To make it possible to implement semaphores on a computer with multiple CPUs that share a common memory, computer architects often provide a Test and Set Lock instruction. TSL X tests the location X. If the contents are zero, they are set to 1 in a single, indivisible memory cycle, and the next instruction is skipped. If it is nonzero, the TSL acts like a no-op. Using TSL it is possible to write procedures lock and unlock with the following properties. lock (x) checks to see if x is locked. If not, it locks x and returns control. If x is already locked, it just waits until it becomes unlocked, then it locks x and returns control. unlock releases an existing lock. If all processes lock the semaphore table before using it, only one process at a time can fiddle with the variables and pointers, thus preventing races. Write lock and unlock in assembly language. (Make any reasonable assumptions you need.)
