[AGENT++] Thread handle leak (SNMP++ 3.2.21a)

[Jochen Katz]

Hi,

> It turned out that the same piece of code causes a memory leak under
> Linux too. The reason is that taskDelete() is not called (in
> stop_poll_thread()) for the thread created by taskSpawn() (in
> start_poll_thread()).
> 
> Here's some info on taskSpawn/Delete -
> http://www.slac.stanford.edu/exp/glast/flight/docs/VxWorks/docs/vxworks/ref/taskLib.html#taskDelete

within stop_poll_thread() m_bThreadRunning flag is set to false, and the
poll thread will stop itself. Is it then legal and a "must do" to call
taskDelete() for the already finished thread? The taskSpawn() doc does
not state that taskDelete() has to be called in any case.

Regards,
  Jochen