Aug 06 14:59:44 tracker-extract-3[5725]: Disallowed syscall "getsockopt" caught in sandbox
Aug 06 14:59:44 systemd[1]: Started Process Core Dump (PID 5777/UID 0).
Aug 06 14:59:44 gnome-software[4747]: g_task_return_pointer: assertion 'G_IS_TASK (task)' failed
Aug 06 14:59:44 systemd-coredump[5778]: [🡕] Process 5725 (tracker-extract) of user 1000 dumped core.
Stack trace of thread 5772:
#0 0x00007fec4fc57989 __restore_rt (libc.so.6 + 0x57989)
ELF object binary architecture: AMD x86-64
Aug 06 14:59:44 systemd[1]: systemd-coredump@4-5777-0.service: Deactivated successfully.
Aug 06 14:59:44 systemd[4202]: tracker-extract-3.service: Main process exited, code=dumped, status=31/SYS
Aug 06 14:59:44 systemd[4202]: tracker-extract-3.service: Failed with result 'core-dump'.
And the systemd units:
ox@orca$ systemctl --user status tracker-*
× tracker-extract-3.service - Tracker metadata extractor
Loaded: loaded (/usr/lib/systemd/user/tracker-extract-3.service; static)
Active: failed (Result: core-dump) since Tue 2024-08-06 15:16:22 CDT; 2min 30s ago
Duration: 1.512s
Process: 7372 ExecStart=/usr/lib/tracker-extract-3 (code=dumped, signal=SYS)
Main PID: 7372 (code=dumped, signal=SYS)
CPU: 763ms
● tracker-miner-fs-3.service - Tracker file system data miner
Loaded: loaded (/usr/lib/systemd/user/tracker-miner-fs-3.service; disabled; preset: disabled)
Active: active (running) since Tue 2024-08-06 14:59:33 CDT; 19min ago
Main PID: 5363 (tracker-miner-f)
Tasks: 8 (limit: 4915)
CPU: 2.667s
CGroup: /user.slice/user-1000.slice/user@1000.service/background.slice/tracker-miner-fs-3.service
└─5363 /usr/lib/tracker-miner-fs-3
After adding the “i915.enable_guc=2” kernel boot option, tracker3 no longer crashes with each extraction. Instead, it just dumps core and moves to the next (hundreds of coredumps). It eventually finishes, but it does not actually extract any data because it coredumps on each file.
Yes … after a reset … a CLEAN reset (tracker can be finicky) …
It appears to be a tracker problem, not a “how I have my box set up” problem. And quite possibly, it is a tracker mated with intel problem, not an intel multi-display problem.
and
And if I am following these threads correctly, I might be able to hack something akin to adding " ALLOW_RULE (getsockopt)" in src/libtracker-miners-common/tracker-seccomp.c.
Why not? I already maintain my own version of nautilus … though this does look much trickier than the easy fixes I maintain in nautilus.
Or maybe I’ll get lucky and be able to build a newer version of tracker …
I will certainly report a bug, but I hope to have a better understanding of the problem (and if I’m lucky, the solution) when I do.
I just KNEW I should have ignored your advice, and now I genuinely want to run on intel with switcheroo … and I’ll have to work for it, but the offload is just too sweet!
