jmx-dev RFR: 8188784: javax/management/notification/BroadcasterSupportDeadlockTest.java - TEST FAILED: deadlock

[Leonid Mesnik]

On Tue, 9 Apr 2024 11:08:31 GMT, Kevin Walls <kevinw at openjdk.org> wrote:

> This test incorrectly fails, although rarely, thinking its "thread 2" has deadlocked.
> A change of sleep will likely fix this, but there are other issues, so cleaning up the test a little.
> 
> Remove the probe for the ManagementFactory class, to check we are on jdk5 or later. 8-)
> 
> When sleeping, sleep 100, not 1ms, we don't need to spin fast and actually race with the other thread.
> 
> We have a 1000 iteration loop, but don't seem to use it.  We only check once then either return (pass), fail, or break (which is also fail).  Use the loop to check for the status change, which is likely what was intended.
> 
> Show the stackframes on all failures.

Marked as reviewed by lmesnik (Reviewer).

test/jdk/javax/management/notification/BroadcasterSupportDeadlockTest.java line 122:

> 120:         java.util.Map<Thread,StackTraceElement[]> traces = Thread.getAllStackTraces();
> 121:         showStackTrace("Thread 1", traces.get(t1));
> 122:         showStackTrace("Thread 2", traces.get(t2));

Could you please replace System.exit() with throwing Exception. Other looks good.

-------------

PR Review: https://git.openjdk.org/jdk/pull/18687#pullrequestreview-2002384686
PR Review Comment: https://git.openjdk.org/jdk/pull/18687#discussion_r1566573453