/*****************************************************************************
* SOURCE: This is a translation of a Pthread program from the Lawrence Livermore
* Computing Center POSIX Threads Programming Exercise at:
* https://computing.llnl.gov/tutorials/pthreads/exercise.html
* FILE: bug6.cvl
* DESCRIPTION:
*   This example demonstrates a race condition with a global variable that
*   gives obviously wrong results. See bug6fix.c for one solution.
*   The dotprod_mutex.c example provides a much more efficient way of
*    solving this problem than bug6fix.c (FYI).
* Command line execution:
* civl verify -inputNUMTHRDS=5 -inputVECLEN=10 bug6.cvl
******************************************************************************/

#include "pthread.cvh"
#include <civlc.h>
#include <stdio.h>
#include <stdlib.h>

/* Define global data where everyone can see them */
$input int NUMTHRDS;
$input int VECLEN;
int *a, *b; 
long sum=0;

void *dotprod(void *arg)
{
  /* Each thread works on a different set of data.
  * The offset is specified by the arg parameter. The size of
  * the data for each thread is indicated by VECLEN.
  */
  int i, start, end, offset, len;
  long tid = (long)*arg; // Dereference rather than direct conv
  offset = tid;
  len = VECLEN;
  start = offset*len;
  end   = start + len;

  /* Perform my section of the dot product */
  printf("thread: %d starting. start=%d end=%d\n",tid,start,end-1); // Removed l from %ld, unsupported
  for (i=start; i<end ; i++) 
    sum += (a[i] * b[i]);
  printf("thread: %d done. Global sum now is=%d\n",tid,sum); // Removed l from %ld, unsupported
  pthread_exit(NULL, false, NULL, 0); //Different parameters
}

int main (void)
{
  long i;
  long j[NUMTHRDS];
  void *status;
  pthread_t threads[NUMTHRDS];
  pthread_attr_t attr;

  /* Assign storage and initialize values */
  a = (int*) malloc (NUMTHRDS*VECLEN*sizeof(int));
  b = (int*) malloc (NUMTHRDS*VECLEN*sizeof(int));
  
  for (i=0; i<VECLEN*NUMTHRDS; i++)
    a[i]= b[i]=1;

  /* Create threads as joinable, each of which will execute the dot product
   * routine. Their offset into the global vectors is specified by passing
   * the "i" argument in pthread_create().
   */
  pthread_attr_init(&attr);
  pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_JOINABLE);
  for(i=0; i<NUMTHRDS; i++){
    //Add values into array and use address so that each value is unique as int to void * conversion is not supported
    j[i] = (long)i;
    pthread_create(&threads[i], &attr, dotprod, (void *)&j[i]); 
  }
  pthread_attr_destroy(&attr);
  
  /* Wait on the threads for final result */
  for(i=0; i<NUMTHRDS; i++) 
    pthread_join(threads[i], &status);
  
  /* After joining, print out the results and cleanup */
  printf ("Final Global Sum=%d\n",sum);
  free (a);
  free (b);
  return 0;
}