Assume that the compiler performs no optimization and that all code runs on the

Question

Assume that the compiler performs no optimization and that all code runs on the processor exactly as written. Also assume that no library calls will fail. #include 0001: 0002: #include 0003: #de fine NUM THREADS 4 0004: char array [NUM THREADS+21: 0005: int pos =0; 0006: char outs [9] # 0011 : 0012: 0013: 0014 char outs[index]: 0015 : 0016: return NULL 0017: 0018: int main() 0019: void * work (void* id) int index (* ( (int*) id)) *2; = array [pos++] = c; 0020: char three'z 0021: int i; 0022: pthread t tid [NUM THREADS] 0023 : for (i 0; í

Explanation / Answer

This program ideally should output length of the “array” i.e. when 4 threads modify the array they add 4 elements and ‘z’ is added by main function. So this should output 5 as length of array would be 5.

Race Condition: Since array[] is modified concurrently by these thread, they insert elements using index represented by ‘pos’ which is a shared variable. Line 15 will be broken down in machine language into many instructions s :

If a thread gets interrupted by another thread in between these instructions, say after assigning c to array, then pos will not be incremented and next thread might also end up inserting element at pos location. In that case instead of 2 only 1 element gets inserted in the array and length of array reported in main will be lesser than expected.

Solution: Use mutex. All the code inside function work() should be wrapped inside mutex as below:

pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;

void * work(void* id){

            pthread_mutex_lock(&mutex);

            int index = (*((int*)id))*2;

            char c = outs[index];

            array[pos++] = c;

            return NULL;

            pthread_mutex_unlock(&mutex);

}

Related Questions

Navigate

• Browse (All)
• Browse A
• Subjects

---

---

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.