Question

1. Briefly explain the differences between a semaphore and a condition variable, and what problem in common these two synchronization primitives solve.

Answer 1

• Semaphores and condition variables both try to solve synchronisation problems. There are situations when we want a resource to be accessible by only one process or some processes or by processes in a certain order. These concepts solve the problems that may arise.
• Condition variable provides a queue for processes which are waiting for some resource to access or acquire. When a process wants to access some resource, it tests a condition, if it is true, it uses it, otherwise it adds itself to the queue and waits for its turn. When a process finishes its turn, it wakes up exactly one process waiting in the queue for the resource. If the queue is empty, it does nothing. There is also a broadcast signal mechanism which wakes up all the processes waiting in case of a shareable resource. Only three operations are defined in case of a condition variable: wait, signal and broadcast signal.
• A semaphore is an integer variable which can be initialized to a certain value. But, this value can be changed by only two atomic operations P(x) and W(x). It may provide service like controlled access to the critical section or some resource. P(x) operation checks the value of x, if x>=0, then it decrements x by one and starts executing the critical section, otherwise it becomes blocked. When a process has finished executing the critical section or is done with the resource it executes W(x) which increments the value of x by one. This increment in x will unblock one process which has been blocked by the semaphore. Semaphore can be counting semaphore where it can take more than 2 values, or binary semaphore where it can take only 2 values: 0 and 1.