2016-02-06 00:13:07	TheAppleMan	Oh I'm still in the initramfs ?
2016-02-06 00:18:21	TheAppleMan	My kernel doesn't have ANDROID_RAM_CONSOLE though
2016-02-06 00:59:21	@kido	yes you get a shell from the ramdisk
2016-02-06 00:59:34	@kido	so you can switch root yourself and redirect systemd as you want
2016-02-06 00:59:48	@kido	and/or modify the services scripts before they are executed
2016-02-06 01:02:57	@kido	thanks for the pull request, merged
2016-02-06 01:03:24	@kido	+ you probably noticed that I gave you write access to AsteroidOS/meta-bass-hybris, so that you can work upstream now
2016-02-06 01:21:10	TheAppleMan	:)
2016-02-06 01:24:58	TheAppleMan	I did want to clean up the branch before sending it up, but it's fine?
2016-02-06 11:18:26	@kido	TheAppleMan: you can still clean up the branch now
2016-02-06 11:18:47	@kido	not sure if I missed other messages, my bouncer disconnected
2016-02-06 14:04:59	TheAppleMan	> init started!
2016-02-06 14:04:59	TheAppleMan	> SELinux:  Could not open sepolicy:  No such file or directory
2016-02-06 14:04:59	TheAppleMan	> SELinux: Failed to load policy; rebooting into recovery mode
2016-02-06 14:06:46	@kido	oh it's the android-init
2016-02-06 14:07:31	@kido	https://github.com/AsteroidOS/android_system_core/commit/9bca250105750857e77789d086dd154e4f034497 did you use this repository/patch in your android's manifest ?
2016-02-06 14:08:43	@kido	mhh according to your android_manifest repository it seems you used the asteroidos repository
2016-02-06 14:10:11	TheAppleMan	doesn't that patch mean selinux is still enabled?
2016-02-06 14:10:54	@kido	mhhh... something is clearly wrong with this patch indeed ><
2016-02-06 14:11:32	@kido	so the error comes from here https://github.com/AsteroidOS/android_system_core/blob/30de2ebdb65221a495ebb1b0518db8ee96f5ab29/init/init.cpp#L958 
2016-02-06 14:12:11	@kido	I think that on the g watch I gave an argument to the kernel to disable selinux and selinux_android_load_policy was not used
2016-02-06 14:12:45	TheAppleMan	i left selinux=0 in the linux-bass/img_info cmdline
2016-02-06 14:13:19	@kido	https://android.googlesource.com/platform/external/libselinux/+/master/src/android.c#1517
2016-02-06 14:16:35	TheAppleMan	SECURITY_SELINUX_POOTPARAM=n
2016-02-06 14:18:00	TheAppleMan	s/P/B/
2016-02-06 14:19:29	@kido	ah probably that
2016-02-06 14:19:52	TheAppleMan	does it need to be added to https://github.com/AsteroidOS/meta-asteroid/blob/master/scripts/check-config ?
2016-02-06 14:20:07	@kido	if it solves the problem then yes
2016-02-06 14:20:20	TheAppleMan	i'm recompiling now
2016-02-06 14:21:00	@kido	a couple of config options should be cleared from check-config because I got them from ubuntu touch and some might not be mandatory to asteroid
2016-02-06 14:21:10	@kido	cool :)
2016-02-06 14:33:24	@kido	by the way I noticed your latest tweet about boot process troubleshooting, don't hesitate to add that kind of tips to the wiki if you have the time :)
2016-02-06 14:35:02	TheAppleMan	sure
2016-02-06 14:44:46	TheAppleMan	yay! it worked!
2016-02-06 14:45:08	@kido	do you mean, only android-init or the whole thing?
2016-02-06 14:45:29	TheAppleMan	systemd is starting, just the logo on the screen though
2016-02-06 14:45:38	@kido	yay :D
2016-02-06 14:45:43	@kido	so you can use adb from the rootfs now
2016-02-06 14:46:27	TheAppleMan	logd and battery-statefs have errors though
2016-02-06 14:46:36	@kido	battery-statefs? oO
2016-02-06 14:46:48	TheAppleMan	Feb 06 13:46:26 bass mce[333]: modules/battery-statefs.c: tracker_open(): /run/state/namespaces/Battery/State: open: No such file or directory
2016-02-06 14:46:48	@kido	I'm working on it at this very moment
2016-02-06 14:46:59	@kido	oh yeah alright it also happens on the g watch
2016-02-06 14:47:03	@kido	it comes from statefs
2016-02-06 14:47:21	@kido	what about logd?
2016-02-06 14:47:37	TheAppleMan	Feb 06 13:47:25 bass init: Starting service 'logd'...
2016-02-06 14:47:37	TheAppleMan	Feb 06 13:47:25 bass init: Service 'logd' (pid 470) exited with status 1
2016-02-06 14:47:37	TheAppleMan	Feb 06 13:47:25 bass init: Service 'logd' (pid 470) killing any children in process group
2016-02-06 14:48:46	@kido	alright then you should bitbake strace, adb push your build/tmp-glibc/work/armv7a-vfp-neon-oe-linux-gnueabi/strace/4.10-r0/image/ directory to / and then use strace /system/bin/logd you should see errors
2016-02-06 15:01:33	TheAppleMan	CANNOT LINK EXECUTABLE DEPENDENCIES: cannot locate symbol "__cxa_finalize" referenced by "libdsyscalls.so"...
2016-02-06 15:03:20	@kido	mhh I wish I would never see that error again
2016-02-06 15:08:21	@kido	long story short: libhybris requires a patched android's bionic libc. since android 5.1 patching bionic creates issues when dynacaly linking with libdsyscalls.so (which is needed by libhybris) from android app, and not when linking staticaly
2016-02-06 15:09:10	@kido	the workaround that is currently used on dory is to compile staticaly the only three binaries that are used from android's /system directory
2016-02-06 15:09:23	@kido	which are init, logd and servicemanager
2016-02-06 15:09:42	@kido	for some reasons your logd seems not to be staticaly compiled
2016-02-06 15:10:07	@kido	ok I just didn't push modifications from my local android_system_core
2016-02-06 15:12:31	@kido	"git pull" in the 'android/system/core' directory,  "mmma system/core/logd/" and "adb push out/target/product/generic/system/bin/logd /system/bin/logd"
2016-02-06 15:25:43	TheAppleMan	okay, that's fixed, but i still just have the logo on screen
2016-02-06 15:27:02	@kido	alright so android-init is running
2016-02-06 15:27:06	@kido	what about asteroid-launcher ?
2016-02-06 15:31:39	@kido	from adb shell, su ceres and then "EGL_PLATFORM=hwcomposer XDG_RUNTIME_DIR=/run/user/1000 lipstick -plugin evdevtouch:/dev/input/event0 -platform hwcomposer"
2016-02-06 15:32:02	@kido	if nothing shows then add strace between the envvars and lipstick
2016-02-06 15:32:32	@kido	and paste that somewhere
2016-02-06 15:34:33	TheAppleMan	https://gist.githubusercontent.com/theappleman/daaee5fa70dba4e553a8/raw/dce031dda6f386e549c6edcd46204ae9dbc76632/lipstick.strace
2016-02-06 15:36:53	@kido	mhhhh it actualy reminds me of something
2016-02-06 15:37:26	@kido	try to run it as root just to make sure...
2016-02-06 15:37:50	TheAppleMan	segfault
2016-02-06 15:37:56	@kido	you'll probably have the same error but we never know, at that point of debuging running as root can remove some doubts
2016-02-06 15:37:58	@kido	alright
2016-02-06 15:39:35	@kido	well you will probably have to modify src/meta-asteroid/classes/asteroid-image.bbclass to install asteroid-launcher-dbg and gdb
2016-02-06 15:39:43	@kido	you will get all the debuging symbols from qt
2016-02-06 15:40:13	@kido	I believe that you'll have to recreate a whole rootfs because your current rootfs only has like ~20% of free space and it isn't enough for the qt debuging symbols
2016-02-06 15:40:26	@kido	so adb pushing the qtbase-dbg wouldn't fit the image
2016-02-06 15:40:41	@kido	and asteroid-image-dbg is probably too fat too
2016-02-06 15:41:05	@kido	so modifying asteroid-image to install "only what's needed" would probably make your life easier
2016-02-06 15:42:19	@kido	once you get the symbols and gdb run asteroid-launcher in gdb and then take a look at the backtrace
2016-02-06 15:43:23	@kido	mhhh, before doing that you should try to run test-glesv2 from libhybris as a standalone app
2016-02-06 15:43:36	@kido	with EGL_PLATFORM=hwcomposer
2016-02-06 15:44:00	TheAppleMan	sh-4.3# EGL_PLATFORM=hwcomposer test_glesv2
2016-02-06 15:44:00	TheAppleMan	failed to get gralloc module: (Invalid argument)
2016-02-06 15:44:00	TheAppleMan	test_glesv2: eglplatform_hwcomposer.cpp:27: void hwcomposerws_init_module(ws_egl_interface*): Assertion `0' failed.
2016-02-06 15:44:00	TheAppleMan	Aborted
2016-02-06 15:44:10	@kido	strace on that
2016-02-06 15:46:09	TheAppleMan	https://gist.github.com/theappleman/daaee5fa70dba4e553a8#file-test_glesv2-strace
2016-02-06 15:46:46	@kido	the bothersome part of your qt error is that it doesn't seem to be related to libhybris
2016-02-06 15:50:05	@kido	by the way you can use the -s 10000 option of strace to get longer strings
2016-02-06 15:50:30	@kido	it would give more info on the lines 745, 746 and 747 of your log
2016-02-06 15:51:09	@kido	especialy the 745 in fact, since you gave the full log some lines ago
2016-02-06 15:54:19	TheAppleMan	updated
2016-02-06 15:56:25	@kido	then you get your issue ;)
2016-02-06 16:04:01	@kido	if you "grep -rF "androidGetTid" /system/*" what do you get?
2016-02-06 16:09:06	@kido	TheAppleMan: git pull your android/system/core again, recompiles libutils and push it to your /system directory I added the androidGetTid symbol
2016-02-06 16:11:03	TheAppleMan	sh-4.3# grep -rF "androidGetTid" /system/*
2016-02-06 16:11:03	TheAppleMan	Binary file /system/lib/libandroid_runtime.so matches
2016-02-06 16:11:03	TheAppleMan	Binary file /system/lib/libbinder.so matches
2016-02-06 16:11:03	TheAppleMan	Binary file /system/lib/libstagefright.so matches
2016-02-06 16:12:42	@kido	alright
2016-02-06 16:21:38	TheAppleMan	okay, i gotta go out now so i'll compile it and pick it up later
2016-02-06 16:21:54	@kido	alright, see you later :)
2016-02-06 16:21:58	@kido	have a nice day
2016-02-06 16:22:18	@kido	and good job for your work
2016-02-06 16:23:32	TheAppleMan	btw https://github.com/AsteroidOS/asteroid/wiki/Boot-Process#troubleshooting-the-boot-process
2016-02-06 16:24:31	@kido	:)
2016-02-06 17:43:54	@kido	ah! finally statefs is fixed and we have battery percentage shown on the quick settings page
2016-02-06 17:44:35	@kido	it was such a mess, I wouldn't expect it to be one of the hardest component to integrate... anyway, let's move on to more interesting stuff now
2016-02-06 19:30:46	locusf	I really look forward to pair Nemo fp2 and Asteroid gwatch with a self-made free protocol of my choosing
2016-02-06 19:33:06	@kido	it will happen someday :D