http://blog.csdn.net/ixidof/article/details/6638066
内存泄漏检查方法(for Linux) 
如果你更想读原始文档, 请参考glibc info的"Allocation Debugging" 一章 (执行info libc);
glibc提供了一个检查内存泄漏的方法, 前提是你的程序使用glibc的标准函数 分配内存(如malloc, alloc...): 
1. 在需要内存泄漏检查的代码的开始调用void mtrace(void) (在mcheck.h中有声明). mtrace为malloc等函数安装hook, 用于记录内存分配信息. 在需要内存泄漏检查的代码的结束调用void muntrace(void). 
注意: 一般情况下不要调用muntrace, 而让程序自然结束. 因为可能有些释放内存代码要到muntrace之后才运行. 
2. 用debug模式编译被检查代码(-g或-ggdb) 
3. 设置环境变量MALLOC_TRACE为一文件名, 这一文件将存有内存分配信息. 
4. 运行被检查程序, 直至结束或muntrace被调用. 
5. 用mtrace命令解析内存分配Log文件($MALLOC_TRACE) , (mtrace foo $MALLOC_TRACE, where foo is the executible name) ,如果有内存泄漏, mtrace会输出分配泄漏内存的代码位置,以及分配数量. 其他东西
1. 可以将mtrace, muntrace放入信号处理函数(USR1, USR2), 以动态地进行内存泄漏检查控制. 
2. mtrace是个perl代码, 如果你对符号地址与代码文本的转换感兴趣, 可以读一下. 
3. again, 尽量不要用muntrace()

For C++ Leak: 
检查内存泄漏的方法除glibc提供外还可以试试一些专用的程序;如: 
ccmalloc(http://www.inf.ethz.ch/personal/biere/projects/ccmalloc/ccmalloc-english.html) 
mpatrol(http://www.cbmamiga.demon.co.uk/mpatrol/) 
这俩个工具的功能相当不错,能对程序进行相当全面的检查 
很奇怪,redhat 9 居然不带mtrace perl脚本,只好下载gcc源码编译了
wget --passive-ftp ftp://rpmfind.net/linux/redhat/9/en/os/i386/SRPMS/glibc-2.3.2-11.9.src.rpm
rpm -ivh glibc*.src.rpm
cd /usr/src/redhat/SPECS/
rpmbuild -ba glibc-9.spec 
cd /var/tmp/glibc-2.3.2-root/usr/bin/
cp mtrace /usr/bin/
 
调试方法如下:
vi a.c

  1. 1 #include <mcheck.h>
  2. 2
  3. 3 int main()
  4. 4 {
  5. 5     mtrace();
  6. 6     malloc(10);
  7. 7     malloc(16);
  8. 8     return 0;
  9. 9 }

$gcc -g a.c #记得编译带-g调试选项 
$export MALLOC_TRACE=a.log 
$./a.out 
$unset MALLOC_TRACE #记得执行完后unset变量,否则可能运行其他命令可能覆盖log 
$mtrace a.out a.log 
Memory not freed:
-----------------
   Address     Size     Caller
0x09b08378      0xa  at /XXX/a.c:6
0x09b08388     0x10  at /XXX/a.c:7

可以看到,会显示未释放动态空间的代码具体位置.

  1. MTRACE(3) Linux Programmer's Manual MTRACE(3)

NAME         top

  1. mtrace, muntrace - malloc tracing

SYNOPSIS         top

  1. #include <mcheck.h>
  2.  
  3. void mtrace(void);
  4.  
  5. void muntrace(void);

DESCRIPTION         top

  1. The mtrace() function installs hook functions for the memory-
  2. allocation functions (malloc(3), realloc(3) memalign(3), free(3)).
  3. These hook functions record tracing information about memory
  4. allocation and deallocation. The tracing information can be used to
  5. discover memory leaks and attempts to free nonallocated memory in a
  6. program.
  7.  
  8. The muntrace() function disables the hook functions installed by
  9. mtrace(), so that tracing information is no longer recorded for the
  10. memory-allocation functions. If no hook functions were successfully
  11. installed by mtrace(), muntrace() does nothing.
  12.  
  13. When mtrace() is called, it checks the value of the environment
  14. variable MALLOC_TRACE, which should contain the pathname of a file in
  15. which the tracing information is to be recorded. If the pathname is
  16. successfully opened, it is truncated to zero length.
  17.  
  18. If MALLOC_TRACE is not set, or the pathname it specifies is invalid
  19. or not writable, then no hook functions are installed, and mtrace()
  20. has no effect. In set-user-ID and set-group-ID programs,
  21. MALLOC_TRACE is ignored, and mtrace() has no effect.

ATTRIBUTES         top

  1. For an explanation of the terms used in this section, see
  2. attributes(7).
  3.  
  4. ┌─────────────────────┬───────────────┬───────────┐
  5. Interface Attribute Value
  6. ├─────────────────────┼───────────────┼───────────┤
  7. mtrace(), muntrace() Thread safety MT-Unsafe
  8. └─────────────────────┴───────────────┴───────────┘

CONFORMING TO         top

  1. These functions are GNU extensions.

NOTES         top

  1. In normal usage, mtrace() is called once at the start of execution of
  2. a program, and muntrace() is never called.
  3.  
  4. The tracing output produced after a call to mtrace() is textual, but
  5. not designed to be human readable. The GNU C library provides a Perl
  6. script, mtrace(1), that interprets the trace log and produces human-
  7. readable output. For best results, the traced program should be
  8. compiled with debugging enabled, so that line-number information is
  9. recorded in the executable.
  10.  
  11. The tracing performed by mtrace() incurs a performance penalty (if
  12. MALLOC_TRACE points to a valid, writable pathname).

BUGS         top

  1. The line-number information produced by mtrace(1) is not always
  2. precise: the line number references may refer to the previous or
  3. following (nonblank) line of the source code.

EXAMPLE         top

  1. The shell session below demonstrates the use of the mtrace() function
  2. and the mtrace(1) command in a program that has memory leaks at two
  3. different locations. The demonstration uses the following program:
  4.  
  5. $ cat t_mtrace.c
  6. #include <mcheck.h>
  7. #include <stdlib.h>
  8. #include <stdio.h>
  9.  
  10. int
  11. main(int argc, char *argv[])
  12. {
  13. int j;
  14.  
  15. mtrace();
  16.  
  17. for (j = 0; j < 2; j++)
  18. malloc(100); /* Never freed--a memory leak */
  19.  
  20. calloc(16, 16); /* Never freed--a memory leak */
  21. exit(EXIT_SUCCESS);
  22. }
  23.  
  24. When we run the program as follows, we see that mtrace() diagnosed
  25. memory leaks at two different locations in the program:
  26.  
  27. $ cc -g t_mtrace.c -o t_mtrace
  28. $ export MALLOC_TRACE=/tmp/t
  29. $ ./t_mtrace
  30. $ mtrace ./t_mtrace $MALLOC_TRACE
  31. Memory not freed:
  32. -----------------
  33. Address Size Caller
  34. 0x084c9378 0x64 at /home/cecilia/t_mtrace.c:12
  35. 0x084c93e0 0x64 at /home/cecilia/t_mtrace.c:12
  36. 0x084c9448 0x100 at /home/cecilia/t_mtrace.c:16
  37.  
  38. The first two messages about unfreed memory correspond to the two
  39. malloc(3) calls inside the for loop. The final message corresponds
  40. to the call to calloc(3) (which in turn calls malloc(3)).

SEE ALSO         top

  1. mtrace(1), malloc(3), malloc_hook(3), mcheck(3)

COLOPHON         top

  1. This page is part of release 4.07 of the Linux man-pages project. A
  2. description of the project, information about reporting bugs, and the
  3. latest version of this page, can be found at
  4. https://www.kernel.org/doc/man-pages/.
  5.  
  6. GNU 2015-03-02 MTRACE(3)

用mtrace检查内存泄漏的更多相关文章

  1. mtrace检查内存泄漏

    内存泄漏检查方法(for Linux) 如果你更想读原始文档, 请参考glibc info的"Allocation Debugging" 一章 (执行info libc);glib ...

  2. c++检查内存泄漏

    使用_CrtDumpMemoryLeaks()函数检查内存泄漏 #include <cstdio> #include <cstdlib> #include <crtdbg ...

  3. iOS开发 如何检查内存泄漏

    本文转载至 http://mobile.51cto.com/iphone-423391.htm 在开发的时候内存泄漏是不可避免的,但是也是我们需要尽量减少的,因为内存泄漏可能会很大程度的影响程序的稳定 ...

  4. ios--->instrument的leaks来检查内存泄漏

    instrument来检查内存泄漏 1.第一步打开 或者: 然后选择leaks 2.若此时编译出现如下问题,可能是非debug版本造成的,切换成debug版本即可 打开工程的Edit Scheme选项 ...

  5. UWP开发入门(十三)——用Diagnostic Tool检查内存泄漏

    因为.NET的垃圾回收机制相当完善,通常情况下我们是不需要关心内存泄漏的.问题人一但傻起来,连自己都会害怕,几个页面跳啊跳的,内存蹭蹭的往上涨,拉都拉不住.这种时候我们就需要冷静下来,泡一杯热巧克力. ...

  6. Android内存管理(14)*使用开源库LeakCanary检查内存泄漏

    1.简介 它是一个非常简单好用的内存泄漏检测工具库.可以轻松检测Activity,Fragment的内存泄漏.如果有内存泄漏,它会产生一个通知. 2.资料 官网: https://github.com ...

  7. 用windbg 检查内存泄漏

    1.下载编译https://github.com/0cch/luadbg 2.编写脚本1.txt .load luadbg_v15*.sympath+ srv*c:\MyServerSymbols*h ...

  8. valgrind 代码检查,内存泄漏

    使用平台 linux 下载 http://valgrind.org/ 文档 http://valgrind.org/docs/manual/manual.html 博客 https://www.osc ...

  9. malloc钩子和内存泄漏工具mtrace、Valgrind

    一:malloc钩子函数 static void* (* old_malloc_hook) (size_t,const void *);static void (* old_free_hook)(vo ...

随机推荐

  1. IOS基础之 (十一) 内存管理 ARC

    一 内存管理 1. set 方法内存管理的相关参数 retain: release旧值,retain新值(值适用于OC对象) assign:直接赋值(set方法默认,适用于非OC对象类型,即基本数据类 ...

  2. 快速反射DataTable

    public class SetterWrapper<TTarget, TValue> { private Action<TTarget, TValue> _setter; p ...

  3. Circular Sequence,ACM/ICPC Seoul 2004,UVa 1584

    #include <stdio.h> #include <string.h> #define maxn 105 int lss(const char *s,int p,int ...

  4. C语言之参数传递

    学了四年的计算机,一直让自己比较苦恼的问题是C语言的参数传递问题,之所以说是苦恼,是因为在某年的一个学期,不幸接触到数据结构,光一个链表就把自己弄得死去活来的,而且自已一直就楞以为在操作的过程中,传递 ...

  5. SMTP 简单邮件传输协议

    SMTP 锁定 本词条由“科普中国”百科科学词条编写与应用工作项目 审核 . SMTP(Simple Mail Transfer Protocol)即简单邮件传输协议,它是一组用于由源地址到目的地址传 ...

  6. 最长不下降子序列 O(nlogn) || 记忆化搜索

    #include<stdio.h> ] , temp[] ; int n , top ; int binary_search (int x) { ; int last = top ; in ...

  7. cc表示Cocos核心,ccs代表CocoStudio,ccui代表CocoStudio的UI控件

    cc表示Cocos核心,ccs代表CocoStudio,ccui代表CocoStudio的UI控件

  8. 淘宝(阿里百川)手机客户端开发日记第五篇 SharedPreferences使用详解

    我们知道,Android中数据存储技术由于如下几种 1 使用SharedPreferences存储数据 2 文件存储数据 3 SQLite数据库存储数据 4 使用ContentProvider存储数据 ...

  9. HDU 4857 Couple doubi(找循环节)

    题目链接:http://acm.hdu.edu.cn/showproblem.php?pid=4861 解题报告:桌子上有k个球 ,第i个球的价值wi = 1^i+2^i+...+(p-1)^i (m ...

  10. UVa 11464 - Even Parity

    解题报告:题目大意有一个N×N的矩阵,矩阵中的元素只有1或0,如果说对于一个矩阵,它的所有的点的上下左右的点的和是偶数,则称这个矩阵为偶数矩阵,现在给你一个任意的矩阵,要求的是如果要把这个矩阵变成偶数 ...