?
场景:
1. 开发Mac,iOS程序时,一般都会涉及到多线程开发, 比如工作线程在更新数据模型, 主线程根据数据模型来显示界面.
比如最常用的NSMutableArray, NSMutableDictionary等, 这类集合在调用读写方法时并不是线程安全的，也就是说会出现数据不一致性和崩溃现象.
2. 解决办法就是使用@syncronized()关键字, 有点类似Java.
?
test.cpp
?
//
// main.c
// test_sync
//
// Created by daih on 11/12/15.
// Copyright (c) 2015 infoworld. All rights reserved.
//
#include
#import
#include
#include pthread.h #include
#include
static int THREADCOUNT = 100; // 需要多线程测试的类,类里有集合类型属性和原始类型属性. @interface MyDevice : NSObject { NSMutableArray* usbArray; // 集合类型 NSInteger usbCount; //原始类型 } @property(readonly) NSMutableArray* usbArray; // 原始类型可以设置原子访问只是保证不会读脏数据,但是不能保证算数运算的同步. // 所以还是得加锁. @property(readwrite,assign,atomic) NSInteger usbCount; @end @implementation MyDevice @synthesize usbArray; @synthesize usbCount; -(id)init { [super init]; usbArray = [NSMutableArray new]; usbCount = 0; return self; } @end static MyDevice* gDevice = [MyDevice new]; static int gCount = 0; void* ThreadFunc(void* data) { for (int i = 0; i < 10000; ++i) { // 每次＋1,注意,这里gDevice.usbCount+1并不是原子操作，所以有可能多个线程的值是一样的. [gDevice setUsbCount:gDevice.usbCount+1]; // 如果是多线程访问,读写usbArray都需要加锁. @synchronized(gDevice.usbArray) { [gDevice.usbArray addObject:@(i)]; } } ++gCount; return NULL; } void TestSyncVariable() { for (int i = 0; i < THREADCOUNT; ++i) { pthread_t t; pthread_create(&t,NULL,ThreadFunc,NULL); pthread_detach(t); } } int main(int argc, const char * argv[]) { // insert code here... CFShow(CFSTR(Hello, World! )); TestSyncVariable(); while (gCount != THREADCOUNT) { sleep(2); } std::cout << [gDevice.usbArray count]: << [gDevice.usbArray count] << std::endl; std::cout << gDevice.usbCount: << gDevice.usbCount << std::endl; assert(gDevice.usbCount <= [gDevice.usbArray count]); assert([gDevice.usbArray count] == THREADCOUNT*10000); CFShow(CFSTR(Bye, World! )); return 0; }
输出:
?
?
Hello, World!
[gDevice.usbArray count]: 1000000
gDevice.usbCount: 999997
Bye, World!
Program ended with exit code: 0
?