CODEWITHSUNDEEP
cfor-looppthreads
John
John

Reputation: 71

Threads in for loops

I'm writing a program in C to calculate the access time to certain websites. The sitenames are stored in each element of urls array. If I take out the for (y = 0; y < iterations; y++) loop, then everything runs fine. But if if I keep it. urls[0], the first website, gets messed up after the second for loop completely finishes and increments y

What's causing this?

char *urls[50]; char str1[20];

void *wget(void *argument)
{
  int threadid;
  threadid = *((int *)argument);
  strcpy(str1, "wget -q --spider ");
  strcat(str1, urls[threadid]);
  system(str1);
}


for (y = 0; y < iterations; y++)
{
  for (j = 0; j < numthreads; j++)
  {
        thread_args[j] = j;

        clock_gettime(CLOCK_REALTIME, &bgn);
        rc = pthread_create(&threads[j], NULL, wget, (void *) &thread_args[j]);
        rc = pthread_join(threads[j], NULL);
        clock_gettime(CLOCK_REALTIME, &nd);
        times[j] = timediff(bgn,nd);
  }
}

Upvotes: 1

Views: 182

Answers (2)

Variable Length Coder
Variable Length Coder

Reputation: 8116

My bet is that one of the strings in urls + the wget string are longer than 20 bytes and are overwriting that data. Make str1 larger, and move it into your wget function (multiple threads should not be writing to one shared resource without locking).

Upvotes: 3

QuantumMechanic
QuantumMechanic

Reputation: 13946

Some possibilities...

str1 appears to be shared among all the threads. That's a recipe for trouble right there.

str1 is only 20 chars long. Hard to believe the whole wget command line including the URL will be less than 20 chars. So you're writing off the end of str1.

Consider making str1 a local variable in wget(), and either make it a char array big enough to handle the largest possible wget command line you might have, or dynamically allocate it and free it within wget() with a size based on the length of the constant part of the command line and the current URL.

Upvotes: 5

Related Questions