[Bug runtime/30393] Lockdep error in _stp_ctl_work_callback()/_stp_request_exit() on a fedora debug kernel

agentzh at gmail dot com sourceware-bugzilla@sourceware.org
Mon May 1 18:19:12 GMT 2023


https://sourceware.org/bugzilla/show_bug.cgi?id=30393

--- Comment #3 from agentzh <agentzh at gmail dot com> ---
Seems like the kernel backtrace is a little bit misleading. It actually has
nothing to do with _stp_request_exit() after all:

```
(gdb) l *_stp_ctl_work_callback+0x8a
0x6cbb is in _stp_ctl_work_callback
(/path/to/systemtap/runtime/transport/transport.c:559).
554             stp_spin_unlock_irqrestore(&_stp_ctl_ready_lock, flags);
555
556             _stp_runtime_entryfn_put_context(c);
557
558             if (do_io)
559                     wake_up_interruptible(&_stp_ctl_wq);
560
561             /* if exit flag is set AND we have finished with
systemtap_module_init() */
562             if (unlikely(_stp_exit_flag && _stp_probes_started))
563                     _stp_request_exit();
```

And I also checked the disassembly of the .ko file, it is indeed the
`wake_up_interruptible()` call:

```
0000000000006be1 <_stp_ctl_work_callback>:
...
/path/to/systemtap/runtime/transport/transport.c:559
                wake_up_interruptible(&_stp_ctl_wq);
    6c50:       b9 00 00 00 00          mov    ecx,0x0
    6c55:       ba 01 00 00 00          mov    edx,0x1
    6c5a:       be 01 00 00 00          mov    esi,0x1
    6c5f:       48 c7 c7 00 00 00 00    mov    rdi,0x0
    6c66:       e8 00 00 00 00          call   6c6b
<_stp_ctl_work_callback+0x8a>
    6c6b:       eb c6                   jmp    6c33
<_stp_ctl_work_callback+0x52>
```

And _stp_ctl_work_callback+0x8a is the instruction at the address 6c6b above.

-- 
You are receiving this mail because:
You are the assignee for the bug.


More information about the Systemtap mailing list