Real Time Operating Systems Questions and Answers

Answer: C

Every stage adds: getting into the handler, doing its work, the kernel deciding the task is now the highest priority ready one, and the context switch. A well designed kernel makes that final switch happen on the handler's return rather than at the next tick.

Answer: D

A handful of things to do in turn needs no kernel, and the loop is smaller, cheaper to certify and easier to reason about. The moment the periods and deadlines diverge, hand-scheduling them in a loop becomes the harder problem.

Answer: A

Writing a task that simply waits for what it needs is enormously easier than breaking the same logic into a state machine that must never block. The kernel costs processor time rather than saving it, and shared data needs more protection rather than less.

Answer: B

The kernel provides the mechanisms; whether the deadlines are met depends on the tasks' execution times and priorities, which are the designer's business. A kernel makes a schedulability analysis possible, and does not perform one.

Answer: C

Splitting the offender so the loop comes round often enough restores the timing without restructuring anything else. Introducing a kernel would work but is a far larger change, and moving the deadline is a change to the requirement rather than to the design.

Answer: D

The whole value of the standard priority model is that a task set's schedulability can be decided by calculation rather than by testing, which a hand-written loop cannot offer. That analysability is exactly why the analysis becomes easier.

Answer: A

Bounded kernel operations are necessary but nowhere near sufficient: an unbounded loop in a task or an unprotected shared resource destroys determinism whatever the kernel does. The reason is a true statement about the kernel and not an explanation of a system property.

Answer: B

Correct answer: 1-R, 2-P, 3-Q

Three distinct measurements. Latency is the delay from an event to the start of its handling. Jitter is the variation in that timing from one occurrence to the next, and for a periodic task it is the spread in release times. And a context switch is the saving and restoring that a change of task requires.

Answer: C

Correct answer: 1-Q, 2-R, 3-P

Each state ends in its own way. A ready task waits only for the scheduler to select it. A blocked task waits for the event it asked for, or for its timeout to expire. And a suspended task waits for another task to resume it explicitly - no event or timeout will do it, which is what distinguishes suspension from blocking.

Answer: D

Correct answer: 1 and 3 only are correct

Statements 1 and 3 are correct: each task has its own stack, since each has its own call chain, and tickless operation lets the processor sleep through idle periods instead of waking at every tick. Statement 2 is exactly what a handler may not do, there being no task context to block.