Java。UnSafe类

此片文章为网上资料搜集:

看到END的地方应该就懂了,java受jvm控制,Java的线程模型与系统线程是多对多关系,Java中看到的线程是个线程的壳子,当我们执行thread的中断方法时,是设置了一个标志位,但这个标志位有jvm轮训,并抛一场,其实这样想想看在android中,有个消息循环机制,其中主线程有自己的消息循环,但是怎么又找不到这个线程,那个时候不了解控制主线程消息轮训的线程是由jvm控制的,现在想明白了。

java不能直接访问操作系统底层,而是通过本地方法来访问。Unsafe类提供了硬件级别的原子操作,主要提供了以下功能:

1、通过Unsafe类可以分配内存,可以释放内存;

类中提供的3个本地方法allocateMemory、reallocateMemory、freeMemory分别用于分配内存,扩充内存和释放内存,与C语言中的3个方法对应。

2、可以定位对象某字段的内存位置,也可以修改对象的字段值,即使它是私有的;

[java]  view plain copy print ?
  1. public native long allocateMemory(long l);  
  2. public native long reallocateMemory(long l, long l1);  
  3. public native void freeMemory(long l);  
字段的定位:
JAVA中对象的字段的定位可能通过staticFieldOffset方法实现,该方法返回给定field的内存地址偏移量,这个值对于给定的filed是唯一的且是固定不变的。
getIntVolatile方法获取对象中offset偏移地址对应的整型field的值,支持volatile load语义。
getLong方法获取对象中offset偏移地址对应的long型field的值
数组元素定位:
Unsafe类中有很多以BASE_OFFSET结尾的常量,比如ARRAY_INT_BASE_OFFSET,ARRAY_BYTE_BASE_OFFSET等,这些常量值是通过arrayBaseOffset方法得到的。arrayBaseOffset方法是一个本地方法,可以获取数组第一个元素的偏移地址。Unsafe类中还有很多以INDEX_SCALE结尾的常量,比如 ARRAY_INT_INDEX_SCALE , ARRAY_BYTE_INDEX_SCALE等,这些常量值是通过arrayIndexScale方法得到的。arrayIndexScale方法也是一个本地方法,可以获取数组的转换因子,也就是数组中元素的增量地址。将arrayBaseOffset与arrayIndexScale配合使用,可以定位数组中每个元素在内存中的位置。
[java]  view plain copy print ?
  1. public final class Unsafe {  
  2.     public static final int ARRAY_INT_BASE_OFFSET;  
  3.     public static final int ARRAY_INT_INDEX_SCALE;  
  4.   
  5.     public native long staticFieldOffset(Field field);  
  6.     public native int getIntVolatile(Object obj, long l);  
  7.     public native long getLong(Object obj, long l);  
  8.     public native int arrayBaseOffset(Class class1);  
  9.     public native int arrayIndexScale(Class class1);  
  10.   
  11.     static   
  12.     {  
  13.         ARRAY_INT_BASE_OFFSET = theUnsafe.arrayBaseOffset([I);  
  14.         ARRAY_INT_INDEX_SCALE = theUnsafe.arrayIndexScale([I);  
  15.     }  
  16. }  

3、挂起与恢复

将一个线程进行挂起是通过park方法实现的,调用 park后,线程将一直阻塞直到超时或者中断等条件出现。unpark可以终止一个挂起的线程,使其恢复正常。整个并发框架中对线程的挂起操作被封装在 LockSupport类中,LockSupport类中有各种版本pack方法,但最终都调用了Unsafe.park()方法。

package java.util.concurrent.locks;
import java.util.concurrent.*;
import sun.misc.Unsafe;
 * <p><b>Sample Usage.</b> Here is a sketch of a first-in-first-out
 * non-reentrant lock class:
 * <pre>{@code
 * <span style="color:#ff0000;">class FIFOMutex {
 *   private final AtomicBoolean locked = new AtomicBoolean(false);
 *   private final Queue<Thread> waiters
 *     = new ConcurrentLinkedQueue<Thread>();
 *
 *   public void lock() {
 *     boolean wasInterrupted = false;
 *     Thread current = Thread.currentThread();
 *     waiters.add(current);
 *
 *     // Block while not first in queue or cannot acquire lock
 *     while (waiters.peek() != current ||
 *            !locked.compareAndSet(false, true)) {
 *        LockSupport.park(this);
 *        if (Thread.interrupted()) // ignore interrupts while waiting
 *          wasInterrupted = true;
 *     }
 *
 *     waiters.remove();
 *     if (wasInterrupted)          // reassert interrupt status on exit
 *        current.interrupt();
 *   }
 *
 *   public void unlock() {
 *     locked.set(false);
 *     LockSupport.unpark(waiters.peek());
 *   }
 * }}</pre>
 */</span>

public class LockSupport {
    private LockSupport() {} // Cannot be instantiated.

    // Hotspot implementation via intrinsics API
    private static final Unsafe unsafe = Unsafe.getUnsafe();
    private static final long parkBlockerOffset;

    static {
        try {
            parkBlockerOffset = unsafe.objectFieldOffset
                (java.lang.Thread.class.getDeclaredField("parkBlocker"));
        } catch (Exception ex) { throw new Error(ex); }
    }

    private static void setBlocker(Thread t, Object arg) {
        // Even though volatile, hotspot doesn't need a write barrier here.
        unsafe.putObject(t, parkBlockerOffset, arg);
    }

    /**
     * Makes available the permit for the given thread, if it
     * was not already available.  If the thread was blocked on
     * {@code park} then it will unblock.  Otherwise, its next call
     * to {@code park} is guaranteed not to block. This operation
     * is not guaranteed to have any effect at all if the given
     * thread has not been started.
     *
     * @param thread the thread to unpark, or {@code null}, in which case
     *        this operation has no effect
     */
    public static void unpark(Thread thread) {
        if (thread != null)
            unsafe.unpark(thread);
    }

    /**
     * Disables the current thread for thread scheduling purposes unless the
     * permit is available.
     *
     * <p>If the permit is available then it is consumed and the call returns
     * immediately; otherwise
     * the current thread becomes disabled for thread scheduling
     * purposes and lies dormant until one of three things happens:
     *
     * <ul>
     * <li>Some other thread invokes {@link #unpark unpark} with the
     * current thread as the target; or
     *
     * <li>Some other thread {@linkplain Thread#interrupt interrupts}
     * the current thread; or
     *
     * <li>The call spuriously (that is, for no reason) returns.
     * </ul>
     *
     * <p>This method does <em>not</em> report which of these caused the
     * method to return. Callers should re-check the conditions which caused
     * the thread to park in the first place. Callers may also determine,
     * for example, the interrupt status of the thread upon return.
     *
     * @param blocker the synchronization object responsible for this
     *        thread parking
     * @since 1.6
     */
    public static void park(Object blocker) {
        Thread t = Thread.currentThread();
        setBlocker(t, blocker);
        unsafe.park(false, 0L);
        setBlocker(t, null);
    }

    /**
     * Disables the current thread for thread scheduling purposes, for up to
     * the specified waiting time, unless the permit is available.
     *
     * <p>If the permit is available then it is consumed and the call
     * returns immediately; otherwise the current thread becomes disabled
     * for thread scheduling purposes and lies dormant until one of four
     * things happens:
     *
     * <ul>
     * <li>Some other thread invokes {@link #unpark unpark} with the
     * current thread as the target; or
     *
     * <li>Some other thread {@linkplain Thread#interrupt interrupts}
     * the current thread; or
     *
     * <li>The specified waiting time elapses; or
     *
     * <li>The call spuriously (that is, for no reason) returns.
     * </ul>
     *
     * <p>This method does <em>not</em> report which of these caused the
     * method to return. Callers should re-check the conditions which caused
     * the thread to park in the first place. Callers may also determine,
     * for example, the interrupt status of the thread, or the elapsed time
     * upon return.
     *
     * @param blocker the synchronization object responsible for this
     *        thread parking
     * @param nanos the maximum number of nanoseconds to wait
     * @since 1.6
     */
    public static void parkNanos(Object blocker, long nanos) {
        if (nanos > 0) {
            Thread t = Thread.currentThread();
            setBlocker(t, blocker);
            unsafe.park(false, nanos);
            setBlocker(t, null);
        }
    }

    /**
     * Disables the current thread for thread scheduling purposes, until
     * the specified deadline, unless the permit is available.
     *
     * <p>If the permit is available then it is consumed and the call
     * returns immediately; otherwise the current thread becomes disabled
     * for thread scheduling purposes and lies dormant until one of four
     * things happens:
     *
     * <ul>
     * <li>Some other thread invokes {@link #unpark unpark} with the
     * current thread as the target; or
     *
     * <li>Some other thread {@linkplain Thread#interrupt interrupts} the
     * current thread; or
     *
     * <li>The specified deadline passes; or
     *
     * <li>The call spuriously (that is, for no reason) returns.
     * </ul>
     *
     * <p>This method does <em>not</em> report which of these caused the
     * method to return. Callers should re-check the conditions which caused
     * the thread to park in the first place. Callers may also determine,
     * for example, the interrupt status of the thread, or the current time
     * upon return.
     *
     * @param blocker the synchronization object responsible for this
     *        thread parking
     * @param deadline the absolute time, in milliseconds from the Epoch,
     *        to wait until
     * @since 1.6
     */
    public static void parkUntil(Object blocker, long deadline) {
        Thread t = Thread.currentThread();
        setBlocker(t, blocker);
        unsafe.park(true, deadline);
        setBlocker(t, null);
    }

    /**
     * Returns the blocker object supplied to the most recent
     * invocation of a park method that has not yet unblocked, or null
     * if not blocked.  The value returned is just a momentary
     * snapshot -- the thread may have since unblocked or blocked on a
     * different blocker object.
     *
     * @param t the thread
     * @return the blocker
     * @throws NullPointerException if argument is null
     * @since 1.6
     */
    public static Object getBlocker(Thread t) {
        if (t == null)
            throw new NullPointerException();
        return unsafe.getObjectVolatile(t, parkBlockerOffset);
    }

    /**
     * Disables the current thread for thread scheduling purposes unless the
     * permit is available.
     *
     * <p>If the permit is available then it is consumed and the call
     * returns immediately; otherwise the current thread becomes disabled
     * for thread scheduling purposes and lies dormant until one of three
     * things happens:
     *
     * <ul>
     *
     * <li>Some other thread invokes {@link #unpark unpark} with the
     * current thread as the target; or
     *
     * <li>Some other thread {@linkplain Thread#interrupt interrupts}
     * the current thread; or
     *
     * <li>The call spuriously (that is, for no reason) returns.
     * </ul>
     *
     * <p>This method does <em>not</em> report which of these caused the
     * method to return. Callers should re-check the conditions which caused
     * the thread to park in the first place. Callers may also determine,
     * for example, the interrupt status of the thread upon return.
     */
    public static void park() {
        unsafe.park(false, 0L);
    }

    /**
     * Disables the current thread for thread scheduling purposes, for up to
     * the specified waiting time, unless the permit is available.
     *
     * <p>If the permit is available then it is consumed and the call
     * returns immediately; otherwise the current thread becomes disabled
     * for thread scheduling purposes and lies dormant until one of four
     * things happens:
     *
     * <ul>
     * <li>Some other thread invokes {@link #unpark unpark} with the
     * current thread as the target; or
     *
     * <li>Some other thread {@linkplain Thread#interrupt interrupts}
     * the current thread; or
     *
     * <li>The specified waiting time elapses; or
     *
     * <li>The call spuriously (that is, for no reason) returns.
     * </ul>
     *
     * <p>This method does <em>not</em> report which of these caused the
     * method to return. Callers should re-check the conditions which caused
     * the thread to park in the first place. Callers may also determine,
     * for example, the interrupt status of the thread, or the elapsed time
     * upon return.
     *
     * @param nanos the maximum number of nanoseconds to wait
     */
    public static void parkNanos(long nanos) {
        if (nanos > 0)
            unsafe.park(false, nanos);
    }

    /**
     * Disables the current thread for thread scheduling purposes, until
     * the specified deadline, unless the permit is available.
     *
     * <p>If the permit is available then it is consumed and the call
     * returns immediately; otherwise the current thread becomes disabled
     * for thread scheduling purposes and lies dormant until one of four
     * things happens:
     *
     * <ul>
     * <li>Some other thread invokes {@link #unpark unpark} with the
     * current thread as the target; or
     *
     * <li>Some other thread {@linkplain Thread#interrupt interrupts}
     * the current thread; or
     *
     * <li>The specified deadline passes; or
     *
     * <li>The call spuriously (that is, for no reason) returns.
     * </ul>
     *
     * <p>This method does <em>not</em> report which of these caused the
     * method to return. Callers should re-check the conditions which caused
     * the thread to park in the first place. Callers may also determine,
     * for example, the interrupt status of the thread, or the current time
     * upon return.
     *
     * @param deadline the absolute time, in milliseconds from the Epoch,
     *        to wait until
     */
    public static void parkUntil(long deadline) {
        unsafe.park(true, deadline);
    }
}

对应的jni的实现方式:
Java不能直接访问操作系统底层,而是通过本地方法来访问,Unsafe提供了硬件级别的原子操作,提供了以下功能:
分配、释放内存

操作内存主要依靠下面三个方法:
allocateMemory:分配内存;
reallocateMemory:扩充内存;
freeMemory:释放内存;
操作对象的字段

  Java对象中字段的定位可以通过staticFieldOffset实现,而读取具体便宜位置的字段的值可以使用getLong(根据不同类型选择不同的函数)来完成,可以使用putLong(根据不同的类型选择不同的函数)来设置值,可以使用arrayBaseOffset(获取数组第一个元素的偏移地址)和arrayIndexScale(获取数组中元素增量的地址)来访问数组中的元素。
CAS操作

CAS操作(会在java.util.concurrent中大量地使用)包括三个操作数:
内存位置;
预期原值;
新值;
如果内存位置的值与预期的值相等,那么处理器会将该位置的值设置为新值,否则,处理器不做任何处理。int类型的CAS实现如下:
复制代码
static inline bool compareAndSwap (volatile jint *addr, jint old, jint new_val)
{
    jboolean result = false;
    spinlock lock;
    if ((result = (*addr == old)))
        *addr = new_val;
    return result;
}
复制代码
这段代码看起来根本无法保证操作的原子性,而其中的关键就在于spinlock lock这段看似没有用的代码,这时会调用构造函数,其中不停检查lock的值是否为0,如果不是0则线程让步等到线程下次执行时再次检查;否则,将lock设置为1。在离开函数的时候会调用其析构函数,将lock的值再次设置为0。spinlock的代码如下:
复制代码
class spinlock
{
    static volatile obj_addr_t lock;
public:
    spinlock ()
    {
        while (!compare_and_swap(&lock, 0, 1))
            _Jv_ThreadYield();
    }
    ~spinlock ()
    {
        release_set(&lock, 0);
    }
};
复制代码
而且,不管是在多核还是在单核的情况下,总是需要一些特殊的手段才能保证CAS操作成功吧,比如锁总线,compare_and_swap的实现如下:
复制代码
<span style="color:#ff0000;">inline static bool compare_and_swap(volatile obj_addr_t *addr, obj_addr_t old, obj_addr_t new_val)
{
    char result;
    #ifdef __x86_64__ __asm__ __volatile__("lock; cmpxchgq %2, %0; setz %1"
        : "=m"(*(addr)), "=q"(result)
        : "r" (new_val), "a"(old), "m"(*addr)
        : "memory");
    #else
        __asm__ __volatile__("lock; cmpxchgl %2, %0; setz %1"
        : "=m"(*addr), "=q"(result)
        : "r" (new_val), "a"(old), "m"(*addr)
        : "memory");
    #endif
    return (bool) result;
}
复制代码
挂起和恢复

使用park方法将一个线程挂起,知道超时或者中断等条件出现,实现如下:
复制代码
void sun::misc::Unsafe::park(jboolean isAbsolute, jlong time)
{
    using namespace ::java::lang;
    Thread *thread = Thread::currentThread();
    natThread *nt = (natThread *) thread->data;
    nt->park_helper.park(isAbsolute, time);
}
复制代码
 使用unpark来终止一个挂起的线程,使其恢复正常,实现如下:
sun::misc::Unsafe::unpark (::java::lang::Thread *thread)
{
    natThread *nt = (natThread *) thread->data;
    nt->park_helper.unpark ();
}
----- -- -
END</span>



4、CAS操作
是通过compareAndSwapXXX方法实现的
[java]  view plain copy print ?
  1. /** 
  2. * 比较obj的offset处内存位置中的值和期望的值,如果相同则更新。此更新是不可中断的。 
  3.  
  4. * @param obj 需要更新的对象 
  5. * @param offset obj中整型field的偏移量 
  6. * @param expect 希望field中存在的值 
  7. * @param update 如果期望值expect与field的当前值相同,设置filed的值为这个新值 
  8. * @return 如果field的值被更改返回true 
  9. */  
  10. public native boolean compareAndSwapInt(Object obj, long offset, int expect, int update);  
CAS操作有3个操作数,内存值M,预期值E,新值U,如果M==E,则将内存值修改为B,否则啥都不做。

参考资料:
Unsafe.h

源码剖析之sun.misc.Unsafe

   首先介绍一下什么是Compare And Swap(CAS)?简单的说就是比较并交换。

CAS 操作包含三个操作数 —— 内存位置(V)、预期原值(A)和新值(B)。如果内存位置的值与预期原值相匹配,那么处理器会自动将该位置值更新为新值。否则,处理器不做任何操作。无论哪种情况,它都会在 CAS 指令之前返回该位置的值。CAS 有效地说明了“我认为位置 V 应该包含值 A;如果包含该值,则将 B 放到这个位置;否则,不要更改该位置,只告诉我这个位置现在的值即可。” Java并发包(java.util.concurrent)中大量使用了CAS操作,涉及到并发的地方都调用了sun.misc.Unsafe类方法进行CAS操作。

在看一下volatile, Volatile修饰的成员变量在每次被线程访问时,都强迫从共享内存中重读该成员变量的值。而且,当成员变量发生变化时,强迫线程将变化值回写到共享内存。这样在任何时刻,两个不同的线程总是看到某个成员变量的值是相同的,更简单一点理解就是volatile修饰的变量值发生变化时对于另外的线程是可见的。

如何正确使用volatile可以参考下面这篇文章:

http://www.ibm.com/developerworks/cn/java/j-jtp06197.html Java 理论与实践: 正确使用 Volatile 变量

下面来看看java中具体的CAS操作类sun.misc.Unsafe。Unsafe类提供了硬件级别的原子操作,Java无法直接访问到操作系统底层(如系统硬件等),为此Java使用native方法来扩展Java程序的功能。具体实现使用c++,详见文件sun.misc.natUnsafe.cc();sun.misc包的源代码可以在这里找到:

 http://www.oschina.net/code/explore/gcc-4.5.2/libjava/sun/misc

//下面是sun.misc.Unsafe.java类源码

package sun.misc;

import java.lang.reflect.Field;

/***
 * This class should provide access to low-level operations and its
 * use should be limited to trusted code.  Fields can be accessed using
 * memory addresses, with undefined behaviour occurring if invalid memory
 * addresses are given.
 * 这个类提供了一个更底层的操作并且应该在受信任的代码中使用。可以通过内存地址
 * 存取fields,如果给出的内存地址是无效的那么会有一个不确定的运行表现。
 * 
 * @author Tom Tromey ([email protected])
 * @author Andrew John Hughes ([email protected])
 */
public class Unsafe
{
  // Singleton class.
  private static Unsafe unsafe = new Unsafe();

  /***
   * Private default constructor to prevent creation of an arbitrary
   * number of instances.
   * 使用私有默认构造器防止创建多个实例
   */
  private Unsafe()
  {
  }

  /***
   * Retrieve the singleton instance of <code>Unsafe</code>.  The calling
   * method should guard this instance from untrusted code, as it provides
   * access to low-level operations such as direct memory access.
   * 获取<code>Unsafe</code>的单例,这个方法调用应该防止在不可信的代码中实例,
   * 因为unsafe类提供了一个低级别的操作,例如直接内存存取。
   * 
   * @throws SecurityException if a security manager exists and prevents
   *                           access to the system properties.
   *                           如果安全管理器不存在或者禁止访问系统属性
   */
  public static Unsafe getUnsafe()
  {
    SecurityManager sm = System.getSecurityManager();
    if (sm != null)
      sm.checkPropertiesAccess();
    return unsafe;
  }
  
  /***
   * Returns the memory address offset of the given static field.
   * The offset is merely used as a means to access a particular field
   * in the other methods of this class.  The value is unique to the given
   * field and the same value should be returned on each subsequent call.
   * 返回指定静态field的内存地址偏移量,在这个类的其他方法中这个值只是被用作一个访问
   * 特定field的一个方式。这个值对于 给定的field是唯一的,并且后续对该方法的调用都应该
   * 返回相同的值。
   *
   * @param field the field whose offset should be returned.
   *              需要返回偏移量的field
   * @return the offset of the given field.
   *         指定field的偏移量
   */
  public native long objectFieldOffset(Field field);

  /***
   * Compares the value of the integer field at the specified offset
   * in the supplied object with the given expected value, and updates
   * it if they match.  The operation of this method should be atomic,
   * thus providing an uninterruptible way of updating an integer field.
   * 在obj的offset位置比较integer field和期望的值,如果相同则更新。这个方法
   * 的操作应该是原子的,因此提供了一种不可中断的方式更新integer field。
   * 
   * @param obj the object containing the field to modify.
   *            包含要修改field的对象
   * @param offset the offset of the integer field within <code>obj</code>.
   *               <code>obj</code>中整型field的偏移量
   * @param expect the expected value of the field.
   *               希望field中存在的值
   * @param update the new value of the field if it equals <code>expect</code>.
   *           如果期望值expect与field的当前值相同,设置filed的值为这个新值
   * @return true if the field was changed.
   *                             如果field的值被更改
   */
  public native boolean compareAndSwapInt(Object obj, long offset,
                                          int expect, int update);

  /***
   * Compares the value of the long field at the specified offset
   * in the supplied object with the given expected value, and updates
   * it if they match.  The operation of this method should be atomic,
   * thus providing an uninterruptible way of updating a long field.
   * 在obj的offset位置比较long field和期望的值,如果相同则更新。这个方法
   * 的操作应该是原子的,因此提供了一种不可中断的方式更新long field。
   * 
   * @param obj the object containing the field to modify.
   *              包含要修改field的对象 
   * @param offset the offset of the long field within <code>obj</code>.
   *               <code>obj</code>中long型field的偏移量
   * @param expect the expected value of the field.
   *               希望field中存在的值
   * @param update the new value of the field if it equals <code>expect</code>.
   *               如果期望值expect与field的当前值相同,设置filed的值为这个新值
   * @return true if the field was changed.
   *              如果field的值被更改
   */
  public native boolean compareAndSwapLong(Object obj, long offset,
                                           long expect, long update);

  /***
   * Compares the value of the object field at the specified offset
   * in the supplied object with the given expected value, and updates
   * it if they match.  The operation of this method should be atomic,
   * thus providing an uninterruptible way of updating an object field.
   * 在obj的offset位置比较object field和期望的值,如果相同则更新。这个方法
   * 的操作应该是原子的,因此提供了一种不可中断的方式更新object field。
   * 
   * @param obj the object containing the field to modify.
   *    包含要修改field的对象 
   * @param offset the offset of the object field within <code>obj</code>.
   *         <code>obj</code>中object型field的偏移量
   * @param expect the expected value of the field.
   *               希望field中存在的值
   * @param update the new value of the field if it equals <code>expect</code>.
   *               如果期望值expect与field的当前值相同,设置filed的值为这个新值
   * @return true if the field was changed.
   *              如果field的值被更改
   */
  public native boolean compareAndSwapObject(Object obj, long offset,
                                             Object expect, Object update);

  /***
   * Sets the value of the integer field at the specified offset in the
   * supplied object to the given value.  This is an ordered or lazy
   * version of <code>putIntVolatile(Object,long,int)</code>, which
   * doesn't guarantee the immediate visibility of the change to other
   * threads.  It is only really useful where the integer field is
   * <code>volatile</code>, and is thus expected to change unexpectedly.
   * 设置obj对象中offset偏移地址对应的整型field的值为指定值。这是一个有序或者
   * 有延迟的<code>putIntVolatile</cdoe>方法,并且不保证值的改变被其他线程立
   * 即看到。只有在field被<code>volatile</code>修饰并且期望被意外修改的时候
   * 使用才有用。
   * 
   * @param obj the object containing the field to modify.
   *    包含需要修改field的对象
   * @param offset the offset of the integer field within <code>obj</code>.
   *       <code>obj</code>中整型field的偏移量
   * @param value the new value of the field.
   *      field将被设置的新值
   * @see #putIntVolatile(Object,long,int)
   */
  public native void putOrderedInt(Object obj, long offset, int value);

  /***
   * Sets the value of the long field at the specified offset in the
   * supplied object to the given value.  This is an ordered or lazy
   * version of <code>putLongVolatile(Object,long,long)</code>, which
   * doesn't guarantee the immediate visibility of the change to other
   * threads.  It is only really useful where the long field is
   * <code>volatile</code>, and is thus expected to change unexpectedly.
   * 设置obj对象中offset偏移地址对应的long型field的值为指定值。这是一个有序或者
   * 有延迟的<code>putLongVolatile</cdoe>方法,并且不保证值的改变被其他线程立
   * 即看到。只有在field被<code>volatile</code>修饰并且期望被意外修改的时候
   * 使用才有用。
   * 
   * @param obj the object containing the field to modify.
   *    包含需要修改field的对象
   * @param offset the offset of the long field within <code>obj</code>.
   *       <code>obj</code>中long型field的偏移量
   * @param value the new value of the field.
   *      field将被设置的新值
   * @see #putLongVolatile(Object,long,long)
   */
  public native void putOrderedLong(Object obj, long offset, long value);

  /***
   * Sets the value of the object field at the specified offset in the
   * supplied object to the given value.  This is an ordered or lazy
   * version of <code>putObjectVolatile(Object,long,Object)</code>, which
   * doesn't guarantee the immediate visibility of the change to other
   * threads.  It is only really useful where the object field is
   * <code>volatile</code>, and is thus expected to change unexpectedly.
   * 设置obj对象中offset偏移地址对应的object型field的值为指定值。这是一个有序或者
   * 有延迟的<code>putObjectVolatile</cdoe>方法,并且不保证值的改变被其他线程立
   * 即看到。只有在field被<code>volatile</code>修饰并且期望被意外修改的时候
   * 使用才有用。
   *
   * @param obj the object containing the field to modify.
   *    包含需要修改field的对象
   * @param offset the offset of the object field within <code>obj</code>.
   *       <code>obj</code>中long型field的偏移量
   * @param value the new value of the field.
   *      field将被设置的新值
   */
  public native void putOrderedObject(Object obj, long offset, Object value);

  /***
   * Sets the value of the integer field at the specified offset in the
   * supplied object to the given value, with volatile store semantics.
   * 设置obj对象中offset偏移地址对应的整型field的值为指定值。支持volatile store语义
   * 
   * @param obj the object containing the field to modify.
   *    包含需要修改field的对象
   * @param offset the offset of the integer field within <code>obj</code>.
   *       <code>obj</code>中整型field的偏移量
   * @param value the new value of the field.
   *       field将被设置的新值
   */
  public native void putIntVolatile(Object obj, long offset, int value);

  /***
   * Retrieves the value of the integer field at the specified offset in the
   * supplied object with volatile load semantics.
   * 获取obj对象中offset偏移地址对应的整型field的值,支持volatile load语义。
   * 
   * @param obj the object containing the field to read.
   *    包含需要去读取的field的对象
   * @param offset the offset of the integer field within <code>obj</code>.
   *       <code>obj</code>中整型field的偏移量
   */
  public native int getIntVolatile(Object obj, long offset);

  /***
   * Sets the value of the long field at the specified offset in the
   * supplied object to the given value, with volatile store semantics.
   * 设置obj对象中offset偏移地址对应的long型field的值为指定值。支持volatile store语义
   *
   * @param obj the object containing the field to modify.
   *            包含需要修改field的对象
   * @param offset the offset of the long field within <code>obj</code>.
   *               <code>obj</code>中long型field的偏移量
   * @param value the new value of the field.
   *              field将被设置的新值
   * @see #putLong(Object,long,long)
   */
  public native void putLongVolatile(Object obj, long offset, long value);

  /***
   * Sets the value of the long field at the specified offset in the
   * supplied object to the given value.
   * 设置obj对象中offset偏移地址对应的long型field的值为指定值。
   * 
   * @param obj the object containing the field to modify.
   *     包含需要修改field的对象
   * @param offset the offset of the long field within <code>obj</code>.
   *     <code>obj</code>中long型field的偏移量
   * @param value the new value of the field.
   *     field将被设置的新值
   * @see #putLongVolatile(Object,long,long)
   */
  public native void putLong(Object obj, long offset, long value);

  /***
   * Retrieves the value of the long field at the specified offset in the
   * supplied object with volatile load semantics.
   * 获取obj对象中offset偏移地址对应的long型field的值,支持volatile load语义。
   * 
   * @param obj the object containing the field to read.
   *    包含需要去读取的field的对象
   * @param offset the offset of the long field within <code>obj</code>.
   *       <code>obj</code>中long型field的偏移量
   * @see #getLong(Object,long)
   */
  public native long getLongVolatile(Object obj, long offset);

  /***
   * Retrieves the value of the long field at the specified offset in the
   * supplied object.
   * 获取obj对象中offset偏移地址对应的long型field的值
   * 
   * @param obj the object containing the field to read.
   *    包含需要去读取的field的对象
   * @param offset the offset of the long field within <code>obj</code>.
   *       <code>obj</code>中long型field的偏移量
   * @see #getLongVolatile(Object,long)
   */
  public native long getLong(Object obj, long offset);

  /***
   * Sets the value of the object field at the specified offset in the
   * supplied object to the given value, with volatile store semantics.
   * 设置obj对象中offset偏移地址对应的object型field的值为指定值。支持volatile store语义
   * 
   * @param obj the object containing the field to modify.
   *    包含需要修改field的对象
   * @param offset the offset of the object field within <code>obj</code>.
   *     <code>obj</code>中object型field的偏移量
   * @param value the new value of the field.
   *       field将被设置的新值
   * @see #putObject(Object,long,Object)
   */
  public native void putObjectVolatile(Object obj, long offset, Object value);

  /***
   * Sets the value of the object field at the specified offset in the
   * supplied object to the given value.
   * 设置obj对象中offset偏移地址对应的object型field的值为指定值。
   * 
   * @param obj the object containing the field to modify.
   *    包含需要修改field的对象
   * @param offset the offset of the object field within <code>obj</code>.
   *     <code>obj</code>中object型field的偏移量
   * @param value the new value of the field.
   *       field将被设置的新值
   * @see #putObjectVolatile(Object,long,Object)
   */
  public native void putObject(Object obj, long offset, Object value);

  /***
   * Retrieves the value of the object field at the specified offset in the
   * supplied object with volatile load semantics.
   * 获取obj对象中offset偏移地址对应的object型field的值,支持volatile load语义。
   * 
   * @param obj the object containing the field to read.
   *    包含需要去读取的field的对象
   * @param offset the offset of the object field within <code>obj</code>.
   *       <code>obj</code>中object型field的偏移量
   */
  public native Object getObjectVolatile(Object obj, long offset);

  /***
   * Returns the offset of the first element for a given array class.
   * To access elements of the array class, this value may be used along with
   * with that returned by 
   * <a href="#arrayIndexScale"><code>arrayIndexScale</code></a>,
   * if non-zero.
   * 获取给定数组中第一个元素的偏移地址。
   * 为了存取数组中的元素,这个偏移地址与<a href="#arrayIndexScale"><code>arrayIndexScale
   * </code></a>方法的非0返回值一起被使用。
   * @param arrayClass the class for which the first element's address should
   *                   be obtained.
   *                   第一个元素地址被获取的class
   * @return the offset of the first element of the array class.
   *    数组第一个元素 的偏移地址
   * @see arrayIndexScale(Class)
   */
  public native int arrayBaseOffset(Class arrayClass);

  /***
   * Returns the scale factor used for addressing elements of the supplied
   * array class.  Where a suitable scale factor can not be returned (e.g.
   * for primitive types), zero should be returned.  The returned value
   * can be used with 
   * <a href="#arrayBaseOffset"><code>arrayBaseOffset</code></a>
   * to access elements of the class.
   * 获取用户给定数组寻址的换算因子.一个合适的换算因子不能返回的时候(例如:基本类型),
   * 返回0.这个返回值能够与<a href="#arrayBaseOffset"><code>arrayBaseOffset</code>
   * </a>一起使用去存取这个数组class中的元素
   * 
   * @param arrayClass the class whose scale factor should be returned.
   * @return the scale factor, or zero if not supported for this array class.
   */
  public native int arrayIndexScale(Class arrayClass);
  
  /***
   * Releases the block on a thread created by 
   * <a href="#park"><code>park</code></a>.  This method can also be used
   * to terminate a blockage caused by a prior call to <code>park</code>.
   * This operation is unsafe, as the thread must be guaranteed to be
   * live.  This is true of Java, but not native code.
   * 释放被<a href="#park"><code>park</code></a>创建的在一个线程上的阻塞.这个
   * 方法也可以被使用来终止一个先前调用<code>park</code>导致的阻塞.
   * 这个操作操作时不安全的,因此线程必须保证是活的.这是java代码不是native代码。
   * @param thread the thread to unblock.
   *           要解除阻塞的线程
   */
  public native void unpark(Thread thread);

  /***
   * Blocks the thread until a matching 
   * <a href="#unpark"><code>unpark</code></a> occurs, the thread is
   * interrupted or the optional timeout expires.  If an <code>unpark</code>
   * call has already occurred, this also counts.  A timeout value of zero
   * is defined as no timeout.  When <code>isAbsolute</code> is
   * <code>true</code>, the timeout is in milliseconds relative to the
   * epoch.  Otherwise, the value is the number of nanoseconds which must
   * occur before timeout.  This call may also return spuriously (i.e.
   * for no apparent reason).
   * 阻塞一个线程直到<a href="#unpark"><code>unpark</code></a>出现、线程
   * 被中断或者timeout时间到期。如果一个<code>unpark</code>调用已经出现了,
   * 这里只计数。timeout为0表示永不过期.当<code>isAbsolute</code>为true时,
   * timeout是相对于新纪元之后的毫秒。否则这个值就是超时前的纳秒数。这个方法执行时
   * 也可能不合理地返回(没有具体原因)
   * 
   * @param isAbsolute true if the timeout is specified in milliseconds from
   *                   the epoch.
   *                   如果为true timeout的值是一个相对于新纪元之后的毫秒数
   * @param time either the number of nanoseconds to wait, or a time in
   *             milliseconds from the epoch to wait for.
   *             可以是一个要等待的纳秒数,或者是一个相对于新纪元之后的毫秒数直到
   *             到达这个时间点
   */
  public native void park(boolean isAbsolute, long time);

}

 

//下面是sun.misc.natUnsafe.cc源码

#include <gcj/cni.h>

#include <gcj/field.h>

#include <gcj/javaprims.h>

#include <jvm.h>

#include <sun/misc/Unsafe.h>

#include <java/lang/System.h>

#include <java/lang/InterruptedException.h>

 

#include <java/lang/Thread.h>

#include <java/lang/Long.h>

 

#include "sysdep/locks.h"

 

// Use a spinlock for multi-word accesses

class spinlock

{

  static volatile obj_addr_t lock;

 

public:

 

spinlock ()

  {

    while (! compare_and_swap (&lock, 0, 1))

      _Jv_ThreadYield ();

  }

  ~spinlock ()

  {

    release_set (&lock, 0);

  }

};

 

// This is a single lock that is used for all synchronized accesses if

// the compiler can't generate inline compare-and-swap operations.  In

// most cases it'll never be used, but the i386 needs it for 64-bit

// locked accesses and so does PPC32.  It's worth building libgcj with

// target=i486 (or above) to get the inlines.

volatile obj_addr_t spinlock::lock;

 

 

static inline bool

compareAndSwap (volatile jint *addr, jint old, jint new_val)

{

  jboolean result = false;

  spinlock lock;

  if ((result = (*addr == old)))

    *addr = new_val;

  return result;

}

 

static inline bool

compareAndSwap (volatile jlong *addr, jlong old, jlong new_val)

{

  jboolean result = false;

  spinlock lock;

  if ((result = (*addr == old)))

    *addr = new_val;

  return result;

}

 

static inline bool

compareAndSwap (volatile jobject *addr, jobject old, jobject new_val)

{

  jboolean result = false;

  spinlock lock;

  if ((result = (*addr == old)))

    *addr = new_val;

  return result;

}

 

 

jlong

sun::misc::Unsafe::objectFieldOffset (::java::lang::reflect::Field *field)

{

  _Jv_Field *fld = _Jv_FromReflectedField (field);

  // FIXME: what if it is not an instance field?

  return fld->getOffset();

}

 

jint

sun::misc::Unsafe::arrayBaseOffset (jclass arrayClass)

{

  // FIXME: assert that arrayClass is array.

  jclass eltClass = arrayClass->getComponentType();

  return (jint)(jlong) _Jv_GetArrayElementFromElementType (NULL, eltClass);

}

 

jint

sun::misc::Unsafe::arrayIndexScale (jclass arrayClass)

{

  // FIXME: assert that arrayClass is array.

  jclass eltClass = arrayClass->getComponentType();

  if (eltClass->isPrimitive())

    return eltClass->size();

  return sizeof (void *);

}

 

// These methods are used when the compiler fails to generate inline

// versions of the compare-and-swap primitives.

 

jboolean

sun::misc::Unsafe::compareAndSwapInt (jobject obj, jlong offset,

                                           jint expect, jint update)

{

  jint *addr = (jint *)((char *)obj + offset);

  return compareAndSwap (addr, expect, update);

}

 

jboolean

sun::misc::Unsafe::compareAndSwapLong (jobject obj, jlong offset,

                                            jlong expect, jlong update)

{

  volatile jlong *addr = (jlong*)((char *) obj + offset);

  return compareAndSwap (addr, expect, update);

}

 

jboolean

sun::misc::Unsafe::compareAndSwapObject (jobject obj, jlong offset,

                                                jobject expect, jobject update)

{

  jobject *addr = (jobject*)((char *) obj + offset);

  return compareAndSwap (addr, expect, update);

}

 

void

sun::misc::Unsafe::putOrderedInt (jobject obj, jlong offset, jint value)

{

  volatile jint *addr = (jint *) ((char *) obj + offset);

  *addr = value;

}

 

void

sun::misc::Unsafe::putOrderedLong (jobject obj, jlong offset, jlong value)

{

  volatile jlong *addr = (jlong *) ((char *) obj + offset);

  spinlock lock;

  *addr = value;

}

 

void

sun::misc::Unsafe::putOrderedObject (jobject obj, jlong offset, jobject value)

{

  volatile jobject *addr = (jobject *) ((char *) obj + offset);

  *addr = value;

}

 

void

sun::misc::Unsafe::putIntVolatile (jobject obj, jlong offset, jint value)

{

  write_barrier ();

  volatile jint *addr = (jint *) ((char *) obj + offset);

  *addr = value;

}

 

void

sun::misc::Unsafe::putLongVolatile (jobject obj, jlong offset, jlong value)

{

  volatile jlong *addr = (jlong *) ((char *) obj + offset);

  spinlock lock;

  *addr = value;

}

 

void

sun::misc::Unsafe::putObjectVolatile (jobject obj, jlong offset, jobject value)

{

  write_barrier ();

  volatile jobject *addr = (jobject *) ((char *) obj + offset);

  *addr = value;

}

 

#if 0  // FIXME

void

sun::misc::Unsafe::putInt (jobject obj, jlong offset, jint value)

{

  jint *addr = (jint *) ((char *) obj + offset);

  *addr = value;

}

#endif

 

void

sun::misc::Unsafe::putLong (jobject obj, jlong offset, jlong value)

{

  jlong *addr = (jlong *) ((char *) obj + offset);

  spinlock lock;

  *addr = value;

}

 

void

sun::misc::Unsafe::putObject (jobject obj, jlong offset, jobject value)

{

  jobject *addr = (jobject *) ((char *) obj + offset);

  *addr = value;

}

 

jint

sun::misc::Unsafe::getIntVolatile (jobject obj, jlong offset)

{

  volatile jint *addr = (jint *) ((char *) obj + offset);

  jint result = *addr;

  read_barrier ();

  return result;

}

 

jobject

sun::misc::Unsafe::getObjectVolatile (jobject obj, jlong offset)

{

  volatile jobject *addr = (jobject *) ((char *) obj + offset);

  jobject result = *addr;

  read_barrier ();

  return result;

}

 

jlong

sun::misc::Unsafe::getLong (jobject obj, jlong offset)

{

  jlong *addr = (jlong *) ((char *) obj + offset);

  spinlock lock;

  return *addr;

}

 

jlong

sun::misc::Unsafe::getLongVolatile (jobject obj, jlong offset)

{

  volatile jlong *addr = (jlong *) ((char *) obj + offset);

  spinlock lock;

  return *addr;

}

 

void

sun::misc::Unsafe::unpark (::java::lang::Thread *thread)

{

  natThread *nt = (natThread *) thread->data;

  nt->park_helper.unpark ();

}

 

void

sun::misc::Unsafe::park (jboolean isAbsolute, jlong time)

{

  using namespace ::java::lang;

  Thread *thread = Thread::currentThread();

  natThread *nt = (natThread *) thread->data;

  nt->park_helper.park (isAbsolute, time);

}


你可能感兴趣的:(jvm,thread)