Netplex-wide mutexes
These are mutexes that work in both multi-processing and multi-threading netplex environments. They are implemented on top of [root:Netplex_semaphore], and use the same name space for named objects.
For implementation issues, see [root:Netplex_semaphore].
It is not required to create mutexes: A mutex is automatically created when the name is used for the first time. Mutexes are initially always in unlocked state.
Mutexes do not stack: A container cannot lock a mutex several times. (The second lock attempt would result in a deadlock.)
There is no deadlock detection.
If a container holding mutexes terminates the mutexes are automatically unlocked.
Mutex are not owned by their container. It is allowed that a different container unlocks the mutex.
Technically, a mutex named n is implemented by a semaphore n.
If the mutex is locked, the semaphore has the value 0, and if it is
unlocked, it has value 1.
Thread safety: Full. The functions can be called from any thread.
To enable mutexes, call the controller's add_plugin method
with this object as argument. This can e.g. be done in the
post_add_hook of the processor.
The following functions can only be invoked in container contexts. Outside of such a context the exception Netplex_cenv.Not_in_container_thread is raised.