**** BEGIN LOGGING AT Fri Jan 17 02:59:57 2020
Jan 17 08:11:52 <Herrie> Morning, seems I don't have CONFIG_PROC_DEVICETREE in my defconfig. Let me add that
Jan 17 08:15:50 <Tofe> Morning!
Jan 17 08:15:56 <Tofe> Herrie: ah, that might explain it
Jan 17 08:19:22 <Herrie> I've also added a ton of logging in init.cpp in android_system_core to see if I can pinpoint where stuff goes AWOL
Jan 17 09:29:16 <novaldex> Morning all!
Jan 17 09:29:39 <novaldex> been advised there's some maintenance planned for this weekend (starting today PST)
Jan 17 09:30:49 <novaldex> it may mean some interruption to service, although i'm not sure how much or for how long
Jan 17 09:30:57 <novaldex> I imagine not long though
Jan 17 09:42:37 <Herrie> novaldex: OK no worries. Is this the disk work that ka6sox was referring to earlier?
Jan 17 09:42:46 <Herrie> So we'll have more space available going forward?
Jan 17 10:07:25 <Herrie> Tofe: Hmmz that didn't fix anything
Jan 17 10:07:28 <Herrie> Still no firmware there
Jan 17 10:17:12 <Tofe> ah, that's annoying
Jan 17 10:19:12 <Tofe> What's a a bit surprising is that you had /proc/device-tree itself, without the CONFIG_PROC_DEVICETREE  option enabled.... it shouldn't be possible from what I read here and there
Jan 17 10:20:07 <Herrie> Tofe: Well it seems that the issue is some SELinux bits
Jan 17 10:20:33 <Herrie> I have a "Start running restorecon" but not an "End running restorecon" which I added in the logs
Jan 17 10:20:41 <Herrie> s/logs/logging
Jan 17 10:20:48 <Herrie> So I'll just disable the whole restorecon for now
Jan 17 10:20:55 <Herrie> See where I end up
Jan 17 10:21:39 <Tofe> ok
Jan 17 10:22:07 <Herrie> BEcause right after the failing restorecon I get "Nov 20 21:42:04 mido kernel[967]: [   36.774055] init: panic: rebooting to bootloader"
Jan 17 10:22:33 <Tofe> could be a hint yes
Jan 17 10:22:50 <Herrie> This is all a bit unchartered territory because we used to disable SELinux completely, but Mer at their side in 8.1 enables it in systemd at some point (which we might also want to do, but now for debugging I prefer to disable it).
Jan 17 10:23:13 <Herrie> I guess UBPorts also doesn't want SELinux since they use AppArmor and you cannot use both at the same time if I understood correctly?
Jan 17 10:24:14 <Tofe> I think Mer enables it because of the way they run android; it's not in a container, so they need some more security
Jan 17 10:24:52 <Tofe> I think they're in conflict with each other yes
Jan 17 10:25:05 <Tofe> or at least it becomes a hell to set it up
Jan 17 10:26:39 <Herrie> Yeah
Jan 17 10:26:44 <Herrie> So I'll try to disable it for now
Jan 17 10:26:55 <Herrie> 7.1 only had a 1st stage, now there's also a 2nd stage it seems
Jan 17 10:26:59 <Herrie> Or it was elsewhere in 7.1
Jan 17 10:27:06 <Herrie> Things were moved around quite a bit on Android side
Jan 17 10:32:27 <novaldex> Herrie: we'll wrap that in to the work this weekend
Jan 17 10:32:41 <novaldex> but going forward you'll have far more space, once we've enacted some of the other changes
Jan 17 10:33:17 <novaldex> think something on the order of a share of 10TB
Jan 17 10:33:39 <novaldex> rather than the 1 or 2 TB we can give you now
Jan 17 10:36:18 <Herrie> Great :)
Jan 17 10:36:38 <Herrie> Well still keep purging what's not needed, but it's nice that we won't run out of space all the time
Jan 17 10:36:51 <Tofe> Yes :)
Jan 17 10:37:18 <novaldex> housekeeping is always welcome! ;)
Jan 17 10:38:37 <Tofe> I think we could investigate setting up a script that removes testing/unstable builds older than a given age
Jan 17 10:42:17 <Herrie> Tofe: Yeah that doesn't hurt for sure
Jan 17 10:54:36 <Herrie> Tofe: OK I get a bit further now
Jan 17 11:10:17 <Herrie> Let me disable some more SELInux stuff
Jan 17 17:24:20 <Herrie> Tofe: This seems to look a bit better: https://paste.ubuntu.com/p/V8ts4VRc2Z/
Jan 17 17:26:19 <Herrie> Seems hwservicemanager is still failing at some point
Jan 17 17:26:40 <Herrie> And it cannot find vndservicemanager it seems: "Nov 20 21:42:04 mido kernel[971]: [   37.272411] init: do_start: Service vndservicemanager not found"
Jan 17 17:38:16 <Tofe> Herrie: do we have a miniservicemanager in halium?
Jan 17 17:39:31 <Tofe> Herrie: https://source.android.com/devices/architecture/hidl/binder-ipc#ipc
Jan 17 17:43:00 <Herrie> Tofe I don't think there's a miniservicemanager as such
Jan 17 17:43:08 <Herrie> There are 3 in mer-hybris
Jan 17 17:43:54 <Herrie> minisf, miniaf, minimediaservice as per: https://github.com/mer-hybris/android_system_core/pull/13/files
Jan 17 17:44:14 <Herrie> minisf = surfaceflinger
Jan 17 17:44:19 <Herrie> miniaf = audio
Jan 17 17:44:49 <Tofe> Herrie: interesting, I found this in your log: init: cannot find '/system/bin/touch', disabling 'bootdone': No such file or directory
Jan 17 17:45:07 <Herrie> Build by droidmedia: https://github.com/sailfishos/droidmedia/blob/master/Android.mk
Jan 17 17:45:16 <Herrie> Tofe: Yeah I saw that one too
Jan 17 17:45:19 <Herrie> That's weird indeed
Jan 17 17:45:27 <Tofe> there's no "touch" binary?
Jan 17 17:45:43 <Tofe> in that case, you'll never read "bootdone" :)
Jan 17 17:46:22 <Herrie> I'll look into that one
Jan 17 17:46:24 <Herrie> It's weird
Jan 17 17:46:32 <Herrie> Could be a silly path issue somehow
Jan 17 17:48:01 <Tofe> maybe they moved it, simply
Jan 17 18:18:45 <Herrie> Tofe: Seems so: https://paste.ubuntu.com/p/5q9ScqsX4K/
Jan 17 18:18:53 <Herrie> Just trying to figure out where the bootdone comes from
Jan 17 18:22:15 <Herrie> Tofe: Ah comes from our side, not Android
Jan 17 18:22:29 <Herrie> Was already wondering why my grep in Halium didn't give me anything
Jan 17 18:23:33 <Herrie> https://github.com/shr-distribution/meta-smartphone/blob/zeus/meta-android/recipes-core/android-system/android-system/10-boot-marker#L6
Jan 17 18:23:53 <Herrie> So should simply update thta .sh right?
Jan 17 18:31:53 <Herrie> Well that one is gone, but still not much luck for the rest: https://paste.ubuntu.com/p/YhrfpZSbSF/
Jan 17 18:36:21 <Tofe> Herrie: line 3370, I still see some selinux stuff
Jan 17 18:36:25 <Tofe> is that normal?
Jan 17 18:38:00 <Tofe> "Unable to open '/default.prop': Too many symbolic links encountered" <-- to be investigated too, maybe a symlink pointing on itself ?
Jan 17 18:41:29 <Tofe> the next step seems to be hwservicemanager
Jan 17 18:41:40 <Tofe> with init: property_set("ro.boottime.hwservicemanager", "35084964933") failed: property already set
Jan 17 18:42:08 <Herrie> Tofe: Yeah it's already set because it crashed before I think
Jan 17 18:42:19 <Herrie> Will check those pointers
Jan 17 18:42:52 <Herrie> Not sure why it's crashing, hard tot debug a bit
Jan 17 19:02:52 <Tofe> I bet it's hard to debug, yes
**** ENDING LOGGING AT Sat Jan 18 02:59:59 2020