I applied the last update yesterday (apt-get update && apt-get dist-upgrade) and rebooted. Since then I get the sad face. Looking at the logs it says that kodi returned error with error 139. I have been looking for around and I can’t find anything that could generat this.
If I run manually /usr/lib/kodi/kodi.bin I get the same errors as in the system log, which are not descriptive at all:
$ /usr/lib/kodi/kodi.bin
unknown
unknown
Segfault
I have also tried reinstalling rbp2-mediacenter-osmc package with no luck.
I am running out of ideas - the other things installed in this raspberry seem to work ok.
The other things are openmediavault - this includes samba, netatalk and I have also installed duplicati, but that is not a package and it is installed in /opt
I can’t tell anything else, and overall nothing that could be affecting kodi to the extent of segfaulting. I have a kodi plugin to watch IP TV, but my test moving out /home/osmc/.kodi didn’t work. That should have removed that plugin, shouldn’t that?
I can’t upload the full logs - it says the max size is exceeded - here is the output of grab-logs -XKvp Please tell me if you need more: http://paste.osmc.io/uconirezoz
By not enough data, I mean that I am on holiday, where I currently am getting about 14.4Kbps down, so can’t easily grab your logs.
The SQLite messages don’t look like a crash to me. You get SIGILL because we link Kodi against libcrypto, which tries to test for processor features and uses SIGILL to handle this, but you can ignore this with:
handle SIGILL nostop
Remember to pass full arguments to kodi.bin (check /usr/bin/mediacenter for argv) to reduce the number of variables.
Passing full arguments and ignoring SIGILL, I get the two “unknown” messages that get registered on the log, and this output in GDB (after all the sqlite messages)
Cannot access memory at address 0x0
Program received signal SIGILL, Illegal instruction.
[New Thread 0x74a983f0 (LWP 13676)]
unknown
unknown
[New Thread 0x740ff3f0 (LWP 13690)]
[New Thread 0x736ff3f0 (LWP 13691)]
[New Thread 0x72eff3f0 (LWP 13692)]
[New Thread 0x724ff3f0 (LWP 13693)]
[New Thread 0x71cff3f0 (LWP 13694)]
[New Thread 0x714ff3f0 (LWP 13695)]
[New Thread 0x70cff3f0 (LWP 13696)]
[New Thread 0x704ff3f0 (LWP 13697)]
[New Thread 0x6fcff3f0 (LWP 13698)]
Program received signal SIGSEGV, Segmentation fault.
0x0210a428 in ?? ()
Some advance - after reinstalling armv7-libsqlite-osmc package the error was different:
terminate called after throwing an instance of 'dbiplus::DbErrors'
So I removed again the .kodi dir and now everything works! I think that something happened when installing that package when upgrading, and this made the db to get corrupt (or something).
In my (old) .kodi directory I have both a userdata/Database and userdata/db directories. The former has only one db and a CDDB folder. The userdata/db folder has the usual dbs - are some of them a copy of the others? I have been able to restore all my config just copying them over the Database directory.
This sounds like you have some bad filesystem corruption. Check your SD card, power supply, and consider a reinstall. You will likely have some problems further down the line. You certainly shouldn’t have to reinstall any packages manually via APT.