Real Time Operating Systems Questions and Answers

Answer: A

Correct answer: 1 and 2 only are correct

Statements 1 and 2 give the two kinds of deadline correctly. Statement 3 is the misconception the subject exists to correct: real time means predictable, not fast. A system meeting a one-second deadline every time is real time; one usually completing in a microsecond and occasionally taking ten milliseconds is not.

Answer: B

Correct answer: a compiler for the target processor

A kernel supplies scheduling, task management, timing and inter-task communication - things that exist while the program runs. The compiler is a development tool used before any of that, and it is supplied separately whatever kernel is chosen.

Answer: C

The kernel's own certification is a body of evidence the application's assessor can build on, which saves a great deal of work. The application still has to be certified on its own account, and the safety manual states the conditions under which the evidence applies.

Answer: D

All three run in tens of kilobytes and provide priority preemptive scheduling with the usual communication objects. The general purpose alternatives with real time patches, such as Linux with PREEMPT_RT, need megabytes and a memory management unit.

Answer: A

A full kernel with drivers, a network stack and a filesystem brings enormous capability and latencies measured in tens of microseconds rather than single microseconds. Which matters depends entirely on the deadline in question.

Answer: B

Splitting the concerns lets each half use the software it suits: a rich kernel where capability matters and a bare core or a small kernel where the deadline does. Dual-core parts with exactly this division are now common.

Answer: C

The scheduler must always have something to dispatch, so a task at the lowest priority runs when nothing else can. It is the natural place to put a sleep instruction and, with a hook, to measure how much processor time is spare.

Answer: D

The fraction of wall-clock time the lowest priority task receives is the fraction the application is not using, which is a very cheap load measurement. Putting a sleep instruction in it complicates the arithmetic, since the sleeping time must still be counted.

Answer: A

Those three choices decide the kernel's timing granularity, its memory footprint and how finely priorities can be separated, and all three are compile-time decisions. Clock, pins and compiler settings belong elsewhere in the build.

Answer: B

Each level needs a list head and, in some kernels, a bitmap entry, so a hundred and twenty-eight levels cost real memory on a small part. A well written scheduler finds the highest ready priority in constant time whatever the number of levels.