递归中的内存泄漏

Question

我正在尝试使用递归来实现一种算法。 在递归中，我使用 new 分配内存而不是删除它，但我仍然遇到内存泄漏。我试图理解我做错了什么，但无法弄清楚。 有人可以看看吗？

• 我知道我可以使用向量，但想了解我做错了什么以及如何解决它。

这是代码：

#include <iostream>
#include <fstream>
#include <math.h>
#define _CRTDBG_MAP_ALLOC
#include <stdlib.h>
#include <crtdbg.h>
using namespace std;



int sortInv(int*,int);
int Sort_And_count_split_Inv(int*,int,int,int);
int Sort_And_count(int *,int,int);

int main()
{
    _CrtSetDbgFlag ( _CRTDBG_ALLOC_MEM_DF | _CRTDBG_LEAK_CHECK_DF );
    int Siz = 9;
    int Arr[9] = {66,3,11,76,93,9,22,56,89};    
    int b = Sort_And_count(Arr,0,Siz-1);
    for (int i=0; i<Siz; i++)
        Arr[i] = Arr[i];
    return 0;
}

int Sort_And_count(int *a,int low,int high)
{
    int mid;
    int n = 0;
    if (low >= high)
        return 0;
    else
        mid = (high+low)/2;
        n+= Sort_And_count(a,low, mid);
        n+= Sort_And_count(a, mid+1, high);
        n+= Sort_And_count_split_Inv(a,low,mid,high);
        return n;
}

int Sort_And_count_split_Inv(int* a, int low, int mid, int high)
{
    int i,j,k;
    i=low;
    j=mid+1;
    k=low;
    int count = 0;
    int* tmp = new int[high-low+1];
    while (i<=mid && j<=high)
    {
        if (a[i]<a[j])
        {
            tmp[k] = a[i];
            i++;
        }
        else
        {
            tmp[k] = a[j];
            j++;
            count += mid-i == 0? 1: mid+1-i;
        }
        k++;
    }
    while (i<=mid)
        tmp[k++] = a[i++];
    while (j<=high)
        tmp[k++] = a[j++];
    for (i=low; i<=high; i++)
        a[i] = tmp[i];
    delete[] tmp;
    _CrtDumpMemoryLeaks;
    return count;
}

Answer 1

编译您的代码并在valgrind 中运行它会导致内存泄漏的第一次出现：

==20797== Invalid write of size 4
==20797==    at 0x400BA0: Sort_And_count_split_Inv(int*, int, int, int) (in /home/test/TestCPP)
==20797==    by 0x4009D4: Sort_And_count(int*, int, int) (in /home/test/TestCPP)
==20797==    by 0x4009BC: Sort_And_count(int*, int, int) (in /home/test/TestCPP)
==20797==    by 0x4009A2: Sort_And_count(int*, int, int) (in /home/test/TestCPP)
==20797==    by 0x400921: main (in /home/test/TestCPP)
==20797==  Address 0x5a1d0ec is 4 bytes after a block of size 8 alloc'd
==20797==    at 0x4C2B800: operator new[](unsigned long) (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==20797==    by 0x400AE8: Sort_And_count_split_Inv(int*, int, int, int) (in /home/test/TestCPP)
==20797==    by 0x4009D4: Sort_And_count(int*, int, int) (in /home/test/TestCPP)
==20797==    by 0x4009BC: Sort_And_count(int*, int, int) (in /home/test/TestCPP)
==20797==    by 0x4009A2: Sort_And_count(int*, int, int) (in /home/test/TestCPP)
==20797==    by 0x400921: main (in /home/test/TestCPP)

我把它归结为这一行：

tmp[k] = a[i];

这意味着k 达到的数字大于分配的int[] 数组的长度。由于您分配了high - low + 1 和k = low 的长度，因此您超过了边界情况2*low = high + 1。

我没有详细介绍您的实际排序算法以及为什么您输入了错误的 low（或 high），但这就是您的内存泄漏的来源.