Friday, 2024-02-09

genericv3i'll, you are elros34 on github right?05:02
genericv3Also mal, can i get your github username too05:02
genericv3Also i need to take my phone to service tomorrow. I need to delay everything if we can't fix it today, but i'll have more time than usual today.05:03
T42_<elros34> I am, why do you ask?11:03
T42_<b100dian> Probably related to the upload of droid config on github, but I think there's a misunderstanding, you don't need the usernames to give access to a project, unless it is private, and I wouldn't recommend making a porting effort private.12:27
genericv3I'll make repos private for development stage, not gonna make these efforts entirely private as b100dian said :)12:49
genericv3Elros i sent you a organization invitation, you can leave after everything is done if you want.12:59
T42_<Herrie1982> mal: On the sensorfwd issue: seems we had some race condition with sensorfwd and sam (webOS component) that were both taking 90s (default timeout to kill). Adding "TimeOutStopSec=5" brings the reboot down to an acceptable 10s. Maybe not the most proper solution but does the trick13:31
mal@Herrie1982 any chance you could attach gdb to the sensorfw process when it appears in hanged state and check from backtraces where it's waiting?13:52
maldoes the same happen if you just try to restart the sensorfw service?13:53
T42_<Herrie1982> mal: Yes that also takes forever13:53
T42_<Herrie1982> I'm pretty sure it's one of our components that doesn't do stuff gracefully and not sensorfw13:53
T42_<Herrie1982> So it locks sensorfw13:54
malwe haven't noticed such issues at least, which backend does the device use? hal or binder?13:55
T42_<Herrie1982> Well it's PT2 on mainline in this case13:55
T42_<Herrie1982> But have seen it on Halium with binder too13:55
T42_<Herrie1982> So I suspect it could be our ALS or other component using sensorfw13:56
malso iio?13:56
malthat reminds me, I should finish my iio fixes for sensorfw13:57
T42_<Herrie1982> Yes iio14:07
genericv2i am home finally, starting to work on sparse15:56
genericv2mal: what about your gh username, are going to give it or no?15:57
malit's quite public knowledge and visible in logs16:12
malI still don't know why you have to make the project private16:14
genericv2idk i want to make it private before we could make it boot to gui16:18
genericv2uploaded everything16:31
genericv2am i going fine?17:38
malsome odd things in there17:39
malmissing sparse/etc/selinux/minimum/contexts/files/file_contexts17:41
genericv2isn't it this one17:42
malthe system mount stuff should not really be there, you should fix fstab instead and rebuild hybris-hal and then droid-hal rpms (at the same adding some other small fixes)17:43
malsparse/etc/selinux/minimum/contexts/files/policy.30 should not exists, that file is in sparse/etc/selinux/minimum/policy/policy.3017:43
malalso wondering about all of those files in sparse/usr/libexec/droid-hybris/system/etc/init17:44
genericv2they came with the submodule17:44
genericv2i didin't added them17:44
malif they are in submodule why did you copy them?17:45
genericv2i didin't, they were already there. as i remember17:45
malyou must have copied those from somewhere17:46
malI have no idea where those are from17:47
genericv2me too tbh, i am checking right now17:47
malas you can see in submodule things are disabled like this
maland you don't have to touch that17:48
malalso the repo is missing the task_profiles things17:48
genericv2i know i am gonna add it too17:49
genericv2okay i found where i got those from17:49
genericv2hadk-faq hybris-15.1 section xD17:49
genericv2wait i'll clean this mess17:49
malyou should remove installroot, tmp and documentation.list from git and add this kind of git ignore to prevent them getting there in the future
malnothing in 17.1 or 16.0 instructions said you should do 15.1 also17:52
mal17.1 only said to apply patches like in 16.0 nothing more17:52
genericv2i know i know. i made that mistake in very beginning17:52
genericv2right before we are messing with kernel issue17:52
malwhy did you remove the repo now?17:53
malyou know you can just force push to it17:53
genericv2i'll start from beginning in that repo, i did so much mistakes for that17:53
malthat was not necessary17:54
genericv2ah right17:54
genericv2force push is a thing17:54
malnow you will probably miss things again17:54
genericv2yeah xD17:54
genericv2but i am not that stupid now, so it'll be better17:54
genericv2it must be okay now18:09
genericv2can you check ?18:10
genericv2Now about system mount, i must do it like this right?: "/dev/block/bootdevice/by-name/system /system_root"18:13
T42_<elros34> no, you never use by-name paths, that is whole point of fixup-mountpoints18:28
genericv2this is no fixup-mountpoints. wait wait wait18:28
T42_<elros34> its not but whole explanation from fixup-mountpoints is valid here too and for every systemd mount unit18:29
genericv2Now, to fix system mount, i'll need to change system to system_root from fixup mountpoints right?18:30
genericv2That file was fstab.qcom18:31
malyou only fix fstab18:35
malfixup remains as it is18:35
genericv2as i guessed18:36
genericv2is mounting by-name/system to /system_root enough or18:36
genericv2do i need to add another entry18:36
malyou are now completely misunderstanding what to do18:36
malwhat kind of change did you do to fstab?18:36
genericv2nothing right now18:37
malyou only touch fstab, do not touch fixup18:37
genericv2i know, didin't edited fixup-mountpoints18:37
genericv2i am  talking about fstab18:37
malthe whole point is that if fstab has /system then it becomes /18:37
malI mean you replace it like that18:38
genericv2Directly to "/" or "/system_root"?18:38
malsystem_root is not something that belong to anywhere, it's just internal magic done in droid-hal rpm build18:38
genericv2ah okay then18:38
genericv2"/dev/block/bootdevice/by-name/system                            /"18:38
genericv2like this right18:39
malyes, after building hybris-hal and then droid-hal rpms you should see the correct mount services in droid-hal rpm (one of those, can't remember if the main one)18:39
genericv2okaay, saved it like this then.18:39
genericv2I remember like there was a ld.config thingy, i'll add that to sparse aswell18:40
malthere should be no ld-config things in android 13 base18:40
malare you sure you didn't take that from the 15.1 instructions18:41
T42_<elros34> now I understand why you had ld.config mount unit in your previous logs18:42
genericv2I applied everything from 15.1 instructions in my previous bulds. That config is HILARIOUS18:42
genericv2But you know i am not stupid now.18:43
genericv2now, about usb-moded. Can i obsolote it from config or masking it from sparse was right?18:43
genericv2Then /odm and thats it i think, i'll check irc logs again to see if i missed something18:45
T42_<elros34> are you sure you even need to mask it? Its only sometimes needed. Definitely not something you want on working port18:46
genericv2i am sure, it aggressively kills my usb port when it's enabled.18:47
genericv2If my power button was working. that will not be gonna a big deal but..18:47
malusb-moded disabling is just temporary and should not probably be added to git18:48
T42_<elros34> I am curious why do you create all of these from scratch like mount units? You already had them created and even working18:49
genericv2Just purged droid-configs. Didin't touched to fixup-mountspoints etc and anything else18:49
malI really hope you took a backup of the droid-config changes18:50
genericv2mal: are you sure it'll work now, you know we will need to wait for hours if it fails again18:50
genericv2talking about usb-moded18:51
genericv2not gonna mask it as you said18:52
malno idea, but I said not to add it to git, you can have it locally until you verify you can re-enable it18:52
genericv2oh okay18:52
malall I meant was that adding temporary hacks to git is not good, especially if those are only for very early porting18:53
genericv2yeah yeah i understand what you said after writing that18:53
malplease verify the mount services in droid-hal rpms before rebuilding a new image18:55
genericv2lets see if this beautiful build works after previous mess19:08
T42_<b100dian> mal: I have two droidmedia camera2 PRs: 121 with fixes for Zenfone8 and 122 with a hack for initial params. Just 1. so you know and 2. maybe add camera2 tag to them20:40
T42_<b100dian> draft PRs, by all means20:41
mal@b100dian I had also wip code locally to properly handle parameter setting, need to check how different your code is20:42
genericv2i built and flashed now. Mountpoints are okay. Gonna reboot20:43
T42_<b100dian> mal: hope you have a saner solution, mine's just a hack20:43
T42_<b100dian> but thanks for tagging them20:44
mal@b100dian my changes have some bug which I didn't yet find, I will compare and see which makes more sense20:44
T42_<b100dian> maybe half sane / half hack :)20:45
mal@b100dian with those changes you have does it keep the settings when for example going from viewfinder to checking taken pictures and back?20:45
T42_<b100dian> let me check20:46
T42_<b100dian> Yes. That is, PR 12220:46
T42_<b100dian> I'm afraid a clean solution is to change some things around camera usage from the gst-droid plugin20:47
malyeah, gst-droid is not really that great for camera220:48
malI have even considered making a new one20:48
T42_<b100dian> The whole machinery is kind of involved there, and so is the gstreamer code, and seems hard to refactor, but I do agree it is a wonder that it exists and works!20:49
malI remember the was some talk in gstreamer to refactor some camera things to make it more compatible with camera220:50
T42_<b100dian> You know, last summer I was fixing playback on Zenfone8 by watching the interaction of gecko-camera and droidmedia on my device. And I could, and I did!20:51
T42_<b100dian> Next would have been fixing gst-droid + droidmedia interaction20:51
T42_<b100dian> I failed20:51
T42_<b100dian> Interesting, that is, gstreamer mailing list?20:52
malonce those settings are working in camera2 I still need to fix the video recording issue which is probably something simple, the issue is that the output video only shows first frame because some issue with timestamps or something like that20:52
T42_<b100dian> ah, was gonna ask about that!20:53
mal@b100dian can't remember if there, or issue tracker or somewhere else20:53
T42_<b100dian> So video is also crashing gst-droid/gstreamer?20:53
T42_<b100dian> I know you have made progress around video, but dont' know what the state of the current problem20:54
malit doesn't crash in my development code20:55
malit records video, the size of the file looks ok but playback fails except for first frame20:55
malcamera works also in browser with my code20:55
T42_<b100dian> Oh, maybe I have old versions, first let me check20:56
malit's not in that camera2 branch20:56
T42_<b100dian> yes, there are several20:57
malit might be that camera2-test3 branch but need to check that later if it really was that20:57
malit does say almost working video in commit messages :)20:57
malnice commit messages in that bracnh20:57
T42_<b100dian> yes, I think those I have20:58
T42_<b100dian> the commit messages don't matter now:)20:59
T42_<b100dian> I was seeing crashes, that you may not have, let me try find who/what20:59
T42_<b100dian> First frame does seem promising, maybe you need a decoding patch :D but feel free to share a 'broken' video, maybe just maybe there's something ffmpeg can tell us about it21:00
genericv2mal: droid-hal-init is timing out right now21:01
genericv2made timeoutsec 9999, but it's probably in loop21:02
mal@b100dian I remember seeing messages that playback drops frames a lot21:02
malgenericv2: wondering what you are missing21:02
genericv2xD, me too. i wonder what did i do wrong21:03
genericv2ohmd.service fails to start21:03
malthat selinux config in git is still a mess21:03
genericv2no, i (think) i fixed it21:03
malyou have two policy.30 files there now21:03
malin different places21:03
genericv2yeah it's still a mess i guess21:04
genericv2i am getting loss to fix it, bye for a few minutes21:04
genericv2i need to fix my english aswell21:04
T42_<b100dian> mal: but it matters how the final product is on desktop too. Playback dropping frames is .. well, something for ccodec2 devices I made PR 111 if you remember, but it may not apply here of course21:04
mal@b100dian I tested playing the video on my laptop also, failed there also21:05
malgenericv2: how about you look at the folder structure here for example
T42_<b100dian> Ok. It is not like your device was immediately upgraded from ACodec to CCodec2 if you recorded with camera2 API:) So there is a recording problem21:07
genericv2yeah i was searching for it21:07
T42_<b100dian> I am currently seeing `Thread 56 "CodecLooper" received signal SIGSEGV, Segmentation fault` in jolla-camera but no logcat errors when video recording crashes camera221:07
T42_<b100dian> GST_DEBUG=4 jolla-camera Since there's no logcat error per se I don't post it here.21:13
T42_<b100dian> Anyway, no need to look into that. The info that you can record and I can't is already useful21:13
T42_<elros34> what cause all of these false "library "*" not found". Is it android or some libhybris changes fault?21:17
mal@b100dian need to check if that branch even has all my changes21:21
T42_<b100dian> mal: yes pleaae. make camera2-test4 branch:)21:25
genericv2mal: can you check the current selinux config21:29
malseems reasonable21:30
genericv2droid-hal-init.service: Start operation timed out. Terminating21:34
genericv2great great great21:34
genericv2now theere is no hwservicemanager etc. is that okay"21:36
genericv2/usr # ls ./libexec/droid-hybris/system/etc/init/21:36
genericv2apexd.rc              disabled_services.rc  ecclist.rc            hw                    hybris_extras.rc      minimediaservice.rc   minisfservice.rc      servicemanager.rc21:36
genericv3No luck (: I need to take phone to service tomorrow morning. Maybe it'll get back in a week, or two. This will be long for me but nothing to do. At least i'll continue with a new screen and a working power button :) I'll continue checking here as always. Thanks everyone.22:04

Generated by 2.17.1 by Marius Gedminas - find it at!