Although mapping Ada tasks onto the underlying threads has significant advantages, it does create some complications when it comes to respecting the scheduling semantics specified in the real-time annex (Annex D).
For instance the Annex D requirement for the
scheduling policy states:
`When the active priority of a ready task that is not running changes, or the setting of its base priority takes effect, the task is removed from the ready queue for its old active priority and is added at the tail of the ready queue for its new active priority, except in the case where the active priority is lowered due to the loss of inherited priority, in which case the task is added at the head of the ready queue for its new active priority.'
While most kernels do put tasks at the end of the priority queue when a task changes its priority, (which respects the main FIFO_Within_Priorities requirement), almost none keep a thread at the beginning of its priority queue when its priority drops from the loss of inherited priority.
As a result most vendors have provided incomplete Annex D implementations.
The GNAT run-time, has a nice cooperative solution to this problem which ensures that accurate FIFO_Within_Priorities semantics are respected.
The principle is as follows. When an Ada task T is about to start running, it checks whether some other Ada task R with the same priority as T has been suspended due to the loss of priority inheritance. If this is the case, T yields and is placed at the end of its priority queue. When R arrives at the front of the queue it executes.
Note that this simple scheme preserves the relative order of the tasks that were ready to execute in the priority queue where R has been placed at the end.