-Training Presentation by Rohit Joshi

 Real-Time Systems.
 A few Concepts.
 Real-Time OS.
 Case Study Nucleus – Introduction will cover
in detail in next presentation.

Decreasing order
of Familiarity
Applications
OS
Hardware
System Mostly RTOS

 A system where correctness of computation
depends on logical correctness and the time
upon which the results are produced.
 Guaranteeing timing means predictability.
 A high degree of utilization while timing
constraints are met.
 A bounded latency rather than high average
performance.

 Hard Real-Time Systems
 If timing constraints are not met causes a
catastrophic system failure.
 Soft Real-Time Systems
 If timing constraints are not met causes a
degradation in the system performance.

 Multiple “Task” are scheduled to run.
 Each task maintains its Own state
 CPU Register
 Execution Stack
 Not always beneficial.

 With more than one task, some means is required to
choose among them
 FIFO Scheduling
 Each task is run in sequence until it blocks or is
finished
 (cooperative multitasking)
 Shortest Job First
 The task requiring the least CPU time goes first
 Round Robin
 Scheduled tasks are periodically interrupted and another
task is given the opportunity to execute
 None of these algorithms are suitable for an
embedded system!

 Task may have an inherent precedence.
 Precedence must be observed when allocating
CPU.
 Highest priority task ready to run (not pended or
delayed) is allocated to the CPU
 Reschedule can occur anytime, due to:
 Kernel calls.
 System clock tick

t3 completes
t3 preempts t2
t2 completes
Task t3
t2 preempts t1
Task t2 Task t2
Task t1 Task t1
TIME

t4 completes
t4 preempts t2
Task t4
t1 t2 t3 t1 t2 t2 t3
TIME

 Ability to provide the required level of service
with a bounded response time.
 Multithreaded and preemptible.
 Static priorities.
 Predictable thread synchronization
mechanism.
 Priority inheritance.

 Documented OS behavior
 Interrupt latency
 Interrupt masking times
 System service latency
 System interrupt levels
 Device IRQ levels

 Task Switches
 Number of tasks shouldn’t effect switching time.
 IPC and Synchronization
 Message sizes should not influence time.
 Memory Allocation.
 Fragmentation should be avoided.
 Can’t rely on garbage collectors.

 Multitasking real-time kernel
 Priority, pre-emptive scheduler
 Inter-task Communication
(mailboxes,queues, pipes)
 Inter-task synchronization (semaphores,
signals, events)
 Dynamic creation and deletion of all objects

 Proven
 Complete product line
 No royalties
 Source code
 Vast CPU support