**** BEGIN LOGGING AT Wed Jan 24 03:00:01 2018 Jan 24 07:01:18 morning: Latest log, somehow my qcom.init.rc doesn't stick :S https://bpaste.net/show/c3fa2514bd6c Jan 24 07:36:52 Morning Jan 24 08:53:19 Tofe: piggz came up with the following: Herrie|Laptop: iirc, sigabrt on android services was becuase of incorrect ld_library_path in init Jan 24 08:53:23 Any ideas where to fix that? Jan 24 08:55:42 Herrie|Laptop: in one of the .rc, iirc. Maybe init.environ.rc ? Jan 24 08:58:50 like here https://github.com/Halium/android_system_core/blob/halium-7.1/rootdir/init.environ.rc.in ... But the ld_library_path is quite specific, and android checks it in a pretty restrictive way Jan 24 09:09:44 Tofe: maybe here? https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-android/recipes-core/android-system/android-system/init.rc#L10 Jan 24 09:09:53 Seems we might need to update the path for aarch64? Jan 24 09:10:35 And then we also have this one still: https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-android/recipes-core/base-files/base-files_3.0.14.bbappend Jan 24 09:11:15 piggz: Thnx for pointer, seems we might need to modify these a bit for aarch64: https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-android/recipes-core/base-files/base-files_3.0.14.bbappend and https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-android/recipes-core/android-system/android-system/init.rc#L10 Jan 24 09:12:39 as for init.rc, I doubt we still use it anymore ? Jan 24 09:12:39 Herrie|Laptop: the base-files file is for executing android code from LuneOS, not from the lxc container Jan 24 09:14:38 Tofe: I'm not sure... Need to check Jan 24 09:22:23 Tofe: Yeah init.rc doesn't seem used Jan 24 09:23:26 Tofe: THere's another one in https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-android/recipes-core/android-system/android-system/lxc-config#L26 Jan 24 09:23:37 Not sure which one gets picked up at which stage :P Jan 24 09:24:31 the lxc one can be used to add more directories, yes (like the 64bit ones) Jan 24 09:25:05 Also I forgot to clean it up, actually; now that we don't use overlays anymore Jan 24 11:19:35 Tofe: Seems init.environ.rc.in comes from https://github.com/Halium/android_system_core/blob/halium-7.1/rootdir/init.environ.rc.in Jan 24 11:20:44 yup, makes sense Jan 24 11:21:01 but I think lxc is the best lead here Jan 24 11:31:09 Tofe: Already tried to set it there, same issue Jan 24 11:31:14 ah... Jan 24 11:31:32 Could still be my fstab, udev as well Jan 24 11:31:52 Really no clue where to start... Just doing random things now to figure stuff out Jan 24 11:31:59 remember udev is not required for starting android container Jan 24 11:32:31 also.. wonder why you have your ow init.rc and stuf ? re: https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-android/recipes-core/android-system/android-system/init.rc#L10 <- Jan 24 11:32:49 shouldn't it just come from halium? Jan 24 11:33:19 bshah: it does, that's just legacy unused code Jan 24 11:33:25 we should clean it up Jan 24 11:33:27 bshah: It's not used ;) Jan 24 11:33:28 Ah okay Jan 24 11:33:40 sorry didn't read backlog fully.. just jumped gun Jan 24 11:33:46 carry on :p Jan 24 11:33:55 As you can see here it's not deployed ;) https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-android/recipes-core/android-system/android-system_1.0.bb#L22 Jan 24 11:37:24 bshah: I'm a bit clueless where things go wrong, could be a simple missing mount, udev rule or something... I'm not 100% sure about all the start order to things in lxc etc. https://bpaste.net/show/2b359fe8386e Jan 24 11:37:28 L1197 and onwards Jan 24 11:39:20 Herrie|Laptop: are you able to ssh/adb shell into? Jan 24 11:39:40 not yet, afaik Jan 24 11:40:20 what do you use? rdnis? or adb shell? Jan 24 11:41:24 ...nothing... adb isn't started yet, or not functional, same goes for wifi/ssh/rndis Jan 24 11:41:53 rndis doesn't require the container to be started Jan 24 11:42:34 but it requires networkmanager, isn't it Jan 24 11:42:36 no Jan 24 11:42:46 oh. So maybe there's a chance there. Jan 24 11:42:53 https://github.com/halium-packaging/hybris-usb-packaging/blob/master/usb-tethering you'd want something like this Jan 24 11:43:08 and systemd job starting it: https://github.com/halium-packaging/hybris-usb-packaging/blob/master/usb-tethering.service Jan 24 11:43:18 AFAIU Herrie|Laptop: device is not bootlooping right? Jan 24 11:44:01 bshah: Shows Xiaomi logo, then goes to black screen with backlight on, then quickly to backlight off, but doesn't bootloop Jan 24 11:44:08 Should be time to get it at least for a minute or so Jan 24 11:44:15 Eventually it might reboot I guess Jan 24 11:44:16 Let me try Jan 24 11:44:45 I think currently your UI is unhappy otherwise dmesg looks pretty okay to me Jan 24 11:44:59 Herrie|Laptop: it probably won't reboot; I think it's just stuck in systemd waiting for android container to finish its startup Jan 24 11:45:02 except modem and sensors bits are unhappy but overall nothing crazy Jan 24 11:48:27 Well the many "no such file or directory" and "no space left on device" are worrying me a bit ;) Jan 24 11:48:49 no I see them on my working nexus 5x as well Jan 24 11:48:53 feel free to ingore them Jan 24 11:49:40 Seems it reboots after some 3-4 minutes Jan 24 11:50:05 we have enough time to debug what's going on I suppose... Jan 24 11:50:21 can you show me what starts rdnis connection in webos? Jan 24 11:50:50 We don't have rdnis currently I think Jan 24 11:50:52 Let me grep around Jan 24 11:50:54 if there is nothing adapt to what halium is doing ^^^^^^^ Jan 24 11:53:52 Seems we had it previously Jan 24 11:53:57 For some Samsung & HP targets Jan 24 11:55:49 https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-samsung/recipes-core/initscripts/initscripts_1.0.bbappend Jan 24 11:56:04 https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-samsung/recipes-core/initscripts/initscripts/tuna/rndissetup.sh Jan 24 11:56:35 And even some systemd bits: https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-samsung/recipes-core/systemd/systemd-machine-units_1.0.bbappend Jan 24 11:56:48 https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-samsung/recipes-core/systemd/systemd-machine-units/rndissetup.service Jan 24 11:56:59 https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-samsung/recipes-core/systemd/systemd-machine-units/tuna/rndissetup.sh Jan 24 11:58:00 Full log until reboot just for reference: https://bpaste.net/show/c604d34d54fc Jan 24 12:00:08 it's wlan which is causing reboo Jan 24 12:00:10 hm Jan 24 12:02:55 That could indicate firmware not being mounted? Jan 24 12:03:27 maybe Jan 24 12:03:35 that's why it's best to have shell access Jan 24 12:03:42 so we don't keep shooting in dark Jan 24 12:27:55 bshah: I guess it should be straight forward enough to get the rndis on yocto Jan 24 12:28:01 Let me check, maybe AsteroidOS already has it Jan 24 12:35:56 Seems they don't Jan 24 18:35:20 Tofe: for rndis should just add systemd machine unit? Jan 24 18:38:18 Any reason why I couldn't start adbd earleir? Jan 24 18:40:38 Or the same issue as with initramfs? Jan 24 19:50:01 Herrie: it could be it's the same issue as with initramfs; or that the lxc init doesn't go as far as initializing adb Jan 24 20:00:30 Tofe seems rndis is started as one of first by systemd Jan 24 20:00:45 We could do the same with adbd? Jan 24 20:01:23 yes, we could, I just thought that it was done in the android container, not by systemd Jan 24 20:04:13 Tofe: Well there's some legacy polution of initscript, chroot etc Jan 24 20:04:38 Which might put me on wrong track Jan 24 20:07:00 I didn't look at it deeply, it could well be we do have a systemd adbd service Jan 24 20:08:30 let me look at how it works on N5... Jan 24 20:10:59 ok yes it's android-tools-adbd.service Jan 24 20:11:55 maybe your /usr/bin/android-gadget-setup is faulty ? We saw that for initramfs there were some things missing... Jan 24 20:13:11 Tofe: Not the startafter in .service? Jan 24 20:13:45 There's no such thing, so I guess it's started pretty early already Jan 24 20:25:09 the one from N5 comes from meta-oe Jan 24 20:26:23 https://github.com/openembedded/meta-openembedded/blob/pyro/meta-oe/recipes-devtools/android-tools/android-tools-conf/android-gadget-setup Jan 24 20:27:23 Looks like it could use some of the tweaks you imported for initramfs Jan 24 20:37:27 Tofe: It wouldn't hurt I guess Jan 24 20:37:36 It's only 2 values anyway Jan 24 20:37:53 AsteroidOS guys do it a bit weird Jan 24 20:38:21 They set all values and then android-gadget-setup overwrites all except for 2 :-P Jan 24 20:38:38 At least that's my interpretation Jan 24 20:40:37 https://github.com/AsteroidOS/meta-asteroid/blob/master/recipes-core/initrdscripts/initramfs-boot-android/init.sh#L47 Jan 24 20:42:22 Then they run https://github.com/openembedded/meta-openembedded/blob/master/meta-oe/recipes-devtools/android-tools/android-tools-conf/android-gadget-setup Jan 24 20:56:01 Tofe: There was a way to get the systemd startup order wasn't there? Might be useful to have from N5 so I can debug mido Jan 24 21:13:20 Tofe: Seems we have this one that's orphaned somehow: https://github.com/webOS-ports/meta-webos-ports/blob/pyro/meta-luneos/recipes-android/android-tools/android-tools/android-tools-adbd.service Jan 24 21:14:13 Seeing we end up having https://github.com/openembedded/meta-openembedded/blob/master/meta-oe/recipes-devtools/android-tools/android-tools/android-tools-adbd.service Jan 24 21:15:25 Seems here we explicitly disable adbd inside lxc? https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-android/recipes-core/android-system/android-system/20-remove-services#L7 Jan 24 21:29:00 Herrie: yes, so that it doesn't conflict with the one outside of lxc Jan 24 21:30:53 Tofe: Could simply be we're missing firmware path somehow as well in this one: https://github.com/shr-distribution/meta-smartphone/blob/pyro/meta-android/recipes-core/systemd/systemd_%25.bbappend#L4 **** ENDING LOGGING AT Thu Jan 25 03:00:05 2018