多線程

iOS常見的多線程方案

image.png

GCD源碼：https://github.com/apple/swift-corelibs-libdispatch

image.png

死鎖

- (void)viewDidLoad {
    [super viewDidLoad];
    
    NSLog(@"1");
    
    dispatch_queue_t queue = dispatch_get_main_queue();
    dispatch_sync(queue, ^{
        NSLog(@"2");
    });
    
    NSLog(@"3");
}

當前隊列需要等待viewDidLoad執(zhí)行完,才能執(zhí)行block里面的內(nèi)容,而viewDidLoad又需要等待block執(zhí)行完.雙方都在等待對方完成,而造成的死鎖.

另一個死鎖

- (void)interView02
{
    NSLog(@"執(zhí)行任務1");
    
    dispatch_queue_t queue = dispatch_queue_create("myqueue", DISPATCH_QUEUE_SERIAL);
    
    dispatch_async(queue, ^{
        
        NSLog(@"執(zhí)行任務2");
        
        dispatch_sync(queue, ^{
            
            NSLog(@"執(zhí)行任務3");
            
        });
        
        NSLog(@"執(zhí)行任務4");
        
    });
    
    NSLog(@"執(zhí)行任務5");
}

網(wǎng)上一道面試題

- (void)test
{
    NSLog(@"3");
}

- (void)touchesBegan:(NSSet<UITouch *> *)touches withEvent:(UIEvent *)event
{
    dispatch_queue_t queque = dispatch_get_global_queue(0, 0);
    
    dispatch_async(queque, ^{
        
        NSLog(@"1");
        [self performSelector:@selector(test) withObject:nil
                   afterDelay:0];
        NSLog(@"2");
        
        [[NSRunLoop currentRunLoop]addPort:[NSPort new] forMode:NSDefaultRunLoopMode];
        [[NSRunLoop currentRunLoop]runMode:NSDefaultRunLoopMode beforeDate:[NSDate distantFuture]];
        
    });
    
}

[self performSelector:@selector(test) withObject:nil afterDelay:0];的本質(zhì)是往runloop中添加定時器,子線程默認沒有開啟runloop.

GNUStep

是GNU計劃的項目之一,它將cocoa的oc庫開源實現(xiàn)了一遍,非常接近蘋果的源碼.
源碼地址:http://www.gnustep.org/resources/downloads.php 這里可以看到runloop的相關實現(xiàn)

多線程的隱患

當多個線程訪問同一塊資源時，很容易引發(fā)數(shù)據(jù)錯亂和數(shù)據(jù)安全問題
eg.存錢取錢問題.


@interface ViewController ()
@property (nonatomic,assign) int moeny;
@end

@implementation ViewController

- (void)viewDidLoad {
    [super viewDidLoad];
    
    self.moeny = 100;
    // Do any additional setup after loading the view.
    dispatch_queue_t queue = dispatch_get_global_queue(0, 0);
    
    dispatch_async(queue, ^{
        for (int i = 0; i < 10; i++) {
            [self saveMoeny];
        }
    });
    
    dispatch_async(queue, ^{
        for (int i = 0; i < 10; i++) {
            [self drawMoney];
        }
    });
}

//取錢
- (void)drawMoney
{
    int oldMoeny = self.moeny;
    sleep(0.2);
    oldMoeny = oldMoeny - 20;
    self.moeny = oldMoeny;
    NSLog(@"取20, 還剩%d--%@",oldMoeny,[NSThread currentThread]);
}

//存錢
- (void)saveMoeny
{
    int newMoney = self.moeny;
    sleep(.2);
    newMoney = newMoney + 50;
    self.moeny = newMoney;
    NSLog(@"存50, 還剩%d--%@",newMoney,[NSThread currentThread]);
}

線程同步技術

  • OSSpinLock
  • os_unfair_lock
  • pthread_mutex
  • dispatch_semaphore
  • dispatch_queue(DISPATCH_QUEUE_SERIAL)
  • NSLock
  • NSRecursiveLock
  • NSCondition
  • NSConditionLock
  • @synchronized

OSSpinLock

  • OSSpinLock叫做”自旋鎖”，等待鎖的線程會處于忙等（busy-wait）狀態(tài)，一直占用著CPU資源
  • 目前已經(jīng)不再安全，可能會出現(xiàn)優(yōu)先級反轉(zhuǎn)問題
  • 如果等待鎖的線程優(yōu)先級較高，它會一直占用著CPU資源，優(yōu)先級低的線程就無法釋放鎖
  • #import <libkern/OSAtomic.h>

//取錢
- (void)drawMoney
{
    OSSpinLockLock(&_lock);
    
    int oldMoeny = self.moeny;
    sleep(0.2);
    oldMoeny = oldMoeny - 20;
    self.moeny = oldMoeny;
    NSLog(@"取20, 還剩%d--%@",oldMoeny,[NSThread currentThread]);
    
    OSSpinLockUnlock(&_lock);
}

//存錢
- (void)saveMoeny
{
    OSSpinLockLock(&_lock);
    int newMoney = self.moeny;
    
    sleep(.2);
    newMoney = newMoney + 50;
    self.moeny = newMoney;
    NSLog(@"存50, 還剩%d--%@",newMoney,[NSThread currentThread]);
    OSSpinLockUnlock(&_lock);
}

新建基類lockDemo,以后每個種類的lock繼承于它便于演示.

@interface lockDemo : NSObject

- (void)drawAndSaveMoneyTest;
- (void)ticketTest;

- (void)drawMoney;
- (void)saveMoeny;
- (void)saleTicket;

@end

@interface lockDemo ()

@property (nonatomic,assign) int moeny;
@property (nonatomic,assign) int ticketsCount;


@end

@implementation lockDemo

- (void)drawAndSaveMoneyTest
{
    self.moeny = 100;
    dispatch_queue_t queue = dispatch_get_global_queue(0, 0);
    
    dispatch_async(queue, ^{
        for (int i = 0; i < 10; i++) {
            [self saveMoeny];
        }
    });
    
    dispatch_async(queue, ^{
        for (int i = 0; i < 10; i++) {
            [self drawMoney];
        }
    });
}

//取錢
- (void)drawMoney
{
    
    
    int oldMoeny = self.moeny;
    sleep(0.2);
    oldMoeny = oldMoeny - 20;
    self.moeny = oldMoeny;
    NSLog(@"取20, 還剩%d--%@",oldMoeny,[NSThread currentThread]);
    
    
}

//存錢
- (void)saveMoeny
{
    
    int newMoney = self.moeny;
    sleep(.2);
    newMoney = newMoney + 50;
    self.moeny = newMoney;
    NSLog(@"存50, 還剩%d--%@",newMoney,[NSThread currentThread]);
    
}

- (void)saleTicket{
    
    int oldTicketsCount = self.ticketsCount;
    sleep(.2);
    oldTicketsCount--;
    self.ticketsCount = oldTicketsCount;
    NSLog(@"還剩%d張票 - %@", oldTicketsCount, [NSThread currentThread]);
}
/**
 賣票演示
 */
- (void)ticketTest
{
    self.ticketsCount = 15;
    
    dispatch_queue_t queue = dispatch_get_global_queue(0, 0);
    
    dispatch_async(queue, ^{
        for (int i = 0; i < 5; i++) {
            [self saleTicket];
        }
    });
    
    dispatch_async(queue, ^{
        for (int i = 0; i < 5; i++) {
            [self saleTicket];
        }
    });
    
    dispatch_async(queue, ^{
        for (int i = 0; i < 5; i++) {
            [self saleTicket];
        }
    });
}
@end

os_unfair_lock

  • os_unfair_lock用于取代不安全的OSSpinLock ，從iOS10開始才支持
  • 從底層調(diào)用看，等待os_unfair_lock鎖的線程會處于休眠狀態(tài)，并非忙等
  • 需要導入頭文件#import <os/lock.h>
@interface os_unfair_locDemo()
@property (nonatomic,assign) os_unfair_lock  moneyLock;
@property (nonatomic,assign) os_unfair_lock  ticketLock;

@end


@implementation os_unfair_locDemo

- (instancetype)init
{
    self = [super init];
    if (self) {
        self.moneyLock = OS_UNFAIR_LOCK_INIT;
        self.ticketLock = OS_UNFAIR_LOCK_INIT;
    }
    return self;
}

- (void)saleTicket
{
    os_unfair_lock_lock(&_ticketLock);
    [super saleTicket];
    os_unfair_lock_unlock(&_ticketLock);
}

- (void)saveMoeny
{
    os_unfair_lock_lock(&_moneyLock);
    [super saveMoeny];
    os_unfair_lock_unlock(&_moneyLock);
}

- (void)drawMoney
{
    os_unfair_lock_lock(&_moneyLock);
    [super drawMoney];
    os_unfair_lock_unlock(&_moneyLock);
}
@end

pthread_mutex

  • mutex叫做”互斥鎖”，等待鎖的線程會處于休眠狀態(tài)
  • 需要導入頭文件#import <pthread.h>

@interface mutexDemo()

@property (nonatomic,assign) pthread_mutex_t moneyMutex;
@property (nonatomic,assign) pthread_mutex_t ticketMutex;

@end


@implementation mutexDemo

- (instancetype)init
{
    self = [super init];
    if (self) {
        
        [self __initMutex:&_moneyMutex];
        [self __initMutex:&_ticketMutex];

    }
    return self;
}

//這里必須要傳指針 否則傳進來都是同一個pthread_mutex_t
- (void)__initMutex:(pthread_mutex_t *)mutex
{
    //定義屬性
    pthread_mutexattr_t attr;
    //初始化屬性
    pthread_mutexattr_init(&attr);
    //設置屬性
    /*
     #define PTHREAD_MUTEX_NORMAL        0
     #define PTHREAD_MUTEX_ERRORCHECK    1
     #define PTHREAD_MUTEX_RECURSIVE        2 //遞歸鎖
     #define PTHREAD_MUTEX_DEFAULT        PTHREAD_MUTEX_NORMAL
     */
    pthread_mutexattr_settype(&attr, PTHREAD_MUTEX_DEFAULT);
    pthread_mutex_init(mutex, &attr);
    //銷毀屬性
    pthread_mutexattr_destroy(&attr);
    
}

- (void)saleTicket
{
    pthread_mutex_lock(&_ticketMutex);
    [super saleTicket];
    pthread_mutex_unlock(&_ticketMutex);
    
}

- (void)saveMoeny
{
    pthread_mutex_lock(&_moneyMutex);
    [super saveMoeny];
    pthread_mutex_unlock(&_moneyMutex);
}

- (void)drawMoney
{
    pthread_mutex_lock(&_moneyMutex);
    [super drawMoney];
    pthread_mutex_unlock(&_moneyMutex);
}
- (void)dealloc
{
    pthread_mutex_destroy(&_moneyMutex);
    pthread_mutex_destroy(&_ticketMutex);
    
}
@end

遞歸鎖

- (instancetype)init
{
    if (self = [super init]) {
        [self __initMutex:&_mutex];
    }
    return self;
}

/**
 線程1：otherTest（+-）
        otherTest（+-）
         otherTest（+-）
 
 線程2：otherTest（等待）
 */

- (void)otherTest
{
    pthread_mutex_lock(&_mutex);
    
    NSLog(@"%s", __func__);
    
    static int count = 0;
    if (count < 10) {
        count++;
        [self otherTest];
    }
    
    pthread_mutex_unlock(&_mutex);
}

NS的鎖基本上是上述鎖的封裝.

dispatch_semaphore_t

  • semaphore叫做”信號量”
  • 信號量的初始值，可以用來控制線程并發(fā)訪問的最大數(shù)量
  • 信號量的初始值為1，代表同時只允許1條線程訪問資源，保證線程同步

@interface semaphore()

@property (nonatomic,strong) dispatch_semaphore_t semaphore;

@end

@implementation semaphore

- (instancetype)init
{
    self = [super init];
    if (self) {
        self.semaphore = dispatch_semaphore_create(5);
    }
    return self;
}

- (void)otherTest
{
    for (int i = 0; i<100; i++) {
        [[[NSThread alloc]initWithTarget:self selector:@selector(test) object:nil] start];
    }
}

- (void)test
{
    dispatch_semaphore_wait(self.semaphore, DISPATCH_TIME_FOREVER);
    sleep(2);
    NSLog(@"12345");
    
    dispatch_semaphore_signal(self.semaphore);
}


@end

synchronized

  • @synchronized是對mutex遞歸鎖的封裝
  • 源碼查看：objc4中的objc-sync.mm文件
  • @synchronized(obj)內(nèi)部會生成obj對應的遞歸鎖，然后進行加鎖、解鎖操作

什么情況使用自旋鎖比較劃算？

- 預計線程等待鎖的時間很短
- 加鎖的代碼（臨界區(qū)）經(jīng)常被調(diào)用，但競爭情況很少發(fā)生
- CPU資源不緊張
- 多核處理器

什么情況使用互斥鎖比較劃算？

- 預計線程等待鎖的時間較長
- 單核處理器
- 臨界區(qū)有IO操作
- 臨界區(qū)代碼復雜或者循環(huán)量大
- 臨界區(qū)競爭非常激烈
?著作權歸作者所有,轉(zhuǎn)載或內(nèi)容合作請聯(lián)系作者
【社區(qū)內(nèi)容提示】社區(qū)部分內(nèi)容疑似由AI輔助生成，瀏覽時請結(jié)合常識與多方信息審慎甄別。
平臺聲明：文章內(nèi)容（如有圖片或視頻亦包括在內(nèi)）由作者上傳并發(fā)布，文章內(nèi)容僅代表作者本人觀點，簡書系信息發(fā)布平臺，僅提供信息存儲服務。

相關閱讀更多精彩內(nèi)容

  • 一、簡介：多線程在之前進行過一篇詳細的基礎博客 iOS多線程 二、多線程的基礎知識回顧 1.1、iOS中的常見多線...
    IIronMan閱讀 1,026評論 0贊 4
  • 基本認識 在計算機的發(fā)展長河中，為了解決充分能讓 CPU 得到利用，出現(xiàn)了多線程的概念，其目的就是為了提高 CPU...
    valentizx閱讀 628評論 5贊 4
  • 目錄：1.為什么要線程安全2.多線程安全隱患分析3.多線程安全隱患的解決方案4.鎖的分類-13種鎖4.1.1OSS...
    二斤寂寞閱讀 1,243評論 0贊 3
  • 線程安全是怎么產(chǎn)生的 常見比如線程內(nèi)操作了一個線程外的非線程安全變量，這個時候一定要考慮線程安全和同步。 - (v...
    幽城88閱讀 771評論 0贊 0
  • demo下載 建議一邊看文章，一邊看代碼。 聲明：關于性能的分析是基于我的測試代碼來的，我也看到和網(wǎng)上很多測試結(jié)果...
    炸街程序猿閱讀 852評論 0贊 2

友情鏈接更多精彩內(nèi)容