The smbd_scavenger crashes with a core dump if its parent dies first

[Richard Sharpe]

On Thu, Aug 30, 2018 at 10:07 AM, Ralph Böhme <slow at samba.org> wrote:
> On Thu, Aug 30, 2018 at 10:00:50AM -0700, Richard Sharpe wrote:
>>
>> (gdb) where
>> #0  0x00007f42b9986277 in raise () from /lib64/libc.so.6
>> #1  0x00007f42b9987968 in abort () from /lib64/libc.so.6
>> #2  0x00007f42bb32d5ae in dump_core () at ../source3/lib/dumpcore.c:338
>> #3  0x00007f42bb31ddb7 in smb_panic_s3 (why=<optimized out>)
>>    at ../source3/lib/util.c:827
>> #4  0x00007f42bd40a02f in smb_panic (
>>    why=why at entry=0x7f42bd126e34 "smbd_scavenger_parent_dead")
>>    at ../lib/util/fault.c:166
>> #5  0x00007f42bcff23e1 in exit_server_common
>> (how=how at entry=SERVER_EXIT_ABNORMAL,
>>    reason=0x7f42bd126e34 "smbd_scavenger_parent_dead")
>>    at ../source3/smbd/server_exit.c:238
>> #6  0x00007f42bcff26c1 in smbd_exit_server (explanation=<optimized out>)
>>    at ../source3/smbd/server_exit.c:256
>> #7  0x00007f42ba348b92 in exit_server (
>>    reason=reason at entry=0x7f42bd126e34 "smbd_scavenger_parent_dead")
>>    at ../source3/lib/smbd_shim.c:123
>> #8  0x00007f42bcff497e in smbd_scavenger_parent_dead (event_ctx=<optimized
>> out>,
>>    fde=<optimized out>, flags=<optimized out>, private_data=<optimized
>> out>)
>>    at ../source3/smbd/scavenger.c:103
>
>
> guess we may want to call exit_server_cleanly() instead of exit_server() in
> smbd_scavenger_parent_dead().

Added samba-technical back because I probably dropped it somewhere.

Well, that could be done ... but it would possibly obscure those cases
where the parent was killed without us knowing.

That will certainly fix my current problem. We could add a DEBUG(0,...
or equivalent.

-- 
Regards,
Richard Sharpe
(何以解憂？唯有杜康。--曹操)(传说杜康是酒的发明者)