**** BEGIN LOGGING AT Wed Apr 26 03:00:02 2017 Apr 26 04:22:33 Tofe: Mako actually boots with systemd232 Apr 26 04:22:41 Just the dropdowns in FirstUse have no data Apr 26 04:24:52 So it might be a N5 issue Apr 26 04:25:07 I also built qemux86 overnight, testing that in a minute Apr 26 04:25:38 Morning BTW ;) Apr 26 04:36:33 qemux86 boots, but graphics glitches... Maybe MESA not sure... Sent JaMa email with screenshot. Apr 26 05:05:56 Tofe: Mako log for comparison: https://bpaste.net/show/6cc97deeb896 Apr 26 05:16:54 Herrie: ok thanks, I'll have a look. That's weird. Apr 26 07:57:53 Tofe: There are plenty of other issues but those were expected (PulseAudio, ConnMan/oFono) Apr 26 07:58:15 But should be fairly easy to fix those by downgrading ConnMan/oFono. Apr 26 08:08:22 yep; I'm still trying to spot the difference between mako and hammerhead, it's far from obvious... Somehow, the Android layer doesn't start, but I don't know why Apr 26 08:10:23 I've pushed the qtmultimedia dependencies to krogoth branch Apr 26 08:10:40 JaMa: Thnx Apr 26 08:10:41 and added one more change to rebased pyro branch which might fix the fb in qemux86 Apr 26 08:11:00 I hope your fix fixes qemux86 Apr 26 08:11:10 Seems it booted alright to firstuse :P Apr 26 08:11:16 build is running now, will see soon Apr 26 08:11:18 So that's not a bad start Apr 26 08:11:28 Just couldn't ssh into it Apr 26 08:57:59 morning all Apr 26 08:58:27 Herrie, Tofe: you remember I posted a note about not being able to access a particular set of websites from my N4? Apr 26 08:58:51 turns out to be server side SSL ciphers that were at fault, the nginx out of the box ones aren't good enough Apr 26 08:59:12 i'm still trying to build a decent cipher string to satisfy ssllabs tests, but i've got it working so no need to worry Apr 26 08:59:25 useful to know though if we come across it again in future though Apr 26 09:00:32 ok, good :) Apr 26 09:31:52 Herrie|Pre3: Tofe: with the fix the graphics is shown correctly, but either mouse doesn't work or the UI is stuck Apr 26 09:34:10 JaMa: Well it's some progress but not a lot Apr 26 09:34:19 You also cannot SSH in? Apr 26 09:34:30 no, ssh also doesn't work for me Apr 26 09:35:07 That doesn't help the troubleshooting much :'( Apr 26 09:53:01 I'm trying with serial and that also doesn't work for whatever reason Apr 26 10:04:43 hmm sshd should be up Apr 26 10:04:45 [^[[0;32m OK ^[[0m] Listening on sshd.socket. Apr 26 10:05:09 only sntp failed Apr 26 10:05:10 [FAILED] Failed to start Simple Network Time Service Client. Apr 26 10:06:09 Could be port forwarding issue somehow? Apr 26 10:11:50 Apr 26 12:11:38 qemux86 sntp[618]: kod_init_kod_db(): Cannot open KoD db file /var/db/ntp-kod: No such file or directory Apr 26 10:11:53 Apr 26 12:11:38 qemux86 sntp[618]: [[0;1;39mkod_init_kod_db(): Cannot open KoD db file /var/db/ntp-kod: No such file or directory[[0m Apr 26 10:11:56 Apr 26 12:11:38 qemux86 sntp[618]: ntpserver.example.org lookup error Temporary failure in name resolution Apr 26 10:12:36 JaMa: I think this error has been around for last couple of OE releases tbh Apr 26 10:12:44 Would be good to get fixed of course Apr 26 10:12:47 there are issues with systemd and resolving http://lists.openembedded.org/pipermail/openembedded-core/2017-April/136074.html Apr 26 10:13:00 depending on sshd configuration it might be the same root cause Apr 26 10:13:27 at least I have working serial port now Apr 26 10:15:16 and locally the sshd works Apr 26 10:30:46 really strange, even running sshd manually doesn't show any incomming connection even with that tmpfiles.d fix applied Apr 26 10:31:16 might be virtualbox breaking the port forwarding (I've recently upgraded to 5.1.20.114628) Apr 26 10:34:49 hmm that was it, changing the emulated network card from "PCnet-FAST III (Am79C973)" which was default in my VirtualBox after importing our appliance to "Intel PRO/1000 MT Desktop (82540EM)" it works fine Apr 26 10:35:07 possibly missing or not loaded driver for the first, because only now I see: Apr 26 10:35:10 enp0s3 Link encap:Ethernet HWaddr 08:00:27:AF:65:31 Apr 26 10:35:20 before that there was only lo and sit0 Apr 26 10:47:38 mystery possibly explained, latest linux-yocto has: Apr 26 10:47:40 CONFIG_PCNET32=m Apr 26 10:47:59 which should be the correct driver for PCnet-FAST cards Apr 26 10:48:06 but this module isn't even installed in our images Apr 26 10:49:43 installing it manually fixes it Apr 26 10:49:43 [ 3.052968] pcnet32: PCnet/FAST III 79C973 at 0xd000, 08:00:27:af:65:31 assigned IRQ 19 Apr 26 10:49:46 [ 3.056893] pcnet32: Found PHY 0022:561b at address 0 Apr 26 10:52:05 and the mouse issue might be different event device (dunno how it's configured in our image) Apr 26 10:52:09 N: Name="VirtualBox mouse integration" Apr 26 10:52:10 is now event6 Apr 26 11:00:30 switch to E1000 driver pushed to pyro branch (not completely tested) Apr 26 11:30:41 great :) now I have to solve that annoying hammerhead mystery... Apr 26 11:33:18 JaMa: Great Apr 26 18:19:53 ah, I think I have a lead Apr 26 18:20:24 meta-smartphone/meta-android/recipes-core/udev/udev-extraconf/65-android.rules isn't in the resulting rootfs, somehow Apr 26 18:24:41 ah, no, that's not it Apr 26 18:25:26 but the /dev/log_* character devices are not created, and it sure doesn't help Apr 26 18:47:40 Herrie: we'll need to take some decision about upgrading systemd & glibc, at some point, I guess... I think we'll manage with pyro, but there will be a point where our embedded devices won't match anymore Apr 26 18:58:52 ok, I managed to get to the UI Apr 26 18:59:07 it looks like /system isn't mounted yet when lxc starts, somehow Apr 26 19:06:19 yes, if I change a bit android-system.service, it's fine Apr 26 19:09:13 As for the empty lists, I see lots of "shared memfd open() failed: Function not implemented", could it be the cause? Apr 26 19:09:34 This kind of functionality only comes with kernels >= 3.17... Apr 26 19:11:47 (the lxc dependency on /system, /persist etc is device-dependent, we should be a bit careful with the systemd config) Apr 26 19:57:53 Tofe: Well OE has quite some other devices with older kernels Apr 26 19:58:02 They will probably hack stuff to make it work :P Apr 26 19:58:18 Tofe: Well shared memfd open() not sure about that Apr 26 19:58:42 It could be that luna-sysservice is the issue somehow Apr 26 19:59:28 That might actually relate to PulseAudio somehow Apr 26 20:00:24 Tofe: As per https://github.com/kunquat/kunquat/issues/580 and https://github.com/gcompris/GCompris-qt/issues/187 it seems PA related Apr 26 20:01:08 For the empty lists we could see if we can simply luna-send the commands that LNC does to populate the values? Apr 26 20:02:16 Do we have configurator running properly? Apr 26 20:02:31 I can imagine iwth configurator not running we don't have this info populated in the right places **** ENDING LOGGING AT Thu Apr 27 03:00:00 2017