How Data locality of locality level PROCESS_LOCAL works in Spark with dynamic allocation enabled

In case of dynamic allocation executors are removed after a certain idle time so there could be time when there are zero executor running. So when a task is submitted at time when there is no executor running will there by any task with data locality as PROCESS_LOCAL? Because PROCESS_LOCAL means data is located inside a JVM, and currently no executors are UP.