openmp并行中的函数调用会导致分段错误

Question

我里面有一个for循环，还有另外两个for循环，我不太渴望并行化。但是在外部for循环内部有一个函数调用，我想称之为并行。但是，我遇到了段错误。下面是代码片段，我正在尝试并行化。

        #pragma omp parallel for private(j,k) 
        for (k=0; k < numatoms; k++) { // Loops over all atoms (k) 
        # I dont case about parallelizing this loop, tried both with and without critical section still segfault. 
        #pragma omp critical
        {
        for (j=0; j < numatoms; j++) { // Loop over all atoms
            all_nbrs_index[j] = j;  // Get the indeces of neighbours in nbrs_ind struct
            all_nbrs_values[j] = dist_mat[k][j];// Get the distances of neighbours in nbrs_ind struct
            }
            }
            #I want this function call to be executed in parallel 
            QuickSort(all_nbrs_values, all_nbrs_index, 0, numatoms-1);
        //              if (k ==0) {
        //                      printf("\nThe sorted neighbour indices are. \n");
        //              }                       
            /// From sorted distances and indices get get the 4 closest.
            #This loop also need not to be parallelized but if possible its a bonus    
            #pragma omp critical
            {
            for(j=0; j <=4 ;j++) {
                    nbrs_ind[k][j] = all_nbrs_index[j];
          //                      printf("%d\t ", all_nbrs_index[i] );                                    
            }
            }
            //              printf("\n");
            }

我认为这与私有化有关。我的错误是由于没有在内部循环中私有化那些数组而引起的。如果是这种情况，如何将数组设为私有？任何帮助将非常有用。