人海原始碼系列 Reentrantlock

今天講reentrantlock原始碼,講解的思路是先自己手寫最簡單lock,然後一步一步增加功能,然後再去看原始碼

手寫版本1,最基礎的lock,搶不到鎖就自旋

手寫版本2,搶不到鎖就把自己放入佇列並且阻塞自己

先寫基礎部分,獲取unsafe類和stateoffset,這個主要是也用於cas操作

private final static unsafe unsafe = unsafeutils.getunsafe();//1. 獲取unsafe類
private final static long stateoffset; //2. 定義偏移量
static catch (nosuchfieldexception e) 
}

接著寫lock方法和unlock方法,並且定義乙個int型別的變數state當做鎖

private int state = 0;    
public void lock() catch (interruptedexception e) }}
public void unlock()

這樣,乙個最最簡單的lock就完成了,讓我們測試一下

public static void main(string args) throws interruptedexception 
});thread t2 = new thread(() -> 
});t1.start();
t2.start();
t1.join();
t2.join();
system.out.println(k);
}

輸出結果

connected to the target vm, address: '127.0.0.1:65421', transport: 'socket' 112277 disconnected from the target vm, address: '127.0.0.1:65421', transport: 'socket' process finished with exit code 0

然後再看加上lock的測試結果

public static void main(string args) throws interruptedexception 
});thread t2 = new thread(() -> 
});t1.start();
t2.start();
t1.join();
t2.join();
system.out.println(k);
}

測試結果

connected to the target vm, address: '127.0.0.1:59131', transport: 'socket' 200000 disconnected from the target vm, address: '127.0.0.1:59131', transport: 'socket' process finished with exit code 0

測試結果證明我們的lock生效了,但是這個版本的lock存在乙個問題,就是當搶不到鎖的時候會進入自旋,如果搶到鎖的任務耗時太長,那麼就會浪費很多cpu資源,所以我們要增加乙個佇列用於存放搶不到鎖的執行緒,並且用locksupport.park()方法阻塞它,然後解鎖的時候去喚醒佇列中的執行緒,這樣在獲取鎖線程執行時間再長也不會浪費資源,因為搶不到鎖的執行緒是在佇列中阻塞著的.

lock方法的改動

unlock方法的改動

整體思路:增加乙個變數存放持有鎖的執行緒,當持有鎖的執行緒獲取鎖時直接把state+1,持有鎖的執行緒要釋放鎖的時候就把state-1

private int state = 0;
private linkedblockingdequewaiters = new linkedblockingdeque<>();
//持有鎖的執行緒
private thread holdthread;
public void lock() 
locksupport.park(); // 搶鎖失敗,阻塞自己,等待被喚醒}}
}public boolean trylock() 
if (unsafe.compareandswapint(this, stateoffset, 0, 1)) 
return false;
}public void unlock() 
}

測試**

public static void main(string args) throws interruptedexception  catch (interruptedexception e) 
lock.unlock();
logger.info("解鎖");
});thread t2 = new thread(() -> );
t1.start();
t2.start();
t1.join();
t2.join();
}

測試結果,執行緒1是等執行緒0釋放兩次鎖才搶到鎖的,釋放第二次鎖之前執行緒阻塞1秒鐘特意給執行緒1留出時間,然而那時候執行緒0還沒釋放完畢,還持有一把鎖,所以即使阻塞一秒鐘也沒辦法搶到鎖

connected to the target vm, address: '127.0.0.1:51910', transport: 'socket' 12:37:48.362 [thread-0] info com.lz.demo.concurrent.renhai.renhailockv2reentrant - 搶到鎖 12:37:48.366 [thread-0] info com.lz.demo.concurrent.renhai.renhailockv2reentrant - 搶到鎖 12:37:48.366 [thread-0] info com.lz.demo.concurrent.renhai.renhailockv2reentrant - 解鎖 12:37:49.369 [thread-0] info com.lz.demo.concurrent.renhai.renhailockv2reentrant - 解鎖 12:37:49.369 [thread-1] info com.lz.demo.concurrent.renhai.renhailockv2reentrant - 搶到鎖 0disconnected from the target vm, address: '127.0.0.1:51910', transport: 'socket' process finished with exit code 0

人海原始碼系列 Reentrantlock

HashMap原始碼系列

Eureka原始碼系列 2 原始碼啟動入口

Redis原始碼分析系列

人海原始碼系列 Reentrantlock

HashMap原始碼系列

Eureka原始碼系列 2 原始碼啟動入口

Redis原始碼分析系列

相關推薦