On 14/02/2018 02:58, Prasanta Sadhukhan wrote:
Maybe the robot mouse events are not getting enough time to be processed which is why the toolkit shutdown is not able to happen as those are still in the event queue. Adding waitForIdle() in the test allows those robot events to be processed and then toolkit thread gets chance to get disposed and system shutdown takes place ok.

Maybe this is the case, but robot just generates an events, it can be possible that the bug can be reproduced by the user and mouse(I guess the problem is in the timings). But even in this case the shutdown hook should not hang, because it block the whole jdk.

On 2/14/2018 12:51 PM, Sergey Bylokhov wrote:
Hi, Prasanta.
I think that the suspicious place in the stack is below, it tries to execute hooks during system shutdown and hangs.

The hook which hangs is "ToolkitShutdown #18 daemon" from the WToolkit class. I guess  that the code which hangs is a loop in the:
    while (!tk.IsDisposed()) {

So it does not look like a test bug.

On 13/02/2018 22:38, Prasanta Sadhukhan wrote:
"main" #1 prio=5 os_prio=0 tid=0x000001b83d8ad000 nid=0x3e84 in Object.wait() [0x0000003e642fe000]
    java.lang.Thread.State: WAITING (on object monitor)
     at java.lang.Object.wait(java.base@10-ea/Native Method)
     - waiting on <0x00000006d2880200> (a java.lang.Thread)
     at java.lang.Thread.join(java.base@10-ea/Thread.java:1353)
     - waiting to re-lock in wait() <0x00000006d2880200> (a java.lang.Thread)
     at java.lang.Thread.join(java.base@10-ea/Thread.java:1427)
     at java.lang.ApplicationShutdownHooks.runHooks(java.base@10-ea/ApplicationShutdownHooks.java:107)      at java.lang.ApplicationShutdownHooks$1.run(java.base@10-ea/ApplicationShutdownHooks.java:46)
     at java.lang.Shutdown.runHooks(java.base@10-ea/Shutdown.java:123)
     at java.lang.Shutdown.sequence(java.base@10-ea/Shutdown.java:167)
     at java.lang.Shutdown.exit(java.base@10-ea/Shutdown.java:212)

Best regards, Sergey.

Reply via email to