| T42 | <rahulyx> @mal I checked journalctl. I see a bunch of AVC avc: denied and another bunch of I/O errors | 04:26 |
|---|---|---|
| T42 | <rahulyx> @elros34 apexd-bootstrap is not failing, checked dmesg logs | 04:26 |
| T42 | <rahulyx> AVC avc: denied { mount } for pid=1180 comm="mount" name="/" dev="mmcblk0p52" ino=2 scontext=u:r:kernel:s0 tcontext=u:object_r:labeledfs:s0 tclass=filesystem permissive=1....... Is this critical? We changed /system to / in fstab yesterday | 04:28 |
| T42 | <rahulyx> The journalctl dump is here: https://paste.mozilla.org/XBKuhCka | 04:31 |
| *** amccarthy is now known as Guest3445 | 09:35 | |
| *** amccarthy_ is now known as amccarthy | 09:35 | |
| *** amccarthy is now known as Guest3446 | 09:44 | |
| *** amccarthy_ is now known as amccarthy | 09:44 | |
| T42 | <elros34> this is unreadable, read hadk-hot, disable audit via kernel param, enable persistet logging and get full journalctl --no-page. Do you have your repositories in github? | 10:45 |
| T42 | <rahulyx> I am rebuilding with audit=0. Yeah, the repos are on github: https://github.com/Sailfish-on-karatep | 16:02 |
| T42 | <rahulyx> I've disabled audit on kernel param and enabled journal persist. Here are the logs: https://paste.mozilla.org/Ss67K0jr | 17:51 |
| T42 | <rahulyx> This looks bad: "Device /dev/sailfish/home does not exist or access denied." | 17:51 |
| T42 | <rahulyx> Also, a bunch of symlinking are failing. Example: "Process '/bin/ln -s /dev/block/mmcblk0p28 /dev/block/platform/soc/7824900.sdhci/by-num/p28' failed with exit code 1." Something wrong with the mountpoints? | 17:52 |
| T42 | <rahulyx> I've masked droid-hal-init btw | 17:53 |
| mal | that "Device /dev/sailfish/home does not exist or access denied." is normal | 18:56 |
| mal | those symlink things are not bad either | 18:57 |
| T42 | <rahulyx> Alright. Do you see anything else that looks off? | 19:06 |
| T42 | <rahulyx> Let me start droid-hal-init manually and capture the logs | 19:07 |
| T42 | <rahulyx> This what I captured: https://paste.mozilla.org/dCw5ALpw | 19:14 |
| T42 | <rahulyx> I see "Service 'apexd-bootstrap' (pid 1900) exited with status 0 waiting took 0.054000 seconds". Is it failing? | 19:16 |
| T42 | <elros34> status 0 is ok | 19:18 |
| T42 | <rahulyx> Exactly | 19:18 |
| T42 | <elros34> but last message suggest it's apex issue | 19:19 |
| T42 | <elros34> does it reboot? | 19:19 |
| T42 | <elros34> IIRC failure of boringssl_selft_test usually leads to reboot | 19:20 |
| T42 | <rahulyx> Yeah, rebooted back to Android. I haven't flashed hybris-boot.img yet | 19:20 |
| T42 | <elros34> apply patch, there is no reason to debug it more, I bet this is the issue. apex also prints that device does not support updatable apex, device is quiete old so | 19:22 |
| T42 | <elros34> it's common issue | 19:22 |
| T42 | <rahulyx> The apex patch that you've shared earlier? | 19:22 |
| T42 | <elros34> yes | 19:22 |
| T42 | <rahulyx> Cool, let me apply | 19:23 |
| T42 | <elros34> build hybris-hal and droid-hal and this should help | 19:23 |
| T42 | <rahulyx> Got it | 19:23 |
| T42 | <rahulyx> Just to be sure, I apply this patch, right? - "If apexd-bootstrap fails beacause of not updatable/flattened apexes here is experimental fix: https://paste.debian.net/hidden/a2302262/" Also, I patch system/core/init/init.cpp? | 19:33 |
| T42 | <rahulyx> I mean, files at system/core/init/ | 19:34 |
| T42 | <elros34> I thinks so, it was long time ago when I wrote it but you can't accidentaly change incorrect file so don't worry | 19:35 |
| T42 | <rahulyx> Okayy | 19:35 |
| *** ggabriel is now known as Guest3480 | 19:45 | |
| *** ggabriel is now known as Guest3481 | 20:01 | |
| *** ggabriel is now known as Guest3483 | 20:09 | |
| T42 | <rahulyx> @elros34 that helped. So droid-hal-init is running now. The display is still blank. Let me check journalctl logs | 20:45 |
| mal | just to be sure, you built 4.6.0 sailfish? | 20:46 |
| T42 | <rahulyx> Yeah, 4.6.0.13 | 20:48 |
| T42 | <elros34> based on previous log you may need some symlink for missing library but for some reason sometimes it's false error | 20:48 |
| T42 | <rahulyx> Can you please help me out with this? Which library is missing? | 20:51 |
| T42 | <rahulyx> I see this: Could not restart 'surfaceflinger': Cannot find 'surfaceflinger_HYBRIS_DISABLED': No such file or directory | 20:51 |
| T42 | <elros34> thak is ok, _HYBRIS_DISABLED services are disabled for a reason. Get fresh journal | 20:53 |
| T42 | <elros34> sometimes you need create libandroidicu symlink or some similar lib | 20:54 |
| T42 | <rahulyx> Here you go: https://paste.mozilla.org/9tsrGFA0 | 20:57 |
| T42 | <rahulyx> Okay, let me check the libandroidicu.so symlink | 20:58 |
| T42 | <rahulyx> I see these two in the logs: library "libhardware.so" not found / library "libandroidicu.so" not found | 20:59 |
| T42 | <rahulyx> Ran a regex, the following are missing: libcutils.so, libz.so, libc++.so, libbinder.so, libandroidicu.so, libhardware.so | 21:04 |
| T42 | <rahulyx> Let me locate and symlink them | 21:04 |
| T42 | <elros34> probably libhardware is not an issue, check logcat and dont symlink everythin | 21:04 |
| T42 | <rahulyx> Sure | 21:05 |
| T42 | <elros34> you maybe do this blindly for libandroidicu, leave all others unless you debugged this with strace | 21:05 |
| T42 | <rahulyx> Alright | 21:06 |
| T42 | <elros34> you maybe at least temporary disable this livedisplay service to stop crashing it. search for hwc/hwcomposer in logcat | 21:07 |
| T42 | <rahulyx> Let me check | 21:14 |
| T42 | <rahulyx> I can't locate libandroidicu.so. There are some files in .intermediates, but nothing in apex/ https://paste.mozilla.org/cN8sh1s1#L1 | 21:26 |
| T42 | <elros34> on device. That was just suggestion. check logcat | 21:30 |
| T42 | <rahulyx> Sorry, just realised that. Located it at: apex/com.android.art/lib64/libandroidicu.so on device | 21:31 |
| T42 | <rahulyx> Checked logcat: libraries "libandroidicu.so" and "libnativehelper.so" are not found | 21:39 |
| T42 | <rahulyx> I also see: ServiceManager: Waiting for service 'vendor.qcom.PeripheralManager' on '/dev/vndbinder' and ServiceManager: Waiting for service 'display.qservice' on '/dev/vndbinder'... at the end | 21:40 |
| T42 | <rahulyx> Searched for hwcomposer: droid-hal-init: Service 'vendor.hwcomposer-2-1' (pid 6211) exited with status 1 / https://paste.mozilla.org/OuH8ruqS#L23,29 | 21:43 |
| T42 | <rahulyx> Let me symlink libandroidicu.so | 21:44 |
| T42 | <rahulyx> I'm not sure if I'm symlinking it right. So I create sparse/odm/lib64/ on the machine's dcd dir, right? Or is it on device? | 22:25 |
| T42 | <rahulyx> I've done it inside dcd, like in https://github.com/sailfishos-on-tucana/droid-config-tucana/tree/master/sparse/odm/lib64 . logcat still says libandroidicu.so is missing | 22:28 |
| T42 | <rahulyx> linker : CANNOT LINK EXECUTABLE "/usr/libexec/droid-hybris/system/bin/minisfservice": library "libandroidicu.so" not found: needed by /system/lib/libmedia.so in namespace (default) | 22:29 |
| T42 | <elros34> do it on device, forget about droid-config until it really helps | 22:29 |
| T42 | <rahulyx> On device, is it at /sparse/odm/lib64/? | 22:30 |
| T42 | <elros34> no, sparse in droid config is / on device | 22:30 |
| mal | no, just /odm/lib64 | 22:30 |
| T42 | <rahulyx> Thanks! | 22:31 |
| T42 | <rahulyx> It is symlinked: https://paste.mozilla.org/03d5x6Uj | 22:31 |
| T42 | <rahulyx> Still, the library seems to be missing | 22:32 |
| mal | and that path /apex/com.android.art/lib64/libandroidicu.so is the correct one? | 22:32 |
| T42 | <rahulyx> Yeah | 22:32 |
| T42 | <rahulyx> These are the contents of /apex/com.android.art/lib64/ https://paste.mozilla.org/gkMZwMid | 22:33 |
| mal | add one to /odm/lib also, of course pointing to lib version of the same file in apex | 22:34 |
| T42 | <rahulyx> Let me try | 22:34 |
| mal | library "libandroidicu.so" not found: needed by /system/lib/libmedia.so | 22:34 |
| mal | that suggests it tries to find 32-bit one from lib | 22:35 |
| mal | note that /system/lib/ | 22:35 |
| T42 | <rahulyx> Ahhhh yes. I did not see that! | 22:35 |
| mal | if you have 64-bit sailfish then you also probably need those in lib64 anyway | 22:45 |
| T42 | <rahulyx> I'm symlinking both 32 and 64 to be safe | 22:46 |
| T42 | <rahulyx> The library missing errors are gone now, still no display | 22:55 |
| T42 | <rahulyx> I see "vendor.lineage.livedisplay@2.0-impl-sdm: Failed to initialize SDM backend" https://paste.mozilla.org/NPghY1eX | 22:56 |
| T42 | <rahulyx> How do I disable livedisplay? | 23:02 |
| T42 | <rahulyx> Here is the logcat since boot: https://paste.mozilla.org/hktnabrf | 23:17 |
| T42 | <rahulyx> @mal, saw this on irc archive "Parcel : Expecting header 0x53595354 but found 0x564e4452. Mixing copies of libbinder?". Killed the process. Now throws "CANNOT LINK EXECUTABLE "/vendor/bin/vndservicemanager": library "/usr/libexec/droid-hybris/system/lib64/libselinux_stubs.so" needed or dlopened by "/vendor/bin/vndservicemanager" is not accessible for the namespace "(default)"" | 23:26 |
| T42 | <rahulyx> E linker : library "/usr/libexec/droid-hybris/system/lib64/libselinux_stubs.so" ("/usr/libexec/droid-hybris/system/lib64/libselinux_stubs.so") needed or dlopened by "/vendor/bin/vndservicemanager" is not accessible for the namespace: [name="(default)", ld_library_paths="", default_library_paths="/odm/lib64:/vendor/lib64:/vendor/lib64/hw:/vendor/lib64/egl", permitted_paths="/odm:/vendor:/system/vendor"] | 23:29 |
| mal | hmm, try killing that vndservicemanager process and see if restart and if anything changes | 23:32 |
| T42 | <rahulyx> vndservicemanager doesn't restart | 23:42 |
| mal | hmm | 23:42 |
| mal | usually those should restart | 23:42 |
| T42 | <rahulyx> I think it is trying to restart | 23:44 |
| T42 | <rahulyx> library "/usr/libexec/droid-hybris/system/lib64/libselinux_stubs.so" ("/usr/libexec/droid-hybris/system/lib64/libselinux_stubs.so") needed or dlopened by "/vendor/bin/vndservicemanager" is not accessible for the namespace: [name="(default)", ld_library_paths="", default_library_paths="/odm/lib64:/vendor/lib64:/vendor/lib64/hw:/vendor/lib64/egl", permitted_paths="/odm:/vendor:/system/vendor"] | 23:44 |
| T42 | <rahulyx> It keeps throwing this | 23:44 |
| mal | ok | 23:45 |
| mal | as a temporary ugly test copy that libselinux_stubs.so to /odm/lib64/ | 23:45 |
| T42 | <elros34> why do you even have these selinux stubs there? did you add them or they are from submodule? | 23:45 |
| mal | aren't those built by default | 23:46 |
| T42 | <elros34> I mean in *servicemanager*.rc files | 23:46 |
| T42 | <rahulyx> haven't added them | 23:46 |
| T42 | <elros34> ok my bad | 23:46 |
| mal | which android base is that again? | 23:48 |
| T42 | <rahulyx> 18.1 | 23:49 |
| mal | faq has things related to those errors | 23:50 |
| T42 | <elros34> https://github.com/Sailfish-on-karatep/droid-config-karatep/commit/1a3461fea59740fd1c9f8794ab19b30aeee68515 | 23:51 |
| mal | hmm, wondering why that was done | 23:51 |
| mal | seems odd | 23:51 |
| mal | try commenting out the selinux stubs lines from that /usr/libexec/droid-hybris/system/etc/init/vndservicemanager.rc | 23:52 |
| mal | ah, you have taken some old stuff from 15.1 instructions in faq? | 23:53 |
| T42 | <rahulyx> Ah I did apply the patches from hybris-16.0 | 23:53 |
| T42 | <rahulyx> They're unnecessary? | 23:54 |
| mal | probably break things | 23:54 |
| mal | only relevant part from 16 in 18.1 base is cloning libhybris | 23:54 |
| mal | and applying patches | 23:54 |
| T42 | <rahulyx> Ah alright. Let me reverse them | 23:55 |
| mal | remove them from device first for testing | 23:55 |
| T42 | <rahulyx> Let me try commenting first | 23:55 |
| T42 | <rahulyx> Yes | 23:55 |
| mal | that disabled service rc is probably fine | 23:55 |
| T42 | <rahulyx> Damn! The UI is up | 23:59 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!