[omniORB] bug fixes in 4.1.0

[Wernke zur Borg]

> 
> We are experiencing a strange bug, servant binary crashing 
> (in generated 
> stub file) when accepting incoming connection, last debug from ORB 
> (traceLevel set to 40):
> 
> omniORB: Dispatching remote call '_is_a' to: 
> key<fsaClientAgent> (active)
> 
> Note that one connect to the same servant (cca 40 seconds 
> before) was ok.
> 
> What could be happening here? Would deleting servant 
> (explicitly, not by 
> removing local reference) result in this behavior (servants are not 
> being deleted, but there could be a problem in our code).
> 

Deleting a servant will almost certainly result in a segfault during the
next upcall. So check your code first. I always found it useful to
insert debug printouts in servant destructors, also to prove that they
are properly deleted when it is time to do so.

Wernke