Java多线程：锁的底层实现
最近准备招聘，在网上搜集了很多关于JVM获取的锁流程的文章，基本上所有关于偏向锁获取的部分都是错的，所以专门花时间自己看了看代码，整理出来供大家参考。
本文章是Java多线程系列的一篇文章，其他文章：
Java多线程：锁的底层实现
Java多线程：synchronized和volatile
Java多线程：JUC包-锁的封装
Java多线程：Thread的使用，以及wait(),notify(),notifyAll()
Java多线程：线程池
一、 基本概念
CAS(compare and swap)
参考资料：
https://blog.csdn.net/ls5718/article/details/52563959
缓存一致性协议
1. 什么是CAS
为了提高性能，JVM很多操作都依赖CAS实现，一种乐观锁的实现。本文锁优化中大量用到了CAS，故有必要先分析一下CAS的实现。
CAS：Compare and Swap。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。
JNI来完成CPU指令的操作：
openjdk\hotspot\src\os_cpu\windows_x86\vm\atomic_windows_x86.inline.hpp
//这里只粘贴了windowsx86的一个实现
//即：若dest和compare_value的值相同，就将dest的值替换为exchange_value
inline intptr_t Atomic::cmpxchg_ptr(intptr_t exchange_value, volatile intptr_t* dest, intptr_t compare_value) {
return (intptr_t)cmpxchg((jlong)exchange_value, (volatile jlong*)dest, (jlong)compare_value);
}
inline jlong Atomic::cmpxchg (jlong exchange_value, volatile jlong* dest, jlong compare_value) {
int mp = os::is_MP();//返回是否为多处理器
jint ex_lo = (jint)exchange_value;
jint ex_hi = *( ((jint*)&exchange_value) + 1 );
jint cmp_lo = (jint)compare_value;
jint cmp_hi = *( ((jint*)&compare_value) + 1 );
__asm {
push ebx
push edi
mov eax, cmp_lo
mov edx, cmp_hi
mov edi, dest
mov ebx, ex_lo
mov ecx, ex_hi
LOCK_IF_MP(mp) //会根据当前处理器的类型来决定是否为cmpxchg指令添加lock前缀。如果程序是在多处理器上运行，就为cmpxchg指令加上lock前缀（lock cmpxchg）。反之，如果程序是在单处理器上运行，就省略lock前缀（单处理器自身会维护单处理器内的顺序一致性，不需要lock前缀提供的内存屏障效果）。
cmpxchg8b qword ptr [edi]
pop edi
pop ebx
}
}CAS有3个操作数，内存值V，旧的预期值A，要修改的新值B。如果A=V，那么把B赋值给V，返回V；如果A！=V，直接返回V。
2. CAS的目的
利用CPU的CAS指令，同时借助JNI来完成Java的非阻塞算法。其它原子操作都是利用类似的特性完成的。而整个J.U.C都是建立在CAS之上的，因此对于synchronized阻塞算法，J.U.C在性能上有了很大的提升。
3. CAS存在的问题
CAS虽然很高效的解决原子操作，但是CAS仍然存在三大问题。ABA问题，循环时间长开销大和只能保证一个共享变量的原子操作
- ABA问题。因为CAS需要在操作值的时候检查下值有没有发生变化，如果没有发生变化则更新，但是如果一个值原来是A，变成了B，又变成了A，那么使用CAS进行检查时会发现它的值没有发生变化，但是实际上却变化了。ABA问题的解决思路就是使用版本号。在变量前面追加上版本号，每次变量更新的时候把版本号加一，那么A－B－A 就会变成1A-2B－3A。
从Java1.5开始JDK的atomic包里提供了一个类AtomicStampedReference来解决ABA问题。这个类的compareAndSet方法作用是首先检查当前引用是否等于预期引用，并且当前标志是否等于预期标志，如果全部相等，则以原子方式将该引用和该标志的值设置为给定的更新值。
关于ABA问题参考文档: http://blog.hesey.net/2011/09/resolve-aba-by-atomicstampedreference.html
循环时间长开销大。自旋CAS如果长时间不成功，会给CPU带来非常大的执行开销。如果JVM能支持处理器提供的pause指令那么效率会有一定的提升，pause指令有两个作用，第一它可以延迟流水线执行指令（de-pipeline）,使CPU不会消耗过多的执行资源，延迟的时间取决于具体实现的版本，在一些处理器上延迟时间是零。第二它可以避免在退出循环的时候因内存顺序冲突（memory order violation）而引起CPU流水线被清空（CPU pipeline flush），从而提高CPU的执行效率。
只能保证一个共享变量的原子操作。当对一个共享变量执行操作时，我们可以使用循环CAS的方式来保证原子操作，但是对多个共享变量操作时，循环CAS就无法保证操作的原子性，这个时候就可以用锁，或者有一个取巧的办法，就是把多个共享变量合并成一个共享变量来操作。比如有两个共享变量i＝2,j=a，合并一下ij=2a，然后用CAS来操作ij。从Java1.5开始JDK提供了AtomicReference类来保证引用对象之间的原子性，你可以把多个变量放在一个对象里来进行CAS操作。
安全点(safe point)
GC时的Stop the World(STW)是大家最大的敌人。但可能很多人没留意，除了GC，JVM底下还会发生这样那样的停顿。
JVM里有一条特殊的线程－－VM Threads，专门用来执行一些特殊的VM Operation，比如分派GC的STW，thread dump等，这些任务，都需要整个Heap，以及所有线程的状态是静止的，一致的才能进行。所以JVM引入了安全点(Safe Point)的概念，想办法在需要进行VM Operation时，通知所有的线程进入一个静止的安全点。 如何做到的见 聊聊JVM（九）理解进入safepoint时如何让Java线程全部阻塞
除了GC，其他触发安全点的VM Operation包括：
- 1. JIT相关，比如Code deoptimization, Flushing code cache
- 2. Class redefinition (e.g. javaagent，AOP代码植入的产生的instrumentation)
- 3. Biased lock revocation 取消偏向锁
- 4. Various debug operation (e.g. thread dump or deadlock check)
打开JDK7源码的vm_operations.hpp，有一个长长的列表，比如ReportJavaOutOfMemory等等不能尽录。
Java中的Monitor
参考：
java线程同步原理
Monitors – The Basic Idea of Java Synchronization => 译文：监视器–JAVA同步基本概念
java会为每个object对象分配一个monitor，当某个对象的同步方法（synchronized methods ）被多个线程调用时，该对象的monitor将负责处理这些访问的并发独占要求。
当一个线程调用一个对象的同步方法时，JVM会检查该对象的monitor。如果monitor没有被占用，那么这个线程就得到了monitor的占有权，可以继续执行该对象的同步方法；如果monitor被其他线程所占用，那么该线程将被挂起，直到monitor被释放。
当线程退出同步方法调用时，该线程会释放monitor，这将允许其他等待的线程获得monitor以使对同步方法的调用执行下去。
注意：Java对象的monitor机制和传统的临界检查代码区技术不一样。java的一个同步方法并不意味着同时只有一个线程独占执行，但临界检查代码区技术确实会保证同步方法在一个时刻只被一个线程独占执行。Java的monitor机制的准确含义是：任何时刻，对一个指定object对象的某同步方法只能由一个线程来调用。
java对象的monitor是跟随object实例来使用的，而不是跟随程序代码。两个线程可以同时执行相同的同步方法，比如：一个类的同步方法是xMethod()，有a,b两个对象实例，一个线程执行a.xMethod()，另一个线程执行b.xMethod(). 互不冲突。
hashCode相关
StackOverFlow：When does the jvm assign hashcode value in the object header
StackOverFlow：How do hashCode() and identityHashCode() work at the back end?
知乎：当Java处在偏向锁、重量级锁状态时，hashcode值存储在哪？
hashCode和identityHashCode底层是怎么生成的
二、 基础数据结构
markword
oopDesc
openjdk\hotspot\src\share\vm\oops\oop.hpp => oopDesc
class oopDesc {
friend class VMStructs;
private:
volatile markOop _mark;//markOop:Mark Word标记字段
union _metadata {
Klass* _klass;//对象类型元数据的指针
narrowKlass _compressed_klass;
} _metadata;
// Fast access to barrier set. Must be initialized.
static BarrierSet* _bs;
//以下省略
//...
}锁状态mask
openjdk\hotspot\src\share\vm\oops\markOop.hpp => markOopDesc
enum {
locked_value = 0,//00偏向锁
unlocked_value = 1,//01无锁
monitor_value = 2,//10监视器锁，又叫重量级锁
marked_value = 3,//11GC标记
biased_lock_pattern = 5//101偏向锁
};三、 获取锁的流程
1. monitorenter()
synchronized关键字修饰的代码段，在JVM被编译为monitorenter、monitorexit指令来获取和释放互斥锁.。
解释器执行monitorenter时会进入到InterpreterRuntime.cpp的InterpreterRuntime::monitorenter函数，具体实现如下：
openjdk/hotspot/src/share/vm/interpreter/interpreterRuntime.cpp => monitor_enter()
//------------------------------------------------------------------------------------------------------------------------
// Synchronization
//
// The interpreter's synchronization code is factored out so that it can
// be shared by method invocation and synchronized blocks.
//%note synchronization_3
//%note monitor_1
IRT_ENTRY_NO_ASYNC(void, InterpreterRuntime::monitorenter(JavaThread* thread, BasicObjectLock* elem))
#ifdef ASSERT
thread->last_frame().interpreter_frame_verify_monitor(elem);
#endif
if (PrintBiasedLockingStatistics) {
Atomic::inc(BiasedLocking::slow_path_entry_count_addr());
}
Handle h_obj(thread, elem->obj());
assert(Universe::heap()->is_in_reserved_or_null(h_obj()),
"must be NULL or an object");
if (UseBiasedLocking) {//如果开启了偏向锁，则直接使用fast_enter()
// Retry fast entry if bias is revoked to avoid unnecessary inflation
ObjectSynchronizer::fast_enter(h_obj, elem->lock(), true, CHECK);
} else {//否则使用slow_enter()
ObjectSynchronizer::slow_enter(h_obj, elem->lock(), CHECK);
}
assert(Universe::heap()->is_in_reserved_or_null(elem->obj()),
"must be NULL or an object");
#ifdef ASSERT
thread->last_frame().interpreter_frame_verify_monitor(elem);
#endif
IRT_END总的来说，就是：如果开启了偏向锁，则直接进入fast_enter()方法，否则进入slow_enter()方法
2. 尝试获取偏向锁：fast_enter()
openjdk/hotspot/src/share/vm/interpreter/synchronizer.cpp => ObjectSynchronizer::fast_enter()
// -----------------------------------------------------------------------------
// Fast Monitor Enter/Exit
// This the fast monitor enter. The interpreter and compiler use
// some assembly copies of this code. Make sure update those code
// if the following function is changed. The implementation is
// extremely sensitive to race condition. Be careful.
void ObjectSynchronizer::fast_enter(Handle obj, BasicLock* lock, bool attempt_rebias, TRAPS) {
if (UseBiasedLocking) {//如果开启了偏向锁
if (!SafepointSynchronize::is_at_safepoint()) {//如果不在安全点
BiasedLocking::Condition cond = BiasedLocking::revoke_and_rebias(obj, attempt_rebias, THREAD);
if (cond == BiasedLocking::BIAS_REVOKED_AND_REBIASED) {//如果调用结果为BIAS_REVOKED_AND_REBIASED
return;
}
} else {//当前处于安全点，调用revoke_at_safepoint
assert(!attempt_rebias, "can not rebias toward VM thread");
BiasedLocking::revoke_at_safepoint(obj);
}
assert(!obj->mark()->has_bias_pattern(), "biases should be revoked by now");
}
slow_enter (obj, lock, THREAD) ;//如果未开启偏向锁，或竞争偏向锁失败
}这段代码很短，就不作太多解释
2.1 revoke_and_rebias
openjdk/hotspot/src/share/vm/runtime/biasedLocking.cpp => revoke_and_rebias()
/**
* @param obj: 要执行操作的对象
* @param attempt_rebias: 尝试将该对象偏向到当前线程
* @param TRAPS:
*/
BiasedLocking::Condition BiasedLocking::revoke_and_rebias(Handle obj, bool attempt_rebias, TRAPS) {
assert(!SafepointSynchronize::is_at_safepoint(), "must not be called while at safepoint");
// We can revoke the biases of anonymously-biased objects
// efficiently enough that we should not cause these revocations to
// update the heuristics because doing so may cause unwanted bulk
// revocations (which are expensive) to occur.
markOop mark = obj->mark();
// 如果没有其他线程占用该对象（线程ID为0，后三位为101）且不尝试重新偏向（注意这里是不尝试重偏向，在`fast_enter()`中传过来的参数是true,也就是尝试重新偏向）
if (mark->is_biased_anonymously() && !attempt_rebias) {
// We are probably trying to revoke the bias of this object due to
// an identity hash code computation. Try to revoke the bias
// without a safepoint. This is possible if we can successfully
// compare-and-exchange an unbiased header into the mark word of
// the object, meaning that no other thread has raced to acquire
// the bias of the object.
//一般来讲，只有在重新计算对象hashCode的时候才会进入该分支，
//所以直接用用CAS操作将对象设置为无锁状态
markOop biased_value = mark;
markOop unbiased_prototype = markOopDesc::prototype()->set_age(mark->age());
markOop res_mark = (markOop) Atomic::cmpxchg_ptr(unbiased_prototype, obj->mark_addr(), mark);
if (res_mark == biased_value) {//如果CAS操作失败，说明存在竞争，交给调用者处理
return BIAS_REVOKED;
}
} else if (mark->has_bias_pattern()) {//如果对象为可偏向状态101，且尝试重新偏向（不管线程ID是不是为空）
Klass* k = obj->klass();
markOop prototype_header = k->prototype_header();
if (!prototype_header->has_bias_pattern()) {//如果有线程对该对象进行了全局锁定（即同步了静态方法/属性），则取消偏向操作
// This object has a stale bias from before the bulk revocation
// for this data type occurred. It's pointless to update the
// heuristics at this point so simply update the header with a
// CAS. If we fail this race, the object's bias has been revoked
// by another thread so we simply return and let the caller deal
// with it.
markOop biased_value = mark;
markOop res_mark = (markOop) Atomic::cmpxchg_ptr(prototype_header, obj->mark_addr(), mark);
assert(!(*(obj->mark_addr()))->has_bias_pattern(), "even if we raced, should still be revoked");
return BIAS_REVOKED;
} else if (prototype_header->bias_epoch() != mark->bias_epoch()) {//如果偏向时间戳没有过期，是不会重新偏向的，因为这个时候有另一个线程通过偏向锁获取到了这个对象的锁
// The epoch of this biasing has expired indicating that the
// object is effectively unbiased. Depending on whether we need
// to rebias or revoke the bias of this object we can do it
// efficiently enough with a CAS that we shouldn't update the
// heuristics. This is normally done in the assembly code but we
// can reach this point due to various points in the runtime
// needing to revoke biases.
if (attempt_rebias) {//尝试重新偏向
assert(THREAD->is_Java_thread(), "");
markOop biased_value = mark;
markOop rebiased_prototype = markOopDesc::encode((JavaThread*) THREAD, mark->age(), prototype_header->bias_epoch());
markOop res_mark = (markOop) Atomic::cmpxchg_ptr(rebiased_prototype, obj->mark_addr(), mark);
if (res_mark == biased_value) {
return BIAS_REVOKED_AND_REBIASED;
}
} else {
markOop biased_value = mark;
markOop unbiased_prototype = markOopDesc::prototype()->set_age(mark->age());
markOop res_mark = (markOop) Atomic::cmpxchg_ptr(unbiased_prototype, obj->mark_addr(), mark);
if (res_mark == biased_value) {
return BIAS_REVOKED;
}
}
}
}
HeuristicsResult heuristics = update_heuristics(obj(), attempt_rebias);
if (heuristics == HR_NOT_BIASED) {
return NOT_BIASED;
} else if (heuristics == HR_SINGLE_REVOKE) {
Klass *k = obj->klass();
markOop prototype_header = k->prototype_header();
if (mark->biased_locker() == THREAD &&
prototype_header->bias_epoch() == mark->bias_epoch()) {
// A thread is trying to revoke the bias of an object biased
// toward it, again likely due to an identity hash code
// computation. We can again avoid a safepoint in this case
// since we are only going to walk our own stack. There are no
// races with revocations occurring in other threads because we
// reach no safepoints in the revocation path.
// Also check the epoch because even if threads match, another thread
// can come in with a CAS to steal the bias of an object that has a
// stale epoch.
ResourceMark rm;
if (TraceBiasedLocking) {
tty->print_cr("Revoking bias by walking my own stack:");
}
BiasedLocking::Condition cond = revoke_bias(obj(), false, false, (JavaThread*) THREAD);
((JavaThread*) THREAD)->set_cached_monitor_info(NULL);
assert(cond == BIAS_REVOKED, "why not?");
return cond;
} else {
VM_RevokeBias revoke(&obj, (JavaThread*) THREAD);
VMThread::execute(&revoke);
return revoke.status_code();
}
}
assert((heuristics == HR_BULK_REVOKE) ||
(heuristics == HR_BULK_REBIAS), "?");
VM_BulkRevokeBias bulk_revoke(&obj, (JavaThread*) THREAD,
(heuristics == HR_BULK_REBIAS),
attempt_rebias);
VMThread::execute(&bulk_revoke);
return bulk_revoke.status_code();
}这个函数定义了获取偏向锁的流程，具体逻辑总结：
1. 判断当前对象是否为可偏向（101），且偏向时间戳已过期（没有其他线程在占用该对象），如果是，则进入步骤2，否则进入步骤3
2. 执行CAS操作将markword中的线程ID替换为本线程ID。如果成功则进入步骤4，否则进入步骤3
3. 存在竞争，当达到全局安全点（safepoint），获得偏向锁的线程被挂起，撤销偏向锁，并升级为轻量级，升级完成后被阻塞在安全点的线程继续执行同步代码块；
4. 执行同步代码
2.2 偏向锁的撤销： revoke_at_safepoint
只有当其它线程尝试竞争偏向锁时，持有偏向锁的线程才会释放锁，偏向锁的撤销由BiasedLocking::revoke_at_safepoint方法实现：
openjdk\hotspot\src\share\vm\runtime\synchronizer.cpp
void BiasedLocking::revoke_at_safepoint(Handle h_obj) {
assert(SafepointSynchronize::is_at_safepoint(), "must only be called while at safepoint");
oop obj = h_obj();
HeuristicsResult heuristics = update_heuristics(obj, false);
if (heuristics == HR_SINGLE_REVOKE) {
revoke_bias(obj, false, false, NULL);
} else if ((heuristics == HR_BULK_REBIAS) ||
(heuristics == HR_BULK_REVOKE)) {
bulk_revoke_or_rebias_at_safepoint(obj, (heuristics == HR_BULK_REBIAS), false, NULL);
}
clean_up_cached_monitor_info();
}偏向锁的释放，需要等待全局安全点（在这个时间点上没有正在执行的字节码），它会首先暂停拥有偏向锁的线程，然后检查持有偏向锁的线程是否还活着，如果线程不处于活动状态，则将对象头设置成无锁状态。如果线程仍然活着，拥有偏向锁的栈会被执行，遍历偏向对象的所记录。栈帧中的锁记录和对象头的Mark Word要么重新偏向其他线程，要么恢复到无锁，或者标记对象不适合作为偏向锁。最后唤醒暂停的线程。
偏向锁在Java 1.6之后是默认启用的，但在应用程序启动几秒钟之后才激活，可以使用-XX:BiasedLockingStartupDelay=0参数关闭延迟，如果确定应用程序中所有锁通常情况下处于竞争状态，可以通过XX:-UseBiasedLocking=false参数关闭偏向锁。
3.1 轻量级锁的获取： slow_enter
如果关闭了偏向锁功能，则会直接再monitorenter方法中进入slow_enter。否则，在faster_enter中竞争偏向锁导致偏向锁升级为轻量级锁后进入。
openjdk\hotspot\src\share\vm\runtime\synchronizer.cpp => slow_enter()
// -----------------------------------------------------------------------------
// Interpreter/Compiler Slow Case
// This routine is used to handle interpreter/compiler slow case
// We don't need to use fast path here, because it must have been
// failed in the interpreter/compiler code.
void ObjectSynchronizer::slow_enter(Handle obj, BasicLock* lock, TRAPS) {
markOop mark = obj->mark();
assert(!mark->has_bias_pattern(), "should not see bias pattern here");
if (mark->is_neutral()) {
// Anticipate successful CAS -- the ST of the displaced mark must
// be visible <= the ST performed by the CAS.
lock->set_displaced_header(mark);
if (mark == (markOop) Atomic::cmpxchg_ptr(lock, obj()->mark_addr(), mark)) {
TEVENT (slow_enter: release stacklock) ;
return ;
}
// Fall through to inflate() ...
} else
if (mark->has_locker() && THREAD->is_lock_owned((address)mark->locker())) {
assert(lock != mark->locker(), "must not re-lock the same lock");
assert(lock != (BasicLock*)obj->mark(), "don't relock with same BasicLock");
lock->set_displaced_header(NULL);
return;
}
#if 0
// The following optimization isn't particularly useful.
if (mark->has_monitor() && mark->monitor()->is_entered(THREAD)) {
lock->set_displaced_header (NULL) ;
return ;
}
#endif
// The object header will never be displaced to this lock,
// so it does not matter what the value is, except that it
// must be non-zero to avoid looking like a re-entrant lock,
// and must not look locked either.
lock->set_displaced_header(markOopDesc::unused_mark());
ObjectSynchronizer::inflate(THREAD, obj())->enter(THREAD);//锁膨胀升级
}1、markOop mark = obj->mark()方法获取对象的markOop数据mark；
2、mark->is_neutral()方法判断mark是否为无锁状态：mark的偏向锁标志位为 0，锁标志位为 01；
3、如果mark处于无锁状态，则进入步骤（4），否则执行步骤（6）；
4、把mark保存到BasicLock对象的_displaced_header字段；
5、通过CAS尝试将Mark Word更新为指向BasicLock对象的指针，如果更新成功，表示竞争到锁，则执行同步代码，否则执行步骤（6）；
6、如果当前mark处于加锁状态，且mark中的ptr指针指向当前线程的栈帧，则执行同步代码，否则说明有多个线程竞争轻量级锁，轻量级锁需要膨胀升级为重量级锁；
假设线程A和B同时执行到临界区if (mark->is_neutral())：
1、线程AB都把Mark Word复制到各自的_displaced_header字段，该数据保存在线程的栈帧上，是线程私有的；
2、Atomic::cmpxchg_ptr原子操作保证只有一个线程可以把指向栈帧的指针复制到Mark Word，假设此时线程A执行成功，并返回继续执行同步代码块；
3、线程B执行失败，退出临界区，通过ObjectSynchronizer::inflate方法开始膨胀锁；
3.2轻量级锁的释放：fast_exit
void ObjectSynchronizer::fast_exit(oop object, BasicLock* lock, TRAPS) {
assert(!object->mark()->has_bias_pattern(), "should not see bias pattern here");
// if displaced header is null, the previous enter is recursive enter, no-op
markOop dhw = lock->displaced_header();
markOop mark ;
if (dhw == NULL) {
// Recursive stack-lock.
// Diagnostics -- Could be: stack-locked, inflating, inflated.
mark = object->mark() ;
assert (!mark->is_neutral(), "invariant") ;
if (mark->has_locker() && mark != markOopDesc::INFLATING()) {
assert(THREAD->is_lock_owned((address)mark->locker()), "invariant") ;
}
if (mark->has_monitor()) {
ObjectMonitor * m = mark->monitor() ;
assert(((oop)(m->object()))->mark() == mark, "invariant") ;
assert(m->is_entered(THREAD), "invariant") ;
}
return ;
}
mark = object->mark() ;
// If the object is stack-locked by the current thread, try to
// swing the displaced header from the box back to the mark.
if (mark == (markOop) lock) {
assert (dhw->is_neutral(), "invariant") ;
if ((markOop) Atomic::cmpxchg_ptr (dhw, object->mark_addr(), mark) == mark) {//成功的释放了锁
TEVENT (fast_exit: release stacklock) ;
return;
}
}
//如果没有成功释放锁，则说明有其他线程竞争，进行锁膨胀
ObjectSynchronizer::inflate(THREAD, object)->exit (true, THREAD) ;
}1、确保处于偏向锁状态时不会执行这段逻辑；
2、取出在获取轻量级锁时保存在BasicLock对象的mark数据dhw；
3、通过CAS尝试把dhw替换到当前的Mark Word，如果CAS成功，说明成功的释放了锁，否则执行步骤（4）；
4、如果CAS失败，说明有其它线程在尝试获取该锁，这时需要将该锁升级为重量级锁，并释放；
4. 重量级锁：
重量级锁通过对象内部的监视器（monitor）实现，其中monitor的本质是依赖于底层操作系统的Mutex Lock实现，操作系统实现线程之间的切换需要从用户态到内核态的切换，切换成本非常高。
4.1 重量级锁的膨胀过程：ObjectSynchronizer::inflate
// Note that we could encounter some performance loss through false-sharing as
// multiple locks occupy the same $ line. Padding might be appropriate.
ObjectMonitor * ATTR ObjectSynchronizer::inflate (Thread * Self, oop object) {
// Inflate mutates the heap ...
// Relaxing assertion for bug 6320749.
assert (Universe::verify_in_progress() ||
!SafepointSynchronize::is_at_safepoint(), "invariant") ;
for (;;) {//自旋
const markOop mark = object->mark() ;
assert (!mark->has_bias_pattern(), "invariant") ;
// The mark can be in one of the following states:
// * Inflated - just return
// * Stack-locked - coerce it to inflated
// * INFLATING - busy wait for conversion to complete
// * Neutral - aggressively inflate the object.
// * BIASED - Illegal. We should never see this
// CASE: inflated --已膨胀，即重量级锁
if (mark->has_monitor()) {//判断当前是否为重量级锁
ObjectMonitor * inf = mark->monitor() ;//获取指向ObjectMonitor的指针
assert (inf->header()->is_neutral(), "invariant");
assert (inf->object() == object, "invariant") ;
assert (ObjectSynchronizer::verify_objmon_isinpool(inf), "monitor is invalid");
return inf ;
}
// CASE: inflation in progress - inflating over a stack-lock. .膨胀等待（其他线程正在从轻量级锁转为膨胀锁）
// Some other thread is converting from stack-locked to inflated.
// Only that thread can complete inflation -- other threads must wait.
// The INFLATING value is transient.
// Currently, we spin/yield/park and poll the markword, waiting for inflation to finish.
// We could always eliminate polling by parking the thread on some auxiliary list.
if (mark == markOopDesc::INFLATING()) {
TEVENT (Inflate: spin while INFLATING) ;
ReadStableMark(object) ;
continue ;
}
// CASE: stack-locked 栈锁（轻量级锁）
// Could be stack-locked either by this thread or by some other thread.
//
// Note that we allocate the objectmonitor speculatively, _before_ attempting
// to install INFLATING into the mark word. We originally installed INFLATING,
// allocated the objectmonitor, and then finally STed the address of the
// objectmonitor into the mark. This was correct, but artificially lengthened
// the interval in which INFLATED appeared in the mark, thus increasing
// the odds of inflation contention.
//
// We now use per-thread private objectmonitor free lists.
// These list are reprovisioned from the global free list outside the
// critical INFLATING...ST interval. A thread can transfer
// multiple objectmonitors en-mass from the global free list to its local free list.
// This reduces coherency traffic and lock contention on the global free list.
// Using such local free lists, it doesn't matter if the omAlloc() call appears
// before or after the CAS(INFLATING) operation.
// See the comments in omAlloc().
if (mark->has_locker()) {
ObjectMonitor * m = omAlloc (Self) ;//获取一个可用的ObjectMonitor
// Optimistically prepare the objectmonitor - anticipate successful CAS
// We do this before the CAS in order to minimize the length of time
// in which INFLATING appears in the mark.
m->Recycle();
m->_Responsible = NULL ;
m->OwnerIsThread = 0 ;
m->_recursions = 0 ;
m->_SpinDuration = ObjectMonitor::Knob_SpinLimit ; // Consider: maintain by type/class
markOop cmp = (markOop) Atomic::cmpxchg_ptr (markOopDesc::INFLATING(), object->mark_addr(), mark) ;
if (cmp != mark) {//CAS失败，说明冲突了，自旋等待
omRelease (Self, m, true) ;//释放监视器锁
continue ; // Interference -- just retry
}
// We've successfully installed INFLATING (0) into the mark-word.
// This is the only case where 0 will appear in a mark-work.
// Only the singular thread that successfully swings the mark-word
// to 0 can perform (or more precisely, complete) inflation.
//
// Why do we CAS a 0 into the mark-word instead of just CASing the
// mark-word from the stack-locked value directly to the new inflated state?
// Consider what happens when a thread unlocks a stack-locked object.
// It attempts to use CAS to swing the displaced header value from the
// on-stack basiclock back into the object header. Recall also that the
// header value (hashcode, etc) can reside in (a) the object header, or
// (b) a displaced header associated with the stack-lock, or (c) a displaced
// header in an objectMonitor. The inflate() routine must copy the header
// value from the basiclock on the owner's stack to the objectMonitor, all
// the while preserving the hashCode stability invariants. If the owner
// decides to release the lock while the value is 0, the unlock will fail
// and control will eventually pass from slow_exit() to inflate. The owner
// will then spin, waiting for the 0 value to disappear. Put another way,
// the 0 causes the owner to stall if the owner happens to try to
// drop the lock (restoring the header from the basiclock to the object)
// while inflation is in-progress. This protocol avoids races that might
// would otherwise permit hashCode values to change or "flicker" for an object.
// Critically, while object->mark is 0 mark->displaced_mark_helper() is stable.
// 0 serves as a "BUSY" inflate-in-progress indicator.
// fetch the displaced mark from the owner's stack.
// The owner can't die or unwind past the lock while our INFLATING
// object is in the mark. Furthermore the owner can't complete
// an unlock on the object, either.
markOop dmw = mark->displaced_mark_helper() ;
assert (dmw->is_neutral(), "invariant") ;
// Setup monitor fields to proper values -- prepare the monitor
//CAS成功，设置ObjectMonitor的_header、_owner和_object等
m->set_header(dmw) ;
// Optimization: if the mark->locker stack address is associated
// with this thread we could simply set m->_owner = Self and
// m->OwnerIsThread = 1. Note that a thread can inflate an object
// that it has stack-locked -- as might happen in wait() -- directly
// with CAS. That is, we can avoid the xchg-NULL .... ST idiom.
m->set_owner(mark->locker());
m->set_object(object);
// TODO-FIXME: assert BasicLock->dhw != 0.
// Must preserve store ordering. The monitor state must
// be stable at the time of publishing the monitor address.
guarantee (object->mark() == markOopDesc::INFLATING(), "invariant") ;
object->release_set_mark(markOopDesc::encode(m));
// Hopefully the performance counters are allocated on distinct cache lines
// to avoid false sharing on MP systems ...
if (ObjectMonitor::_sync_Inflations != NULL) ObjectMonitor::_sync_Inflations->inc() ;
TEVENT(Inflate: overwrite stacklock) ;
if (TraceMonitorInflation) {
if (object->is_instance()) {
ResourceMark rm;
tty->print_cr("Inflating object " INTPTR_FORMAT " , mark " INTPTR_FORMAT " , type %s",
(void *) object, (intptr_t) object->mark(),
object->klass()->external_name());
}
}
return m ;
}
// CASE: neutral 无锁
// TODO-FIXME: for entry we currently inflate and then try to CAS _owner.
// If we know we're inflating for entry it's better to inflate by swinging a
// pre-locked objectMonitor pointer into the object header. A successful
// CAS inflates the object *and* confers ownership to the inflating thread.
// In the current implementation we use a 2-step mechanism where we CAS()
// to inflate and then CAS() again to try to swing _owner from NULL to Self.
// An inflateTry() method that we could call from fast_enter() and slow_enter()
// would be useful.
assert (mark->is_neutral(), "invariant");
ObjectMonitor * m = omAlloc (Self) ;
// prepare m for installation - set monitor to initial state
m->Recycle();
m->set_header(mark);
m->set_owner(NULL);
m->set_object(object);
m->OwnerIsThread = 1 ;
m->_recursions = 0 ;
m->_Responsible = NULL ;
m->_SpinDuration = ObjectMonitor::Knob_SpinLimit ; // consider: keep metastats by type/class
if (Atomic::cmpxchg_ptr (markOopDesc::encode(m), object->mark_addr(), mark) != mark) {
m->set_object (NULL) ;
m->set_owner (NULL) ;
m->OwnerIsThread = 0 ;
m->Recycle() ;
omRelease (Self, m, true) ;
m = NULL ;
continue ;
// interference - the markword changed - just retry.
// The state-transitions are one-way, so there's no chance of
// live-lock -- "Inflated" is an absorbing state.
}
// Hopefully the performance counters are allocated on distinct
// cache lines to avoid false sharing on MP systems ...
if (ObjectMonitor::_sync_Inflations != NULL) ObjectMonitor::_sync_Inflations->inc() ;
TEVENT(Inflate: overwrite neutral) ;
if (TraceMonitorInflation) {
if (object->is_instance()) {
ResourceMark rm;
tty->print_cr("Inflating object " INTPTR_FORMAT " , mark " INTPTR_FORMAT " , type %s",
(void *) object, (intptr_t) object->mark(),
object->klass()->external_name());
}
}
return m ;
}
}膨胀过程的实现比较复杂，大概实现过程如下：
1、整个膨胀过程在自旋下完成；
2、mark->has_monitor()方法判断当前是否为重量级锁（上图18-25行），即Mark Word的锁标识位为 10，如果当前状态为重量级锁，执行步骤（3），否则执行步骤（4）；
3、膨胀过程已经完成：mark->monitor()方法获取指向ObjectMonitor的指针，并返回。
4、当前锁处于膨胀中（上图33-37行），说明该锁正在被其它线程执行膨胀操作，则当前线程就进行自旋等待锁膨胀完成，这里需要注意一点，虽然是自旋操作，但不会一直占用cpu资源，每隔一段时间会通过os::NakedYield方法放弃cpu资源，或通过park方法挂起；如果其他线程完成锁的膨胀操作，则退出自旋并返回；
5、当前是轻量级锁状态（上图58-138行），即锁标识位为 00，膨胀过程如下：
通过omAlloc方法，获取一个可用的ObjectMonitor monitor，并重置monitor数据；
通过CAS尝试将Mark Word设置为markOopDesc:INFLATING，标识当前锁正在膨胀中，如果CAS失败，说明同一时刻其它线程已经将Mark Word设置为markOopDesc:INFLATING，当前线程进行自旋等待膨胀完成；
如果CAS成功，设置monitor的各个字段：_header、_owner和_object等，并返回；
6、无锁（中立，上图150-186行），重置监视器值；
4.2 重量级锁的竞争：ObjectMonitor::enter
当锁膨胀完成并返回对应的monitor时，并不表示该线程竞争到了锁，真正的锁竞争发生在ObjectMonitor::enter方法中。
openjdk\hotspot\src\share\vm\runtime\objectMonitor.cpp => ObjectMonitor::enter()
void ATTR ObjectMonitor::enter(TRAPS) {
// The following code is ordered to check the most common cases first
// and to reduce RTS->RTO cache line upgrades on SPARC and IA32 processors.
Thread * const Self = THREAD ;
void * cur ;
cur = Atomic::cmpxchg_ptr (Self, &_owner, NULL) ;
if (cur == NULL) {//CAS成功
// Either ASSERT _recursions == 0 or explicitly set _recursions = 0.
assert (_recursions == 0 , "invariant") ;
assert (_owner == Self, "invariant") ;
// CONSIDER: set or assert OwnerIsThread == 1
return ;
}
if (cur == Self) {//重入锁
// TODO-FIXME: check for integer overflow! BUGID 6557169.
_recursions ++ ;
return ;
}
if (Self->is_lock_owned ((address)cur)) {
assert (_recursions == 0, "internal state error");
_recursions = 1 ;
// Commute owner from a thread-specific on-stack BasicLockObject address to
// a full-fledged "Thread *".
_owner = Self ;
OwnerIsThread = 1 ;
return ;
}
// We've encountered genuine contention.
assert (Self->_Stalled == 0, "invariant") ;
Self->_Stalled = intptr_t(this) ;
// Try one round of spinning *before* enqueueing Self
// and before going through the awkward and expensive state
// transitions. The following spin is strictly optional ...
// Note that if we acquire the monitor from an initial spin
// we forgo posting JVMTI events and firing DTRACE probes.
if (Knob_SpinEarly && TrySpin (Self) > 0) {
assert (_owner == Self , "invariant") ;
assert (_recursions == 0 , "invariant") ;
assert (((oop)(object()))->mark() == markOopDesc::encode(this), "invariant") ;
Self->_Stalled = 0 ;
return ;
}
assert (_owner != Self , "invariant") ;
assert (_succ != Self , "invariant") ;
assert (Self->is_Java_thread() , "invariant") ;
JavaThread * jt = (JavaThread *) Self ;
assert (!SafepointSynchronize::is_at_safepoint(), "invariant") ;
assert (jt->thread_state() != _thread_blocked , "invariant") ;
assert (this->object() != NULL , "invariant") ;
assert (_count >= 0, "invariant") ;
// Prevent deflation at STW-time. See deflate_idle_monitors() and is_busy().
// Ensure the object-monitor relationship remains stable while there's contention.
Atomic::inc_ptr(&_count);
EventJavaMonitorEnter event;
{ // Change java thread status to indicate blocked on monitor enter.
JavaThreadBlockedOnMonitorEnterState jtbmes(jt, this);
DTRACE_MONITOR_PROBE(contended__enter, this, object(), jt);
if (JvmtiExport::should_post_monitor_contended_enter()) {
JvmtiExport::post_monitor_contended_enter(jt, this);
}
OSThreadContendState osts(Self->osthread());
ThreadBlockInVM tbivm(jt);
Self->set_current_pending_monitor(this);
// TODO-FIXME: change the following for(;;) loop to straight-line code.
for (;;) {
jt->set_suspend_equivalent();
// cleared by handle_special_suspend_equivalent_condition()
// or java_suspend_self()
EnterI (THREAD) ;
if (!ExitSuspendEquivalent(jt)) break ;
//
// We have acquired the contended monitor, but while we were
// waiting another thread suspended us. We don't want to enter
// the monitor while suspended because that would surprise the
// thread that suspended us.
//
_recursions = 0 ;
_succ = NULL ;
exit (false, Self) ;
jt->java_suspend_self();
}
Self->set_current_pending_monitor(NULL);
}
Atomic::dec_ptr(&_count);
assert (_count >= 0, "invariant") ;
Self->_Stalled = 0 ;
// Must either set _recursions = 0 or ASSERT _recursions == 0.
assert (_recursions == 0 , "invariant") ;
assert (_owner == Self , "invariant") ;
assert (_succ != Self , "invariant") ;
assert (((oop)(object()))->mark() == markOopDesc::encode(this), "invariant") ;
// The thread -- now the owner -- is back in vm mode.
// Report the glorious news via TI,DTrace and jvmstat.
// The probe effect is non-trivial. All the reportage occurs
// while we hold the monitor, increasing the length of the critical
// section. Amdahl's parallel speedup law comes vividly into play.
//
// Another option might be to aggregate the events (thread local or
// per-monitor aggregation) and defer reporting until a more opportune
// time -- such as next time some thread encounters contention but has
// yet to acquire the lock. While spinning that thread could
// spinning we could increment JVMStat counters, etc.
DTRACE_MONITOR_PROBE(contended__entered, this, object(), jt);
if (JvmtiExport::should_post_monitor_contended_entered()) {
JvmtiExport::post_monitor_contended_entered(jt, this);
}
if (event.should_commit()) {
event.set_klass(((oop)this->object())->klass());
event.set_previousOwner((TYPE_JAVALANGTHREAD)_previous_owner_tid);
event.set_address((TYPE_ADDRESS)(uintptr_t)(this->object_addr()));
event.commit();
}
if (ObjectMonitor::_sync_ContendedLockAttempts != NULL) {
ObjectMonitor::_sync_ContendedLockAttempts->inc() ;
}
}1、通过CAS尝试把monitor的_owner字段设置为当前线程；
2、如果设置之前的_owner指向当前线程，说明当前线程再次进入monitor，即重入锁，执行_recursions ++ ，记录重入的次数；
3、如果之前的_owner指向的地址在当前线程中，这种描述有点拗口，换一种说法：之前_owner指向的BasicLock在当前线程栈上，说明当前线程是第一次进入该monitor，设置_recursions为1，_owner为当前线程，该线程成功获得锁并返回；
4、如果获取锁失败，则等待锁的释放；
4.3. monitor等待
monitor竞争失败的线程，通过自旋执行ObjectMonitor::EnterI方法等待锁的释放，EnterI方法的部分逻辑实现如下：
ObjectWaiter node(Self) ;
Self->_ParkEvent->reset() ;
node._prev = (ObjectWaiter *) 0xBAD ;
node.TState = ObjectWaiter::TS_CXQ ;
// Push "Self" onto the front of the _cxq.
// Once on cxq/EntryList, Self stays on-queue until it acquires the lock.
// Note that spinning tends to reduce the rate at which threads
// enqueue and dequeue on EntryList|cxq.
ObjectWaiter * nxt ;
for (;;) {
node._next = nxt = _cxq ;
if (Atomic::cmpxchg_ptr (&node, &_cxq, nxt) == nxt) break ;
// Interference - the CAS failed because _cxq changed. Just retry.
// As an optional optimization we retry the lock.
if (TryLock (Self) > 0) {
assert (_succ != Self , "invariant") ;
assert (_owner == Self , "invariant") ;
assert (_Responsible != Self , "invariant") ;
return ;
}
}1、当前线程被封装成ObjectWaiter对象node，状态设置成ObjectWaiter::TS_CXQ；
2、在for循环中，通过CAS把node节点push到_cxq列表中，同一时刻可能有多个线程把自己的node节点push到_cxq列表中；
3、node节点push到_cxq列表之后，通过自旋尝试获取锁，如果还是没有获取到锁，则通过park将当前线程挂起，等待被唤醒，实现如下：
for (;;) {
if (TryLock (Self) > 0) break ;
assert (_owner != Self, "invariant") ;
if ((SyncFlags & 2) && _Responsible == NULL) {
Atomic::cmpxchg_ptr (Self, &_Responsible, NULL) ;
}
// park self
if (_Responsible == Self || (SyncFlags & 1)) {
TEVENT (Inflated enter - park TIMED) ;
Self->_ParkEvent->park ((jlong) RecheckInterval) ;
// Increase the RecheckInterval, but clamp the value.
RecheckInterval *= 8 ;
if (RecheckInterval > 1000) RecheckInterval = 1000 ;
} else {
TEVENT (Inflated enter - park UNTIMED) ;
Self->_ParkEvent->park() ;//当前线程挂起
}
if (TryLock(Self) > 0) break ;
// The lock is still contested.
// Keep a tally of the # of futile wakeups.
// Note that the counter is not protected by a lock or updated by atomics.
// That is by design - we trade "lossy" counters which are exposed to
// races during updates for a lower probe effect.
TEVENT (Inflated enter - Futile wakeup) ;
if (ObjectMonitor::_sync_FutileWakeups != NULL) {
ObjectMonitor::_sync_FutileWakeups->inc() ;
}
++ nWakeups ;
// Assuming this is not a spurious wakeup we'll normally find _succ == Self.
// We can defer clearing _succ until after the spin completes
// TrySpin() must tolerate being called with _succ == Self.
// Try yet another round of adaptive spinning.
if ((Knob_SpinAfterFutile & 1) && TrySpin (Self) > 0) break ;
// We can find that we were unpark()ed and redesignated _succ while
// we were spinning. That's harmless. If we iterate and call park(),
// park() will consume the event and return immediately and we'll
// just spin again. This pattern can repeat, leaving _succ to simply
// spin on a CPU. Enable Knob_ResetEvent to clear pending unparks().
// Alternately, we can sample fired() here, and if set, forgo spinning
// in the next iteration.
if ((Knob_ResetEvent & 1) && Self->_ParkEvent->fired()) {
Self->_ParkEvent->reset() ;
OrderAccess::fence() ;
}
if (_succ == Self) _succ = NULL ;
// Invariant: after clearing _succ a thread *must* retry _owner before parking.
OrderAccess::fence() ;
}4、当该线程被唤醒时，会从挂起的点继续执行，通过ObjectMonitor::TryLock尝试获取锁，TryLock方法实现如下：
int ObjectMonitor::TryLock (Thread * Self) {
for (;;) {
void * own = _owner ;
if (own != NULL) return 0 ;
if (Atomic::cmpxchg_ptr (Self, &_owner, NULL) == NULL) {//CAS成功，获取锁
// Either guarantee _recursions == 0 or set _recursions = 0.
assert (_recursions == 0, "invariant") ;
assert (_owner == Self, "invariant") ;
// CONSIDER: set or assert that OwnerIsThread == 1
return 1 ;
}
// The lock had been free momentarily, but we lost the race to the lock.
// Interference -- the CAS failed.
// We can either return -1 or retry.
// Retry doesn't make as much sense because the lock was just acquired.
if (true) return -1 ;
}
}其本质就是通过CAS设置monitor的_owner字段为当前线程，如果CAS成功，则表示该线程获取了锁，跳出自旋操作，执行同步代码，否则继续被挂起；
4.4. monitor释放
当某个持有锁的线程执行完同步代码块时，会进行锁的释放，给其它线程机会执行同步代码，在HotSpot中，通过退出monitor的方式实现锁的释放，并通知被阻塞的线程，具体实现位于ObjectMonitor::exit方法中。
void ATTR ObjectMonitor::exit(bool not_suspended, TRAPS) {
Thread * Self = THREAD ;
if (THREAD != _owner) {
if (THREAD->is_lock_owned((address) _owner)) {
// Transmute _owner from a BasicLock pointer to a Thread address.
// We don't need to hold _mutex for this transition.
// Non-null to Non-null is safe as long as all readers can
// tolerate either flavor.
assert (_recursions == 0, "invariant") ;
_owner = THREAD ;
_recursions = 0 ;
OwnerIsThread = 1 ;
} else {
// NOTE: we need to handle unbalanced monitor enter/exit
// in native code by throwing an exception.
// TODO: Throw an IllegalMonitorStateException ?
TEVENT (Exit - Throw IMSX) ;
assert(false, "Non-balanced monitor enter/exit!");
if (false) {
THROW(vmSymbols::java_lang_IllegalMonitorStateException());
}
return;
}
}
if (_recursions != 0) {
_recursions--; // this is simple recursive enter
TEVENT (Inflated exit - recursive) ;
return ;
}
...省略...1、如果是重量级锁的释放，monitor中的_owner指向当前线程，即THREAD == _owner；
2、根据不同的策略（由QMode指定），从cxq或EntryList中获取头节点，通过ObjectMonitor::ExitEpilog方法唤醒该节点封装的线程，唤醒操作最终由unpark完成，实现如下：
void ObjectMonitor::ExitEpilog (Thread * Self, ObjectWaiter * Wakee) {
assert (_owner == Self, "invariant") ;
// Exit protocol:
// 1. ST _succ = wakee
// 2. membar #loadstore|#storestore;
// 2. ST _owner = NULL
// 3. unpark(wakee)
_succ = Knob_SuccEnabled ? Wakee->_thread : NULL ;
ParkEvent * Trigger = Wakee->_event ;
// Hygiene -- once we've set _owner = NULL we can't safely dereference Wakee again.
// The thread associated with Wakee may have grabbed the lock and "Wakee" may be
// out-of-scope (non-extant).
Wakee = NULL ;
// Drop the lock
OrderAccess::release_store_ptr (&_owner, NULL) ;
OrderAccess::fence() ; // ST _owner vs LD in unpark()
if (SafepointSynchronize::do_call_back()) {
TEVENT (unpark before SAFEPOINT) ;
}
DTRACE_MONITOR_PROBE(contended__exit, this, object(), Self);
Trigger->unpark() ;
// Maintain stats and report events to JVMTI
if (ObjectMonitor::_sync_Parks != NULL) {
ObjectMonitor::_sync_Parks->inc() ;
}
}3、被唤醒的线程，继续执行monitor的竞争；
四、总结
直接看代码虽然可以窥探加锁流程的细节，但是源代码由于复合了对象的多个状态（无锁、偏向、轻量级、重量级）并对其进行了很多优化，所以我们并不能很直观的了解到各个锁状态之间的转换。所以整理出了下面的状态机图帮助理解：
本文代码解析部分参考了 jdk源码剖析二: 对象内存布局、synchronized原理,并修正了其中的部分错误。
其他参考资料：
深入JVM锁机制1-synchronized
Java Language Specification
转载本文请注明出处https://blog.csdn.net/qq_29753285/article/details/81299509
手机扫一扫
移动阅读更方便
你可能感兴趣的文章