[Openswan Users] _plutorun/_plutoload stops after shell logout

[netvipe]

Paul Wouters wrote:
> On Tue, 10 May 2005, netvipe wrote:
> 
>>>> my problem starts after i've left the shell (from where i've started
>>>> the
>>>> ipsec init script). when i logout, the ike process
> 
> 
>>>> some info's about the system:
>>>> - x86 (pentium IIT) hardware with 2.4.29-grsec kernel
>>>> - os resides on a flash module and is based on uClibc
>>>> - /etc, /var and /tmp are writeable (tmpfs)
>>>> - busybox provides most services (incl. init)
>>>> - sh replaced by bash (3.00.16(2))
> 
> 
>> i followed your hint and updated to openswan 2.3.1 but unfortunatly
>> nothing changed.
> 
> 
>> ...but there is a new entry in the logfile:
>>
>> May 10 15:54:35 pluto[16652]: Pluto ignores SIGHUP -- perhaps you want
>> "whack --listen"
>> May 10 15:54:35 pluto[16652]: ADNS process terminated by signal 1
>> May 10 15:54:35 pluto[16652]: closing helper(0) pid=17207 fd=6 exit=0
>> May 10 15:54:35 pluto[16652]: reapchild failed with errno=10 No child
>> processes
>>
>>
>> not sure if sshd is the fault. i would also tend to some "weird shell
>> issue".
> 
> 
> Does your unit have a serial console? Can you try and start it from the
> serial console, and then quit the console and see what happens? Then we can
> at least pinpoint it to openssh's sshd with openswan.
> 
> I opened this as http://bugs.xelerance.com/view.php?id=297
> 
> Paul


i've tried to start the ipsec process at the serial console.
unfortunatly the same event occurs:

May 11 14:08:12 pluto[12954]: Pluto ignores SIGHUP -- perhaps you want
"whack --listen"
May 11 14:08:12 init: ^MProcess '/sbin/getty -L tts/0 115200 vt100' (pid
4045) exited.  Scheduling it for restart.
May 11 14:08:12 pluto[12954]: ADNS process terminated by signal 1
May 11 14:08:12 init: ^MStarting pid 12295, console /dev/tts/0:
'/sbin/getty'
May 11 14:08:12 pluto[12954]: closing helper(0) pid=16939 fd=6 exit=0
May 11 14:08:12 pluto[12954]: reapchild failed with errno=10 No child
processes


but i thought about your "weird shell issue" suggestion and replaced the
bash by an other shell.
after a logout and a session reinitialization i was really surprised...
all the ipsec processes were still running!!!

so i started to trace the problem again as i did before.
finally "strace -p -f" showed the problem.

it seems to be the combination of a busybox init and the bash shell.
init (busybox) free's all ressources after a logout.
even "daemonized" scripts will be stopped/killed.
right now i'm not sure why all ressources will be freed.
maybe the kernel (2.4.29-grsec) is the problem...???

any other process (elf-binary) is running without problems.

nevertheless thanks for your help !!!
specially for the suggestions.

i'll provide the definite reason to you, if i find out.

kindly regards


Florian

-- 
Florian Reinholz                CCSP, INFOSEC Professional
mailto:me at netvipe.com           http://www.netvipe.com
Tel: +49(1801)555777-0840

GnuPG Key:   http://netvipe.com/netvipe-pubkey.asc
Fingerprint: C808 9B2E 002F FE88 A916 0CFF 128A 6EC0 5C97 DE78