A spin lock (see Bacon [2002]) is a boolean variable accessed via an atomic test-and-set instruction, which is used to obtain mutual exclusion. Would you use a spin lock to obtain mutual exclusion between threads on a single-processor computer?

What will be an ideal response?

The problem that might arise is the situation in which a thread spinning on a lock uses up its timeslice, when meanwhile the thread that is about to free the lock lies idle on the READY queue. We can try to avoid this problem by integrating lock management with the scheduling mechanism, but it is doubtful whether this would have any advantages over a mutual exclusion mechanism without busy-waiting.