实例介绍利用valgrind定位内存泄漏问题

我就是我 2022-03-20 13:12 250阅读 0赞

在前面的文章中， 我们简单了解了valgrind工具的用途以及安装， 以便大家能进行实际操作。 在本文中， 我们通过实例来看看如何利用valgrind来定位内存泄漏问题。 先看程序：

#include <stdio.h>#include <stdlib.h>char* getMemory(){ char *p = (char *)malloc(30);  return p;}int main(){ char *p = getMemory(); p = NULL;  return 0;}

只要是懂一点C/C++的人， 就很容易看出上述程序有内存泄漏， 我们用valgrind工具来检测下：

[root@xxx ~/valgrind-3.8.1/bin]# ./valgrind --tool=memcheck --leak-check=yes --show-reachable=yes ./a.out==19226== Memcheck, a memory error detector==19226== Copyright (C) 2002-2012, and GNU GPL'd, by Julian Seward et al.==19226== Using Valgrind-3.8.1 and LibVEX; rerun with -h for copyright info==19226== Command: ./a.out==19226== ==19226== ==19226== HEAP SUMMARY:==19226==     in use at exit: 30 bytes in 1 blocks==19226==   total heap usage: 1 allocs, 0 frees, 30 bytes allocated==19226== ==19226== 30 bytes in 1 blocks are definitely lost in loss record 1 of 1==19226==    at 0x4C278FE: malloc (vg_replace_malloc.c:270)==19226==    by 0x4005B5: getMemory() (in /root/valgrind-3.8.1/bin/a.out)==19226==    by 0x4005CC: main (in /root/valgrind-3.8.1/bin/a.out)==19226== ==19226== LEAK SUMMARY:==19226==    definitely lost: 30 bytes in 1 blocks==19226==    indirectly lost: 0 bytes in 0 blocks==19226==      possibly lost: 0 bytes in 0 blocks==19226==    still reachable: 0 bytes in 0 blocks==19226==         suppressed: 0 bytes in 0 blocks==19226== ==19226== For counts of detected and suppressed errors, rerun with: -v==19226== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 6 from 6)[root@xxx ~/valgrind-3.8.1/bin]#

我们可以很清楚地看到， 在getMemory调用malloc的那里， 有内存泄漏。最左边的19226表示进程号！

不过， 这样看着也挺蛋疼的， 如果代码过多， 肉眼据不太好分析了， 能不能把内存泄漏的代码行给找出来呢？ 当然能！ 回想一下我们之前介绍过得core dump定位到代码行的问题， 两个必要条件是:  编译时必须有-g参数； 编译后不能strip.  我们一起再看看：

[root@xxx ~/valgrind-3.8.1/bin]# g++ -g test.cpp [root@xxx ~/valgrind-3.8.1/bin]# [root@xxx ~/valgrind-3.8.1/bin]# ./valgrind --tool=memcheck --leak-check=yes --show-reachable=yes ./a.out==20448== Memcheck, a memory error detector==20448== Copyright (C) 2002-2012, and GNU GPL'd, by Julian Seward et al.==20448== Using Valgrind-3.8.1 and LibVEX; rerun with -h for copyright info==20448== Command: ./a.out==20448== ==20448== ==20448== HEAP SUMMARY:==20448==     in use at exit: 30 bytes in 1 blocks==20448==   total heap usage: 1 allocs, 0 frees, 30 bytes allocated==20448== ==20448== 30 bytes in 1 blocks are definitely lost in loss record 1 of 1==20448==    at 0x4C278FE: malloc (vg_replace_malloc.c:270)==20448==    by 0x4005B5: getMemory() (test.cpp:5)==20448==    by 0x4005CC: main (test.cpp:11)==20448== ==20448== LEAK SUMMARY:==20448==    definitely lost: 30 bytes in 1 blocks==20448==    indirectly lost: 0 bytes in 0 blocks==20448==      possibly lost: 0 bytes in 0 blocks==20448==    still reachable: 0 bytes in 0 blocks==20448==         suppressed: 0 bytes in 0 blocks==20448== ==20448== For counts of detected and suppressed errors, rerun with: -v==20448== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 6 from 6)[root@xxx ~/valgrind-3.8.1/bin]#

好了， 泄漏的代码行出来了。 这里， 我顺便说说以往介绍过得addr2line命令， 如果用addr2line -e a.out 0x4005B5,  也是能得出代码行的。

fix后的代码如下：

#include <stdio.h>#include <stdlib.h>char* getMemory(){ char *p = (char *)malloc(30);  return p;}int main(){ char *p = getMemory(); if(p != NULL) {  free(p);  p = NULL; }  return 0;}

我们再用valgrind工具检测一下：

可见， 没有内存泄漏了。 爽爽哒 ![羡慕][envy.gif] ![羡慕][envy.gif]

再分享一下我老师大神的人工智能教程吧。零基础！通俗易懂！风趣幽默！还带黄段子！希望你也加入到我们人工智能的队伍中来！[https://blog.csdn.net/jiangjunshow][https_blog.csdn.net_jiangjunshow]

[envy.gif]: /images/20220320/82756eb0a0ee4919a85a1c2e85b2d242.png
[https_blog.csdn.net_jiangjunshow]: https://blog.csdn.net/jiangjunshow/article/details/77338485