Java8的伪共享和缓存行填充--@Contended注释
在我的前一篇文章<伪共享和缓存行填充,从Java 6, Java 7 到Java 8>中, 我们演示了在Java 8中,可以采用@Contended在类级别上的注释,来进行缓存行填充。这样,多线程情况下的伪共享冲突问题。 感兴趣的同学可以查看该文。
其实,@Contended注释还可以应用于字段级别(Field-Level),当应用于字段级别时,被注释的字段将和其他字段隔离开来,会被加载在独立的缓存行上。在字段级别上,@Contended还支持一个“contention group”属性(Class-Level不支持),同一group的字段们在内存上将是连续,但和其他他字段隔离开来。
上面只是泛泛的介绍一下。关于@Contended应用于Field-Level特别是contention group的相关的资料很少,源代码中的注释中有一些,还有关于JEP-142(即关于增加@Contended的提议)的邮件讨论组中的描述(http://mail.openjdk.java.net/pipermail/hotspot-dev/2012-November/007309.html),其中的讲解是非常详细的(由于该讨论发生在@Contended实现的最初阶段,不能保证和现在的实现完全一致), 我摘录和翻译如下:
@Contended注释的行为如下所示:
A,在类上应用Contended:
@Contended public static class ContendedTest2 { private Object plainField1; private Object plainField2; private Object plainField3; private Object plainField4; }
将使整个字段块的两端都被填充:(以下是使用 –XX:+PrintFieldLayout的输出)(翻译注:注意前面的@140表示字段在类中的地址偏移)
TestContended$ContendedTest2: field layout Entire class is marked contended @140 --- instance fields start --- @140 "plainField1" Ljava.lang.Object; @144 "plainField2" Ljava.lang.Object; @148 "plainField3" Ljava.lang.Object; @152 "plainField4" Ljava.lang.Object; @288 --- instance fields end --- @288 --- instance ends ---
注意,我们使用了128bytes的填充 -- 2倍于大多数硬件缓存行的大小 -- 来避免相邻扇区预取导致的伪共享冲突。
B,在字段上应用Contended:
public static class ContendedTest1 { @Contended private Object contendedField1; private Object plainField1; private Object plainField2; private Object plainField3; private Object plainField4; }
将导致该字段从连续的字段块中分离开来并高效的添加填充:
TestContended$ContendedTest1: field layout @ 12 --- instance fields start --- @ 12 "plainField1" Ljava.lang.Object; @ 16 "plainField2" Ljava.lang.Object; @ 20 "plainField3" Ljava.lang.Object; @ 24 "plainField4" Ljava.lang.Object; @156 "contendedField1" Ljava.lang.Object; (contended, group = 0) @288 --- instance fields end --- @288 --- instance ends ---
C, 注解多个字段使他们分别被填充:
public static class ContendedTest4 { @Contended private Object contendedField1; @Contended private Object contendedField2; private Object plainField3; private Object plainField4; }
被注解的2个字段都被独立地填充:
TestContended$ContendedTest4: field layout @ 12 --- instance fields start --- @ 12 "plainField3" Ljava.lang.Object; @ 16 "plainField4" Ljava.lang.Object; @148 "contendedField1" Ljava.lang.Object; (contended, group = 0) @280 "contendedField2" Ljava.lang.Object; (contended, group = 0) @416 --- instance fields end --- @416 --- instance ends ---
在有些cases中,你会想对字段进行分组,同一组的字段会和其他字段有访问冲突,但是和同一组的没有。例如,(同一个线程的)代码同时更新2个字段是很常见的情况。如果同时把2个字段都添加@Contended注解是足够的(翻译注:但是太足够了),但我们可以通过去掉他们之间的填充,来优化它们的内存空间占用。为了区分组,我们有一个参数“contention group”来描述:
所以:
public static class ContendedTest5 { @Contended("updater1") private Object contendedField1; @Contended("updater1") private Object contendedField2; @Contended("updater2") private Object contendedField3; private Object plainField5; private Object plainField6; }
内存布局是:
TestContended$ContendedTest5: field layout @ 12 --- instance fields start --- @ 12 "plainField5" Ljava.lang.Object; @ 16 "plainField6" Ljava.lang.Object; @148 "contendedField1" Ljava.lang.Object; (contended, group = 12) @152 "contendedField2" Ljava.lang.Object; (contended, group = 12) @284 "contendedField3" Ljava.lang.Object; (contended, group = 15) @416 --- instance fields end --- @416 --- instance ends ---
注意$contendedField1 和$contendedField2和其他字段之间有填充,但是它们之间是紧挨着的。
以上是对邮件组中大牛们原始实现解释的翻译。
下面我们来做一个测试,看@Contended在字段级别,并且带分组的情况下,是否能解决伪缓存问题。
import sun.misc.Contended; public class VolatileLong { @Contended("group0") public volatile long value1 = 0L; @Contended("group0") public volatile long value2 = 0L; @Contended("group1") public volatile long value3 = 0L; @Contended("group1") public volatile long value4 = 0L; }
我们用2个线程来修改字段--
测试1:线程0修改value1和value2;线程1修改value3和value4;他们都在同一组中。
测试2:线程0修改value1和value3;线程1修改value2和value4;他们在不同组中。
测试1:
public final class FalseSharing implements Runnable { public final static long ITERATIONS = 500L * 1000L * 1000L; private static VolatileLong volatileLong; private String groupId; public FalseSharing(String groupId) { this.groupId = groupId; } public static void main(final String[] args) throws Exception { // Thread.sleep(10000); System.out.println("starting...."); volatileLong = new VolatileLong(); final long start = System.nanoTime(); runTest(); System.out.println("duration = " + (System.nanoTime() - start)); } private static void runTest() throws InterruptedException { Thread t0 = new Thread(new FalseSharing("t0")); Thread t1 = new Thread(new FalseSharing("t1")); t0.start(); t1.start(); t0.join(); t1.join(); } public void run() { long i = ITERATIONS + 1; if (groupId.equals("t0")) { while (0 != --i) { volatileLong.value1 = i; volatileLong.value2 = i; } } else if (groupId.equals("t1")) { while (0 != --i) { volatileLong.value3 = i; volatileLong.value4 = i; } } } }
测试2:(基于以上代码修改下面的部分)
public void run() { long i = ITERATIONS + 1; if (groupId.equals("t0")) { while (0 != --i) { volatileLong.value1 = i; volatileLong.value3 = i; } } else if (groupId.equals("t1")) { while (0 != --i) { volatileLong.value2 = i; volatileLong.value4 = i; } } }
测试1:
starting....
duration = 16821484056
测试2:
starting....
duration = 39191867777
可以看出,如果同一线程修改的是同一“contention group”中的字段,没有伪共享冲突,比有伪共享冲突的情况要快1倍多。
后记:
测试3:不使用@Contended
public class VolatileLong { public volatile long value1 = 0L; public volatile long value2 = 0L; public volatile long value3 = 0L; public volatile long value4 = 0L; }
结果:
starting....
duration = 38096777198
参考:
http://openjdk.java.net/jeps/142
http://mail.openjdk.java.net/pipermail/hotspot-dev/2012-November/007309.html