[Ns-bugs] [Bug 793] Many still-reachable memory leaks detected in the system
code at nsnam.ece.gatech.edu
Fri Jan 15 15:52:42 PST 2010
--- Comment #1 from Craig Dowell <craigdo at ee.washington.edu> 2010-01-15 18:52:42 EDT ---
I picked on simple-point-to-point-olsr and took a look at what is happening
with these errors.
It seems that Simulator::Destroy is called at the end of the example main ()
and the SimulatorImpl is destroyed and pimpl is zeroed.
After Simulator::Destroy returns, the locals in main() are torn down.
One of these locals is a node container. The container starts tearing its
stuff down. Eventually it works its way to something with a routing protocol
and drops the reference count. When the routing protocol reference count drops
to zero, it is deleted. This means that the Timer objects in the olsr routing
protocol are deleted.
In ~Timer, a check is made for the CANCEL_ON_DESTROY bit, which is true. It
then calls m_event.Cancel () which means a call to Simulator::GetImpl (), where
pimpl is found to be zero and a new simulator impl is created.
Trying to figure out what happened between ns-3.6 and now to trigger this ...
Configure bugmail: http://www.nsnam.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.
More information about the Ns-bugs