[10] RFR for JDK-8170299: Debugger does not stop inside the low memory notifications code
Shafi Ahmad
shafi.s.ahmad at oracle.com
Mon Sep 4 10:11:00 UTC 2017
Hi,
Please review the fix for 'JDK-8170299: Debugger does not stop inside the low memory notifications code' to jdk10.
With the current implementation 'debugger does not stop inside the low memory *notifications* code'
This is expected as the notification code is getting executed with the service thread, which is a hidden thread.
jdk10-dev-new$ jdb -Xmx32m MemoryWarningSystem
Initializing jdb ...
> stop at MemoryWarningSystem$1:25
Deferring breakpoint MemoryWarningSystem$1:25.
It will be set after the class is loaded.
> run
run MemoryWarningSystem
Set uncaught java.lang.Throwable
Set deferred uncaught java.lang.Throwable
>
. . .
> quit
Breakpoint not hit.
With the attached webrev, jdb stop at breakpoint inside notification code.
jdk10-dev-new$ jdb -Xmx64m MemoryWarningSystem
Initializing jdb ...
> stop at MemoryWarningSystem$1:25
Deferring breakpoint MemoryWarningSystem$1:25.
It will be set after the class is loaded.
> run
run MemoryWarningSystem
Set uncaught java.lang.Throwable
Set deferred uncaught java.lang.Throwable
>
VM Started: Set deferred breakpoint MemoryWarningSystem$1:25
Breakpoint hit: "thread=Thread-0", MemoryWarningSystem$1.memoryUsageLow(), line=25 bci=0
25 System.out.println("Memory usage low!!!");
Thread-0[1] step
> Memory usage low!!!
Step completed: "thread=Thread-0", MemoryWarningSystem$1.memoryUsageLow(), line=26 bci=8
26 double percentageUsed = ((double) usedMemory) / maxMemory;
Thread-0[1] where
[1] MemoryWarningSystem$1.memoryUsageLow (MemoryWarningSystem.java:26)
[2] MemoryWarningSystem$2.handleNotification (MemoryWarningSystem.java:48)
[3] javax.management.NotificationBroadcasterSupport.handleNotification (NotificationBroadcasterSupport.java:284)
[4] javax.management.NotificationBroadcasterSupport$SendNotifJob.run (NotificationBroadcasterSupport.java:361)
[5] java.lang.Thread.run (Thread.java:844)
Thread-0[1] step
>
Step completed: "thread=Thread-0", MemoryWarningSystem$1.memoryUsageLow(), line=27 bci=15
27 System.out.println("percentageUsed = " + percentageUsed);
Thread-0[1] list
23 mws.addListener(new MemoryWarningSystem.Listener() {
24 public void memoryUsageLow(long usedMemory, long maxMemory) {
25 System.out.println("Memory usage low!!!");
26 double percentageUsed = ((double) usedMemory) / maxMemory;
27 => System.out.println("percentageUsed = " + percentageUsed);
28 MemoryWarningSystem.setPercentageUsageThreshold(0.8);
29 }
30 });
31
32 Collection<Double> numbers = new LinkedList<Double>();
Thread-0[1]
In the change, the class 'MemoryImpl' is derived from class 'NotificationBroadcasterSupport' instead of class 'NotificationEmitterSupport'
NotificationBroadcastSupport takes an Executor, MemoryImpl constructor call super() with an executor, a visible thread.
The method hasListeners() is referenced inside MemoryImpl.java and defined in NotificationEmitterSupport.
This method is not present in NotificationBroadcasterSupport so I added it to NotificationBroadcasterSupport.
Jdk10 bug: https://bugs.openjdk.java.net/browse/JDK-8170299
Webrev link: http://cr.openjdk.java.net/~shshahma/8170299/webrev.01/
Testing: jprt -testset core, vm tonga tests nsk.jvmti.testlist, nsk.jdi.testlist, nsk.jdwp.testlist and jtreg tests ./jdk/test/com/sun/jdi/
Please let me know if I missed to test some more test suit[s].
Thanks to Mandy and Poonam for helping me in fixing this.
Regards,
Shafi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/serviceability-dev/attachments/20170904/8ca0639e/attachment.html>
More information about the serviceability-dev
mailing list