When the SERIALIZE keyword is specified in a concurrent-thread module, only one thread can run in the procedure at any time. If one thread is running in the procedure and another thread calls the procedure, the second thread will wait to run the procedure until the first thread is no longer running in the procedure. If a thread is running in the procedure and it makes a recursive call to the procedure, then it must return from all the recursive calls to the procedure before another thread can begin running in the procedure.
The SERIALIZE keyword is allowed only when THREAD(*CONCURRENT) is specified on the Control specification.
Specifying SERIALIZE for one procedure is similar to specifying THREAD(*SERIALIZE) on the control specification. The difference is that specifying THREAD(*SERIALIZE) on the Control specification limits access by multiple threads to all the procedures in the module, while specifying the SERIALIZE keyword for a procedure only limits access to that procedure.
If you have more than one procedure in a module with the SERIALIZE keyword, the procedures are independent. One thread can be running in one serialized procedure, while another thread is running in another serialized procedure in the same module. For example, if procedures PROCA and PROCB in the same module both have the SERIALIZE keyword, one thread could be running PROCA while another thread was running PROCB. For more information on using serialized procedures, see THREAD(*CONCURRENT | *SERIALIZE).