Real-Time Operating Systems (RTOS) 101

Real-Time System Characteristics RTOS Architecture Rate Monotonic

• A real-time system is a computer system which is required by
its specification to adhere to: Scheduling (RMS)
– functional requirements (behavior) • A priority is assigned based on the inverse of its pe-
– temporal requirements (timing constraints, deadlines) riod
• Specific deterministic timing (temporal ) requirements – Shorter execution periods = higher priority
– “Deterministic" timing means that RTOS services consume only – Longer execution periods = lower priority
known and expected amounts of time.
• Common way to assign fixed priorities
• Small size (footprint) – If there is a fixed-priority schedule that meets all dead-
lines, then RMS will produce a feasible schedule
Types of Real-Time Systems • Simple to understand and implement
• A generic real-time system requires that results be produced RTOS Task Services • P1 is assigned a higher priority than P2.
within a specified deadline period.
• An embedded system is a computing device that is part of a • Scheduling and Dispatching
larger system.
• Inter-task Communication
• A safety-critical system is a real-time system with catastro-
phic results in case of failure. • Memory System Management
• A hard real-time system guarantees that real-time tasks be • Input / Output System Management
Earliest Deadline First (EDF)
completed within their required deadlines. Failure to meet
a single deadline may lead to a critical catastrophic system • Time Management & Timers Scheduling
failure such as physical damage or loss of life.
• Error Management • Priorities are assigned according to deadlines:
• A firm real-time system tolerates a low occurrence of missing
a deadline. A few missed deadlines will not lead to total • Message Management – the earlier the deadline, the higher the priority
failure, but missing more than a few may lead to complete – the later the deadline, the lower the priority
and catastrophic system failure.
• Priorities are dynamically chosen
• A soft real-time system provides priority of real-time tasks
over non real-time tasks. Performance degradation is toler-
Task Control Block (TCB)
ated by failure to meet several deadline time constraints
with decreased service quality but no critical consequences.

Disciplines that Impact Priority Inversion

Real-Time Systems • Lower-priority task effectively blocks a higher-
• Real-time systems engineering is so multidisciplinary, it priority task
stands out as a highly specialized area.
• Lower-priority task’s ownership of lock prevents
higher-priority task from running
• Nasty: makes high-priority task runtime unpredict-
Controlling a Task able!

Priority Inheritance
• Solution to priority inversion
• Temporarily increase task’s priority when it acquires
a lock
• Level to increase: highest priority of any task that
might want to acquire same lock
– High enough to prevent it from being preempted
• dormant (idle): task has no need for computer time • Danger: Low-priority task acquires lock, gets high
• ready: task is ready to go active, but waiting for processor time priority and hogs the processor
• active (running): task is executing associated activities – So much for RMS
What is a RTOS? • waiting (blocked): task put on temporary hold to allow lower priority task • Basic rule: low-priority tasks should acquire high-
chance to execute priority locks only briefly!
• An RTOS is a preemptive multitasking operating system intended
for real-time applications. • suspended: task is waiting for resource
• It must support a scheduling method that guarantees re- VxWorks Architecture
sponse time
– Especially to critical tasks
Priority-Based
• Tasks must be able to be given a priority
– Static or dynamic
Preemptive Scheduling
• An RTOS has to support predictable task synchronization • Problem: Multiple tasks at the same priority level?
mechanisms
– Shared memory mutexes / semaphores, etc. • Solutions:
• A system of priority inheritance has to exist – Give each task a unique priority
• Manages hardware and software resources. – Time-slice tasks at the same priority
• Deterministic: guarantees task completion at a set deadline. • Extra context-switch overhead
– A system is deterministic if, for each possible state and each set of in- • No starvation dangers at that level
puts, a unique set of outputs and next state of the system can be de-
termined.
– Tasks at the same priority never preempt the other
• Behavior time constraints should be known and minimized • More efficient
– Interrupt latency (i.e., time from interrupt to task run)
– Minimal task-switching time (context switching) • Still meets deadlines if possible

NASA Independent
Richard E. Kowalski, [email protected], TASC Inc. Verification and
Validation Facility
NASA POC: Frank Huy, [email protected] Fairmont, West
Virginia