When more than one AES/SHA requests are sent from different client processors to PLM, it serves the first request and queues the latter one. Serving the latter request is done in the form of task-based event notification functionality. When core (AES/SHA) is busy, PLM yields the task to be executed and notifies it in the form of an event. PLM checks this event whenever the core becomes free and triggers the corresponding task (adds to the task queue). The added task gets dispatched by the task dispatcher in serial mode.
If the core is busy for a longer time i.e. if the client using the core fails to send next command within 2 seconds, PLM frees the core forcefully and the next request notified in the queue is triggered. Scheduler takes care of this 2seconds timeout. If the timeout is reached, it adds the free resource task to the task queue which frees the core (AES/SHA).
Requests from clients for using core are served on first-come first-serve basis.
If partial PDI comes when the core is busy, partial PDI event is also notified, and it gets served when both AES and SHA cores are free.
This feature is disabled in Versal by default and can be enabled through Board Support Package settings. For Versal Net and Telluride, this feature is enabled by default.