TCMalloc 安装和使用
Author:Echo Chen(陈斌)
Email:chenb19870707@gmail.com
Date:October 20th, 2014
前面三篇译文《TCMalloc:线程缓冲的Malloc》、《使用TCMalloc的堆栈检查》、《使用TCMalloc进行堆栈分析》介绍了TCMalloc的基本原理,以及堆栈分析和检查工具,TCMalloc优点很多,比glibc 2.3的malloc快、自带的堆栈工具可以轻松找出内存瓶颈和内存泄漏,给服务器开发指明了一条新的道路。
google-perftools:http://code.google.com/p/google-perftools/gperftools-2.1.tar.gz
libunwind:http://download.savannah.gnu.org/releases/libunwind/libunwind-1.1.tar.gz
64位操作系统请先安装 libunwind库,32位操作系统不要安装。libunwind库为基于64位CPU和操作系统的程序提供了基本的堆栈辗转开解功能,其中包括用于输出堆栈跟踪的API、用于以编程方式辗转开解堆栈的API以及支持C++异常处理机制的API。
1: #tar zxvf libunwind-1.1.tar.gz
2: #cd libunwind-1.1
3: #./configure
4: #make
5: #make install
1: #tar zxvf tar zxvf gperftools-2.1.tar.gz
2: #cd gperftools-2.1
3: #./configure
4: #make
5: #make install
1: echo "/usr/local/lib" > /etc/ld.so.conf.d/usr_local_lib.conf
2: /sbin/ldconfig
在Makefile中 添加TCMalloc链接,注意:这里为了进行内存泄漏分析,一定要将TCMalloc链接在最后,官方文档里说:堆栈检查器可能误解列在它后面的链接库的一些内存。
1: # funshion wuhan game studio
2: # Created by zeng jun fei in 2013-08-08
3:
4: CXX = g++
5: # debug
6: CXXFLAGS = -g -I../BaseCode -I../../CommonSrc -Wall -D_POSIX_MT_ -O0
7: CXXLFLAGS = -g -Wall -L../bin -lBaseCode -lpthread -lprotobuf -rdynamic -ltcmalloc
8:
9: # release
10: # CXXFLAGS = -O3 -g -I../NetworkEngine -Wall
11: # CXXLFLAGS = -O3 -g -Wall -L../NetworkEngine -lnetwork
12:
13: LIB_NETWORK = ../bin/libBaseCode.a
14:
15: OBJS = $(patsubst %.cpp,%.o,$(wildcard *.cpp))
16: SRCS = $(OBJS:%.o=%.cpp)
17: DEPS = $(OBJS:%.o=%.d)
18:
19: ALL_TARGETS = ../bin/GateServer
20:
21: all: $(ALL_TARGETS)
22:
23: -include $(DEPS)
24: $(DEPS): %.d: %.cpp
25: @$(CXX) -MM $(CXXFLAGS) $< > $@.$$$$; sed 's,\($*\)\.o[ :]*,\1.o $@ : ,g' < $@.$$$$ > $@; rm -f $@.$$$$
26:
27: $(OBJS): %.o: %.cpp
28: $(CXX) -c $(CXXFLAGS) $< -o $@
29:
30: $(ALL_TARGETS): $(OBJS) $(LIB_NETWORK)
31: $(CXX) $(OBJS) -o $@ $(CXXLFLAGS)
32:
33: clean:
34: @rm -rf $(OBJS) $(ALL_TARGETS) *.d
35:
首先,设置pperf的环境变量:export PPROF_PATH=/usr/local/bin/pprof
测试代码:
1: #include
2: using namespace std;
3:
4: int main()
5: {
6: int *p = new int();
7: return 0;
8: }
编译:g++ main.cpp -o main -ltcmalloc -g -O0
内存泄漏检查: env HEAPCHECK=normal ./main
结果:
WARNING: Perftools heap leak checker is active -- Performance may suffer
Have memory regions w/o callers: might report false leaks
Leak check _main_ detected leaks of 4 bytes in 1 objects
The 1 largest leaks:
Using local file ./main.
Leak of 4 bytes in 1 objects allocated from:
@ 4007a6 main
@ 7f1734263d1d __libc_start_main
@ 4006d9 _start
If the preceding stack traces are not enough to find the leaks, try running THIS shell command:
pprof ./main "/tmp/main.54616._main_-end.heap" --inuse_objects --lines --heapcheck --edgefraction=1e-10 --nodefraction=1e-10 --gv
If you are still puzzled about why the leaks are there, try rerunning this program with HEAP_CHECK_TEST_POINTER_ALIGNMENT=1 and/or with HEAP_CHECK_MAX_POINTER_OFFSET=-1
If the leak report occurs in a small fraction of runs, try running with TCMALLOC_MAX_FREE_QUEUE_SIZE of few hundred MB or with TCMALLOC_RECLAIM_MEMORY=false, it might help find leaks more repeatably
Exiting with error code (instead of crashing) because of whole-program memory leaks
上面的报告显示有4个字节的内存泄漏,并提示使用pprof进一步跟踪泄漏来源的方法。
包括normal在内总共有4种泄漏检查方式:minimal,忽略进入main函数之前的初始化过程;normal,报告所有的无法再引用的内存对象;strick,在normal的基础上增加一些额外的检查;draconian,在程序退出的时候存在未释放的内存的情况下报错。
根据《使用TCMalloc的堆栈检查》,除了前面使用env命令行的全局内存泄漏检查方式外,还可以作对代码段的更加细粒度的泄漏检查。这里需要先在源代码中包含头文件google/heap-checker.h。 实例代码如下:
1: #include
2: #include
3: #include
4: #include
5: int* fun(int n)
6: {
7: int *p2;
8: HeapLeakChecker heap_checker("fun");
9: {
10: new int[n];
11: p2=new int[n];
12: //delete [] p1;
13: }
14: assert(!heap_checker.NoLeaks());
15: return p2;
16: }
17: int main(int argc,char* argv[])
18: {
19: int n;
20: scanf("%d",&n);
21: int *p=fun(n);
22: delete [] p;
23: return 0;
24: }
此外,还可以忽略某些已知的内存泄漏:
1: #include
2: …
3: void *mark = HeapLeakChecker::GetDisableChecksStart();
4:
5: HeapLeakChecker::DisableChecksToHereFrom(mark);
感觉跟valgrind效果差不多,但是valgrind还能查出内存越界,更加优秀。
本来研究TCMalloc是为了优化游戏服务器,解决游戏服务器后期玩家流失后,占用大量内存的浪费,结果发现由于我们游戏服务器为了防止内存碎片和频繁调用new和delete带来的性能损耗,使用了大量的内存池对象(如装备池、技能池、玩家池),这些池都不会调用delete还给系统,所以即使使用了TCMalloc也不会有内存释放,现在也明白了服务器维护的意义,当然这和服务器框架设计很有关系,如果没有这些缓冲池,直接调用new和delete,TCMalloc会是一个很好的选择。
-
Echo Chen:Blog.csdn.net/chen19870707
-
手机扫一扫
移动阅读更方便
你可能感兴趣的文章