信号:MPI C ++中的分段错误(11) [英] Signal: Segmentation fault (11) in MPI C++

问题描述

我有一个代码，可以计算MPI中整数的平均值:

I have a code, which counts the average value of integers in MPI:

#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，我正在获取错误列表:

When i try to run this (mpirun -c 4 avg 4), i`m getting the error list:

[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] Signal: Segmentation fault (11)

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

[mangeke-mpi-2431940:03372] Signal code: Address not mapped (1)

[mangeke-mpi-2431940:03372]在以下地址失败:(nil)

[mangeke-mpi-2431940:03372] Failing at address: (nil)

[mangeke-mpi-2431940:03372] *错误消息结尾*

[mangeke-mpi-2431940:03372] * End of error message *

我如何解决此问题?

推荐答案

正如Hristo所说，nums被初始化为NULL.如果您浏览由调试器生成的核心文件，它将引发以下语句

As Hristo comments, the nums is initialized to NULL. If you explore the core file generated with the debugger, it raises the following statement

核心由`./m 4'生成.程序以信号SIGSEGV终止， 分段故障. #0 0x0000000000408809在m.cxx:36 36 nums [i] = i;中的主(argc = 2，argv = 0x7ffd4fc87e68)

Core was generated by `./m 4'. Program terminated with signal SIGSEGV, Segmentation fault. #0 0x0000000000408809 in main (argc=2, argv=0x7ffd4fc87e68) at m.cxx:36 36 nums[i] = i;

如果您按如下所示更改以下代码，则可以使其运行而不会出现段错误.

if you change the following code as shown below you'll get to make it run without segfaulting.

    ....

    // 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);
    }

这篇关于信号:MPI C ++中的分段错误(11)的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！