be aware of the the cases in the first link i sent where it can be ambiguous19:53
piggzso, we need to be sure u didnt break anything ;)19:54
mal@adampigg ok, not sure why it was selected like that20:15
mal@adampigg what is your opinion, the sfos qml style guide tells it's ok to have onSomething: foo = bar instead of having curly brackets, should I also change those?20:18
T42<adampigg> I kinda like having braces everywhere for explicitness20:22
mal@adampigg I think the cleanup PR should be tested well, although it's just cosmetic changes mostly there is still some small chance of issues20:29
deathmistmal: HOLY CRAP hybris-16.0 booted on my phone to UI!!!22:14
maldeathmist: nice22:21
malwhat was the fix?22:21
deathmistmal: I honestly have no clue sadly, I updated the init files (, used a kernel with linux-stable merged instead (ver 4.4.197,, possible cause for currently no Wi-Fi, audio, LTE network etc) and just hard copied patched ld.config.28.txt to /system22:24
deathmistafter a reboot lipstick and jolla setup came up :D22:26
malok, maybe the ld.config fix wasn't used before?22:27
malor something was wrong in .rc files22:27
malare you sure the qti service should be disabled? that seems like something that should be running22:28
deathmistI really don't know, I got to UI now on 16.0 with it disabled too and everything functioned fine on 15.1 before with it disabled, oh I also removed all local sources & resynced everything and started fresh with 16.0 patches etc22:29
deathmist :p22:30
malok, wondering what the qti service then does22:35
T42<BusterBg_18> deathmist: nice c:23:21
deathmistmal: it seems to be a direct match with "vendor.dataqti" while "qti" used to be it's old name for me on 15.1, still behaves the same: just keeps on failing and restarting every 10s23:41
deathmistit's late now tho, think I'll look into the proper .mount unit for ld.config.28.txt from logs and do same for the vendor init file I now don't touch / import in libexec init scripts to fix-up cpusets which just need a simple path change23:44

Generated by 2.17.1 by Marius Gedminas - find it at!