Question

我有一个代码，用于计算MPI中整数的平均值：

#include <stdio.h>
#include <stdlib.h>
#include <time.h>
#include <mpi.h>
#include <assert.h>

// Average method
int compute_avg(int *array, int num_elements) {
    int sum = 0;
    int i;
    for (i = 0; i < num_elements; i++) {
    sum += array[i];
    }
    return sum / num_elements;
}

int main(int argc, char** argv) {

    if (argc != 2) {
    fprintf(stderr, "Usage: avg num_elements_per_proc\n");
    exit(1);
    }
    int num_elements_per_proc = atoi(argv[1]);

    MPI_Init(NULL, NULL);
    int world_rank;
    MPI_Comm_rank(MPI_COMM_WORLD, &world_rank);
    int world_size;
    MPI_Comm_size(MPI_COMM_WORLD, &world_size);

// Create array with integers
    int *nums = NULL;
    if (world_rank == 0) {

        for (int i =0; i<5; i++){
            nums[i] = i;
        }
    }


// Subtable from any processes
    int *sub_nums = (int *)malloc(sizeof(int) * num_elements_per_proc);
    assert(sub_nums != NULL);


// distribution numbers for all processes
    MPI_Scatter(nums, num_elements_per_proc, MPI_INT, sub_nums,
    num_elements_per_proc, MPI_INT, 0, MPI_COMM_WORLD);

// Count avg subtable
    int sub_avg = compute_avg(sub_nums, num_elements_per_proc);

// Collectiong averages
    int *sub_avgs = NULL;
    if (world_rank == 0) {
    sub_avgs = (int *)malloc(sizeof(int) * world_size);
    assert(sub_avgs != NULL);
    }
    MPI_Gather(&sub_avg, 1, MPI_INT, sub_avgs, 1, MPI_INT, 0, MPI_COMM_WORLD);


// Calculates the overall average
    if (world_rank == 0) {
    int avg = compute_avg(sub_avgs, world_size);
    printf("Avg of all elements is %d\n", avg);
// Obliczenie średniej na danych oryginalnych i wyświetlenie.
    int original_data_avg =
    compute_avg(nums, num_elements_per_proc * world_size);
    printf("Avg computed across original data is %d\n", original_data_avg);
}

// free memory
    if (world_rank == 0) {
    free(nums);
    free(sub_avgs);
    }
    free(sub_nums);
    MPI_Barrier(MPI_COMM_WORLD);
    MPI_Finalize();

return 0;
}

当我尝试运行此（mpirun -c 4 avg 4）时，我收到了错误列表：

[mangeke-mpi-2431940：03372] *处理收到的信号*

[mangeke-mpi-2431940：03372]信号：分段错误（11）

[mangeke-mpi-2431940：03372]信号代码：未映射的地址（1）

[mangeke-mpi-2431940：03372]地址失败:(无）

[mangeke-mpi-2431940：03372] *错误消息结束*

我如何解决这个问题？

Answer 1

正如Hristo评论的那样，nums被初始化为NULL。如果您浏览使用调试器生成的核心文件，则会引发以下语句

核心是由./m 4＆＃39;生成的。程序以信号SIGSEGV终止，分段故障。＃0 0x0000000000408809 in main（argc = 2，argv = 0x7ffd4fc87e68）at m.cxx：36 36 nums [i] = i;

如果您更改下面的代码，如下所示，您可以让它在没有segfaulting的情况下运行。

    ....

    // Create array with integers
    int nums[num_elements_per_proc]; // <<-- change here
    if (world_rank == 0) {

        for (int i =0; i<5; i++){
            nums[i] = i;
        }
    }

    ....
    // free memory
    if (world_rank == 0) {
    // free(nums);                   // <<-- change here, free not needed
    free(sub_avgs);
    }

信号：MPI C ++中的分段错误（11）

1 个答案: