2.4.19 kernel kills application
 

Hello Linux experts

I was running linux kernel 2.4.14 uptill now. I had an application ( a logger ) that sets up some shared memory initially when it
starts up. Then it waits for someone to send it a message and
then logs that message. This was working fine in 2.4.14.
(I was starting it from the syslog script ).

When I tried the same app under 2.4.19, the application starts
ok, but is killed by something in a very short while. However if
I start it again later, it seems to keep running. There is no
core file or any message in the /var/log/messages.

I have put debug messages in the application and everything is
working fine. So the app is getting killed by the kernel.

Any ideas why this might be happening or how to debug it ?

Thanks in advance
Goody

Did you write the app yourself?

If it's a commercial product, it might not be supported with that kernel.

Yes. I wrote the application myself. I have been running it for
a few months without any problems on 2.4.14 kernel.
Now I switched to 2.4.19 kernel and the darn thing is not
working the way it used to.

Any help will be greatly appreciated.

Goody

Have you recompiled it for the new kernel?
Which libraries does it use?
What are the error messages when you run it from a command line?
(/var/log/messages are kernel messages only).

No, I haven't recompiled it for the new kernel. The development machine I have is a Redhat 7.3 ( with 2.4.18-3 ) kernel on it.
The target machine is with 2.4.19 kernel. I don't have any development tools on the target.

The only libraries that this is using is
libc.so.6 and
ld-linux.so.2

When I run it from command line, it gives no error, it runs fine.
When I start it from syslog script also it gives no error, but the
process gets killed after it starts up ok.

I can start it up again on the command line and it will run fine.
I put a bunch of debug statements in my application code and
all of it is working fine. The application does not seem to have
any errors.

Thanks
Goody

A very long way to debug it is to try it with ebery kernel release from 2.4.14 to 2.4.19 and see at which it release it starts getting killed, then check the kernel changelog for the release of that kernel which kills it, and you might come across some clues in new kernel code that might be causing your app to die.
I know this is a very time-consuming and hard thing to do, but its the only one that comes to mind now..
-NSKL

Is it really the difference in kernel version? You said you where running it on a different machine is everything on the other machine identical to the one where it works, with exception of the kernel version?
How exactly is it started from syslog? And how are the scripts setup which run syslog? In my rotatelogs script I restart syslog to make sure it starts writing to the new log files instead of continueing with the old ones. It could be some script similar to this which is causing syslog along with your app to stop.
Just not sure if it's right to blame the kernel if you say it does work properly after starting it up again later.

I will try with earlier kernels ( like 2.4.18 ) to see if this problem
occurs there as well.

Thanks