RFA 8056248: Improve ForkJoin thread throttling

[Paul Sandoz]

Hi,

After some off-list discussion and contemplation a implementation note is considered the best way forward (a middle ground between saying nothing and updating the spec). 

Here are the proposed doc updates for review to 8u (which requires a CCC) that would be applied on top the the 9 patch:

diff -r 7b40710067fa src/share/classes/java/util/concurrent/ForkJoinPool.java
--- a/src/share/classes/java/util/concurrent/ForkJoinPool.java	Fri Sep 05 10:54:28 2014 +0200
+++ b/src/share/classes/java/util/concurrent/ForkJoinPool.java	Thu Sep 11 14:35:00 2014 +0200
@@ -143,9 +143,6 @@
  * - the class name of a {@link ForkJoinWorkerThreadFactory}
  * <li>{@code java.util.concurrent.ForkJoinPool.common.exceptionHandler}
  * - the class name of a {@link UncaughtExceptionHandler}
- * <li>{@code java.util.concurrent.ForkJoinPool.common.maximumSpares}
- * - the maximum number of allowed extra threads to maintain target
- * parallelism (default 256).
  * </ul>
  * If a {@link SecurityManager} is present and no factory is
  * specified, then the default pool uses a factory supplying
@@ -162,6 +159,15 @@
  * pools with greater than the maximum number result in
  * {@code IllegalArgumentException}.
  *
+ * <p>This implementation supports an additional common pool parameter that may
+ * be controlled by setting the following
+ * {@linkplain System#getProperty system property}:
+ * <ul>
+ * <li>{@code java.util.concurrent.ForkJoinPool.common.maximumSpares}
+ * - the maximum number of allowed extra threads to maintain target
+ * parallelism (default 256).
+ * </ul>
+ *
  * <p>This implementation rejects submitted tasks (that is, by throwing
  * {@link RejectedExecutionException}) only when the pool is shut down
  * or internal resources have been exhausted.

Paul.
 
On Sep 5, 2014, at 11:35 AM, Paul Sandoz <Paul.Sandoz at oracle.com> wrote:

> Hi,
> 
> I would like to backport the improvements to the fork join framework from 9 to 8u40:
> 
>  http://mail.openjdk.java.net/pipermail/core-libs-dev/2014-August/028419.html
> 
>  https://bugs.openjdk.java.net/browse/JDK-8056248
> 
>  http://cr.openjdk.java.net/~psandoz/jdk9/JDK-8056248-fj-thread-throttling/webrev/
> 
>  http://hg.openjdk.java.net/jdk9/dev/jdk/rev/7af64e3e095d
> 
> The changeset (when shuffled) imports cleanly into jdk8u-dev/jdk. 
> 
> However, it also introduced a new system property ("java.util.concurrent.ForkJoinPool.common.maximumSpares"). I am not sure if we can add such a new feature to an 8u release, any advice? If that is not possible I suggest that reference to that property be removed from the JavaDoc, since that makes it easier to support further backports:
> 
> diff -r a6403166fc7f src/share/classes/java/util/concurrent/ForkJoinPool.java
> --- a/src/share/classes/java/util/concurrent/ForkJoinPool.java	Fri Sep 05 10:54:28 2014 +0200
> +++ b/src/share/classes/java/util/concurrent/ForkJoinPool.java	Fri Sep 05 11:26:49 2014 +0200
> @@ -143,9 +143,6 @@
>  * - the class name of a {@link ForkJoinWorkerThreadFactory}
>  * <li>{@code java.util.concurrent.ForkJoinPool.common.exceptionHandler}
>  * - the class name of a {@link UncaughtExceptionHandler}
> - * <li>{@code java.util.concurrent.ForkJoinPool.common.maximumSpares}
> - * - the maximum number of allowed extra threads to maintain target
> - * parallelism (default 256).
>  * </ul>
>  * If a {@link SecurityManager} is present and no factory is
>  * specified, then the default pool uses a factory supplying
> 
> 
> Thanks,
> Paul.