cocos2dx多线程以及线程同步 与 cocos2dx内存管理与多线程问题

摘要: cocos2d-x引擎在内部实现了一个庞大的主循环，每帧之间更新界面，如果耗时的操作放到了主线程中，游戏的界面就会卡，这是不能容忍的，游戏最基本的条件就是流畅性，这就是为什么游戏开发选择C++的原因。另外现在双核手机和四核手机越来越普遍了，是时候使用多线程来挖掘硬件的潜力了

cocos2d-x引擎在内部实现了一个庞大的主循环，每帧之间更新界面，如果耗时的操作放到了主线程中，游戏的界面就会卡，这是不能容忍的，游戏最基本的条件就是流畅性，这就是为什么游戏开发选择C++的原因。另外现在双核手机和四核手机越来越普遍了，是时候使用多线程来挖掘硬件的潜力了。

1.环境搭建

cocos2d-x中的多线程使用pthread就可以实现跨平台，而且也不是很难理解。使用pthread需要先配置一下工程。右击工程----->属性----->配置属性---->链接器----->输入---->附加依赖项中添加pthreadVCE2.lib，如下图



接着添加附加包含目录，右击项目，属性----->C/C++---->常规----->附加包含目录加入pthread头文件所在的目录



这样，环境就搭建起来了。

2.多线程的使用

使用pthread来实现多线程，最重要的一个函数是
[cpp] view plaincopy

PTW32_DLLPORT int PTW32_CDECL pthread_create (pthread_t * tid,//线程的标示

const pthread_attr_t * attr, //创建线程的参数

void *(*start) (void *), //入口函数的指针

void *arg); //传递给线程的数据

在HelloWorldScene.h文件中

[cpp]
view plain
copy

pthread_t pidrun,pidgo;

static void* th_run(void *r);

static void* th_go(void *r);

定义了两个函数和两个线程的标识。

然后自定义了一个类，用于给线程传递数据。Student类如下：

[cpp]
view plain
copy

#pragma once

#include <string>

class Student

{

public:

Student(void);

Student(std::string name,int age,std::string sex);

~Student(void);

std::string name;

int age;

std::string sex;

};

源文件如下

[cpp]
view plain
copy

#include "Student.h"

#include "cocos2d.h"

Student::Student(void)

{

}

Student::~Student(void)

{

cocos2d::CCLog("delete data");

}

Student::Student(std::string name,int age,std::string sex)

{

this->name=name;

this->age=age;

this->sex=sex;

}

在退出菜单的回调函数中启动两个线程：

[cpp]
view plain
copy

void HelloWorld::menuCloseCallback(CCObject* pSender)

{

Student *temp=new Student(std::string("zhycheng"),23,std::string("male"));

pthread_mutex_init(&mutex,NULL);

pthread_create(&pidrun,NULL,th_run,temp);//启动线程

pthread_create(&pidgo,NULL,th_go,0);

}

可以看到，将Student的指针传递给了pidrun线程，那么在pidrun线程中获得Student信息如下：

[cpp]
view plain
copy

Student *s=(Student*)(r);

CCLog("name is %s,and age is %d,sex is %s",s->name.c_str(),s->age,s->sex.c_str());

delete s;

3.线程同步

使用了线程，必然就要考虑到线程同步，不同的线程同时访问资源的话，访问的顺序是不可预知的，会造成不可预知的结果。

这里使用pthread_mutex_t来实现同步，下面我来演示一下使用多线程实现卖票系统。卖票的时候，是由多个窗口同时卖票，这里要做到一张票不要卖出去两次，不要出现有票却无法卖的结果。
在线程函数th_run和th_go中来卖票，票的数量是一个全局变量，每卖出去一张票，就将票的数量减一。其中同步的pthread_mutex_t也是一个全局变量，就用它来实现线程同步。

[cpp]
view plain
copy

void* HelloWorld::th_run(void *r)

{

Student *s=(Student*)(r);

CCLog("name is %s,and age is %d,sex is %s",s->name.c_str(),s->age,s->sex.c_str());

delete s;

while(true)

{

pthread_mutex_lock(&mutex);

if(ticket>0)

{

CCLog("thread run sell %d",ticket);

ticket--;

pthread_mutex_unlock(&mutex);

}

else

{

pthread_mutex_unlock(&mutex);

break;

}

Sleep(1);

//Usleep(10);

}

return NULL;

}

[cpp]
view plain
copy

void* HelloWorld::th_go(void *r)

{

while(true)

{

pthread_mutex_lock(&mutex);

if(ticket>0)

{

CCLog("thread go sell %d",ticket);

ticket--;

pthread_mutex_unlock(&mutex);

}

else

{

pthread_mutex_unlock(&mutex);

break;

}

Sleep(1);

}

return NULL;

}

mutex被锁定后，其他线程若再想锁定mutex的话，必须等待，当该线程释放了mutex之后，其他线程才能锁定mutex。Sleep()函数可以使得该线程休眠，单位是毫秒。下面是卖票的结果：

[cpp]
view plain
copy

name is zhycheng,and age is 23,sex is male

delete data

thread run sell 100

thread run sell 99

thread go sell 98

thread go sell 97

thread run sell 96

thread go sell 95

thread go sell 94

thread run sell 93

thread go sell 92

thread run sell 91

thread go sell 90

thread go sell 89

thread run sell 88

thread go sell 87

thread run sell 86

thread go sell 85

thread run sell 84

thread go sell 83

thread run sell 82

thread go sell 81

thread run sell 80

thread go sell 79

thread run sell 78

thread go sell 77

thread run sell 76

thread go sell 75

thread run sell 74

thread go sell 73

thread run sell 72

thread go sell 71

thread run sell 70

thread go sell 69

thread go sell 68

thread run sell 67

thread go sell 66

thread run sell 65

thread go sell 64

thread run sell 63

thread go sell 62

thread run sell 61

thread go sell 60

thread run sell 59

thread go sell 58

thread run sell 57

thread go sell 56

thread run sell 55

thread go sell 54

thread run sell 53

thread run sell 52

thread go sell 51

thread run sell 50

thread go sell 49

thread run sell 48

thread go sell 47

thread run sell 46

thread go sell 45

thread run sell 44

thread run sell 43

thread go sell 42

thread run sell 41

thread run sell 40

thread go sell 39

thread run sell 38

thread run sell 37

thread run sell 36

thread run sell 35

thread go sell 34

thread run sell 33

thread run sell 32

thread go sell 31

thread run sell 30

thread run sell 29

thread run sell 28

thread run sell 27

thread run sell 26

thread run sell 25

thread go sell 24

thread run sell 23

thread go sell 22

thread go sell 21

thread run sell 20

thread go sell 19

thread run sell 18

thread run sell 17

thread go sell 16

thread run sell 15

thread go sell 14

thread go sell 13

thread run sell 12

thread go sell 11

thread go sell 10

thread run sell 9

thread go sell 8

thread run sell 7

thread go sell 6

thread go sell 5

thread run sell 4

thread go sell 3

thread run sell 2

thread run sell 1

可以看到，这个打印结果正确无误。如果不加mutex会是什么样的结果呢，我将线程同步的mutex注释掉，输出的结果为：

[cpp]
view plain
copy

name is zhycheng,and age is 23,sex is male

delete data

thread run sell 100

thread run sell 99

thread run sell 98

thread go sell 97

thread run sell 96

thread go sell 95

thread run sell 94

thread go sell 94

thread run sell 92

thread run sell 91

thread go sell 90

thread run sell 89

thread go sell 88

thread run sell 87

thread run sell 86

thread go sell 86

thread go sell 84

thread run sell 83

thread go sell 82

thread run sell 81

thread go sell 80

thread run sell 79

thread run sell 78

thread go sell 77

thread run sell 76

thread run sell 75

thread go sell 74

thread run sell 73

thread go sell 72

thread run sell 71

thread go sell 70

thread go sell 69

thread run sell 68

thread go sell 67

thread go sell 66

thread run sell 65

thread go sell 64

thread go sell 63

thread run sell 62

thread go sell 61

thread run sell 60

thread run sell 59

thread run sell 58

thread run sell 57

thread run sell 56

thread run sell 55

thread go sell 54

thread run sell 54

thread go sell 52

thread run sell 52

thread go sell 50

thread run sell 50

thread go sell 49

thread run sell 47

thread go sell 47

thread go sell 45

thread run sell 45

thread run sell 43thread go sell 43

thread run sell 41

thread go sell 41

thread go sell 39

thread run sell 39

thread run sell 37

thread go sell 37

thread go sell 35

thread run sell 35

thread go sell 33thread run sell 33

thread go sell 31thread run sell 31

thread go sell 29

thread run sell 29

thread go sell 27

thread run sell 27

thread go sell 25

thread run sell 25

thread run sell 23

thread go sell 23

thread run sell 21

thread go sell 21

thread go sell 19

thread run sell 19

thread run sell 17

thread go sell 17

thread go sell 15

thread run sell 15

thread run sell 13

thread go sell 13

thread run sell 11thread go sell 11

thread go sell 9

thread run sell 9

thread run sell 7

thread go sell 7

thread go sell 5thread run sell 5

thread go sell 3

thread run sell 3

thread go sell 1

thread run sell 1

可以看到，有的票卖了两次，有的票就没卖。

4.注意

1.Sleep()函数是使得线程休眠的函数，这个函数不跨平台，仅仅在windows上能用，其他平台使用usleep。
2.在非主线程中不能使用cocos2d-x管理内存的CCObject::retain(), CCObject::release() 者CCObject::autorelease()，因为CCAutoreleasePool不是线程安全的，OPENGL的上下文也不是线程安全的，所以不要再非主线程中使用cocos2d-x的API和UI操作。

cocos2dx内存管理与多线程问题:

Cocos2d-x的内存管理采用Objective-C的机制，大喜过望。因为只要坚持Objective-C的原则“谁创建谁释放，谁备份谁释放”的原则即可确保内存使用不易出现Bug。
但是因为本身开放的游戏需要使用到多线程技术，导致测试的时候总是莫名其妙的导致空指针错误。而且是随机出现，纠结了2天无果后，开始怀疑Cocos2d-X的内存本身管理可能存在问题。怀着这样的想法，
一步一步的调试，发现经常出现指针异常的变量总是在调用autorelease后一会就莫名其妙再使用的时候就抛异常。狠下心，在它的析构函数里面断点+Log输出信息。发现对象被释放了。一时也很迷糊，因为对象只是
autorelease,并没有真正释放，是谁导致它释放的？

然后就去看了CCAutoreleasePool的源码，发现存在Cocos2d-X的内存管理在多线程的情况下存在如下问题



如图：thread 1和thread 2是独立的两个线程，它们之间存在CPU分配的交叉集，我们在time 1的时候push一个autorelease的自动释放池，在该线程的末尾，即time 3的时候pop它。同理在thread 2的线程里面，在time 2的时候push一个自动释放池，在time 4的时候释放它，即Pop.
此时我们假设在thread 2分配得到CPU的时候有一个对象obj自动释放，即obj-autorelease().那么在time 3的时候会发生是么事情呢？
答案很简单，就是obj在time 3的时候就被释放了，而我们期望它在time 4的时候才释放。所以就导致我上面说的，在多线程下面，cocos2d-x的autorelease变量会发生莫名其妙的指针异常。

解决办法：在PoolManager给每个线程根据pthread_t的线程id生成一个CCArray的stack的嵌套管理自动释放池。源码如下
所以我在Push的时候根据当前线程的pthread_t的线程id生成一个CCArray的stack来存储该线程对应的Autoreleasepool的嵌套对象
源码如下

//--------------------------------------------------------------------
//
// CCPoolManager
//
//--------------------------------------------------------------------

/////【diff - begin】- by layne//////

CCPoolManager* CCPoolManager::sharedPoolManager()
{
if (s_pPoolManager == NULL)
{
s_pPoolManager = new CCPoolManager();
}
return s_pPoolManager;
}

void CCPoolManager::purgePoolManager()
{
CC_SAFE_DELETE(s_pPoolManager);
}

CCPoolManager::CCPoolManager()
{
//    m_pReleasePoolStack = new CCArray();
//    m_pReleasePoolStack->init();
//    m_pCurReleasePool = 0;

m_pReleasePoolMultiStack = new CCDictionary();
}

CCPoolManager::~CCPoolManager()
{

//    finalize();

//    // we only release the last autorelease pool here
//    m_pCurReleasePool = 0;
//    m_pReleasePoolStack->removeObjectAtIndex(0);
//
//    CC_SAFE_DELETE(m_pReleasePoolStack);

finalize();

CC_SAFE_DELETE(m_pReleasePoolMultiStack);
}

void CCPoolManager::finalize()
{
if(m_pReleasePoolMultiStack->count() > 0)
{
//CCAutoreleasePool* pReleasePool;
CCObject* pkey = NULL;
CCARRAY_FOREACH(m_pReleasePoolMultiStack->allKeys(), pkey)
{
if(!pkey)
break;
CCInteger *key = (CCInteger*)pkey;
CCArray *poolStack = (CCArray *)m_pReleasePoolMultiStack->objectForKey(key->getValue());
CCObject* pObj = NULL;
CCARRAY_FOREACH(poolStack, pObj)
{
if(!pObj)
break;
CCAutoreleasePool* pPool = (CCAutoreleasePool*)pObj;
pPool->clear();
}
}
}
}

void CCPoolManager::push()
{
//    CCAutoreleasePool* pPool = new CCAutoreleasePool();       //ref = 1
//    m_pCurReleasePool = pPool;
//
//    m_pReleasePoolStack->addObject(pPool);                   //ref = 2
//
//    pPool->release();                                       //ref = 1

pthread_mutex_lock(&m_mutex);

CCArray* pCurReleasePoolStack = getCurReleasePoolStack();
CCAutoreleasePool* pPool = new CCAutoreleasePool();         //ref = 1
pCurReleasePoolStack->addObject(pPool);                               //ref = 2
pPool->release();                                           //ref = 1

pthread_mutex_unlock(&m_mutex);
}

void CCPoolManager::pop()
{
//    if (! m_pCurReleasePool)
//    {
//        return;
//    }
//
//    int nCount = m_pReleasePoolStack->count();
//
//    m_pCurReleasePool->clear();
//
//    if(nCount > 1)
//    {
//        m_pReleasePoolStack->removeObjectAtIndex(nCount-1);
//
//        //         if(nCount > 1)
//        //         {
//        //             m_pCurReleasePool = m_pReleasePoolStack->objectAtIndex(nCount - 2);
//        //             return;
//        //         }
//        m_pCurReleasePool = (CCAutoreleasePool*)m_pReleasePoolStack->objectAtIndex(nCount - 2);
//    }
//
//    /*m_pCurReleasePool = NULL;*/

pthread_mutex_lock(&m_mutex);

CCArray* pCurReleasePoolStack = getCurReleasePoolStack();
CCAutoreleasePool* pCurReleasePool = getCurReleasePool();
if (pCurReleasePoolStack && pCurReleasePool)
{
int nCount = pCurReleasePoolStack->count();

pCurReleasePool->clear();

if(nCount > 1)
{
pCurReleasePoolStack->removeObject(pCurReleasePool);
}
}

pthread_mutex_unlock(&m_mutex);
}

void CCPoolManager::removeObject(CCObject* pObject)
{
//    CCAssert(m_pCurReleasePool, "current auto release pool should not be null");
//
//    m_pCurReleasePool->removeObject(pObject);

pthread_mutex_lock(&m_mutex);
CCAutoreleasePool* pCurReleasePool = getCurReleasePool();
CCAssert(pCurReleasePool, "current auto release pool should not be null");

pCurReleasePool->removeObject(pObject);
pthread_mutex_unlock(&m_mutex);
}

void CCPoolManager::addObject(CCObject* pObject)
{
//    getCurReleasePool()->addObject(pObject);

pthread_mutex_lock(&m_mutex);
CCAutoreleasePool* pCurReleasePool = getCurReleasePool(true);
CCAssert(pCurReleasePool, "current auto release pool should not be null");

pCurReleasePool->addObject(pObject);
pthread_mutex_unlock(&m_mutex);
}

CCArray* CCPoolManager::getCurReleasePoolStack()
{
CCArray* pPoolStack = NULL;
pthread_t tid = pthread_self();
if(m_pReleasePoolMultiStack->count() > 0)
{
pPoolStack = (CCArray*)m_pReleasePoolMultiStack->objectForKey((int)tid);
}

if (!pPoolStack) {
pPoolStack = new CCArray();
m_pReleasePoolMultiStack->setObject(pPoolStack, (int)tid);
pPoolStack->release();
}

return pPoolStack;
}

CCAutoreleasePool* CCPoolManager::getCurReleasePool(bool autoCreate)
{
//    if(!m_pCurReleasePool)
//    {
//        push();
//    }
//
//    CCAssert(m_pCurReleasePool, "current auto release pool should not be null");
//
//    return m_pCurReleasePool;

CCAutoreleasePool* pReleasePool = NULL;

CCArray* pPoolStack = getCurReleasePoolStack();
if(pPoolStack->count() > 0)
{
pReleasePool = (CCAutoreleasePool*)pPoolStack->lastObject();
}

if (!pReleasePool && autoCreate) {
CCAutoreleasePool* pPool = new CCAutoreleasePool();         //ref = 1
pPoolStack->addObject(pPool);                               //ref = 2
pPool->release();                                           //ref = 1

pReleasePool = pPool;
}

return pReleasePool;
}