Real Time Operating Systems Questions and Answers

Answer: A

Correct answer: ten times as many tick interrupts and their overhead

Each tick is an interrupt with a context save, a list walk and a possible switch, so ten times the rate is ten times that cost - a real fraction of the processor on a small part. Determinism is unaffected and the memory unchanged; what is spent is processor time, which is why the tick rate is chosen rather than maximised.

Answer: B

Waking a hundred or a thousand times a second to discover there is nothing to do is ruinous for a battery, so the kernel programmes a wake-up for the next event that is actually due and sleeps until then. It is the single most valuable power feature a kernel offers.

Answer: C

A task suspended half way down a call chain must find that chain intact when it resumes, which a shared stack could not guarantee. Sizing each one is a real design problem, since too little corrupts memory and too much wastes it.

Answer: D

Correct answer: its deepest call chain plus the worst-case interrupt frame

The stack must hold every frame on the deepest path through the task’s call tree, together with whatever an interrupt pushes if one arrives at the worst moment - and nested interrupts if they are permitted. Overflow is silent and corrupts whatever lies beyond, which is why a guard region or a stack-checking kernel is worth having.

Answer: A

Filling the stack with a sentinel value and having the kernel look at the last words on every switch catches the overrun soon after it happens. A memory protection unit does better still, faulting at the moment of the overrun rather than afterwards.

Answer: B

A heap allocation that fails halfway through initialisation leaves the system in a state nobody designed, and proving it cannot happen is far easier if there is no heap. Certification standards frequently forbid dynamic allocation after start-up outright.

Answer: C

Free memory broken into many small pieces cannot satisfy a request for one large block, and the failure appears after weeks of running rather than in test. Fixed-size block pools avoid it entirely, which is why they are common in embedded work.

Answer: D

Time spent in a handler is time every other source waits, so the handler does the urgent minimum and signals a task to do the rest. The pattern is variously called deferred interrupt processing, a bottom half or a deferred procedure call.

Answer: A

Calling the ordinary blocking version from a handler would try to block the interrupt context, which has no task to block, so kernels provide non-blocking variants and name them plainly. Mixing them up is a common and hard-to-diagnose fault.

Answer: B

Correct answer: there is no task context to place in the blocked state

Blocking means recording that a task is waiting and scheduling another, and a handler is not a task - it has no control block and no state to save. The kernel therefore provides from-interrupt variants that never block, and calling the ordinary ones from a handler faults or hangs.