RFR: JDK-4903717: nsk/jdi/ThreadReference/isSuspended/issuspended002 failing

[gary.adams at oracle.com]

The commented out "docontinue" was done to match the style
used in the other tests. I'll remove it for this test.

In any event, the test still is failing 8 / 500 testruns on
macosx and windows debug builds.

The change is not sufficient.

Here's a recent failure. You can see in the log that the debugger did
wait until the new "docontinue" arrives outside the lockingObject
synchronized block.

It appears with the main thread suspended and the thread2 double resumed,
it still does not continue executing up to the breakpoint in thread2.
After the eventQueue.remove(timeout), when the vm is resumed
you can see the debuggee messages expected leading up to the breakpoint.

...

--> debugger:       suspending VM with vm.suspend();
*debugee.stderr> **> mainThread: mainThread is out of: synchronized 
(lockingObject) {*
--> debugger:          thread2.suspendCount() == 2
--> debugger:      :   thread2.isSuspended()
--> debugger:          mainThread.suspendCount() == 1
--> debugger:      :   mainThread.isSuspended()
--> debugger:       double resuming thread2 with thread2.resume();
--> debugger:          thread2.suspendCount() == 1
--> debugger:      :   thread2.isSuspended()
--> debugger:      enabling breakpRequest1
--> debugger:          thread2.suspendCount() == 0
--> debugger:      :  !thread2.isAtBreakpoint() before call to breakpoint()
--> debugger:        waiting for BreakpointEvent
*--> debugger: new: eventSet = eventQueue.remove();*
# ERROR: ##> debugger: ERROR:  timeout for waiting for a BreakpintEvent
The following stacktrace is for failure analysis.
nsk.share.TestFailure: ##> debugger: ERROR:  timeout for waiting for a BreakpintEvent
	at nsk.share.Log.logExceptionForFailureAnalysis(Log.java:428)
	at nsk.share.Log.complain(Log.java:399)
	at nsk.jdi.ThreadReference.isSuspended.issuspended002.log3(issuspended002.java:104)
	at nsk.jdi.ThreadReference.isSuspended.issuspended002.breakpoint(issuspended002.java:555)
	at nsk.jdi.ThreadReference.isSuspended.issuspended002.runThis(issuspended002.java:389)
	at nsk.jdi.ThreadReference.isSuspended.issuspended002.run(issuspended002.java:90)
	at nsk.jdi.ThreadReference.isSuspended.issuspended002.main(issuspended002.java:85)
	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
	at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.base/java.lang.reflect.Method.invoke(Method.java:567)
	at PropertyResolvingWrapper.main(PropertyResolvingWrapper.java:104)
	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
	at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.base/java.lang.reflect.Method.invoke(Method.java:567)
	at com.sun.javatest.regtest.agent.MainWrapper$MainThread.run(MainWrapper.java:127)
	at java.base/java.lang.Thread.run(Thread.java:835)
*--> debugger: resuming both second and main thread debugee.stderr> **> 
mainThread: waiting for an instruction from the debugger ... --> 
debugger: the end of testing debugee.stderr> **> thread2: entered into 
block: synchronized (lockingObject) debugee.stderr> **> thread2: exited 
from block: synchronized (lockingObject) debugee.stderr> **> thread2: 
call to the method 'runt1' debugee.stderr> **> thread2: method 'runt1' enter*
--> debugger:      : returned string is 'checkend'
debugee.stderr> **> mainThread: waiting for an instruction from the debugger ...



On 2/20/19 5:15 PM, Alex Menkov wrote:
> +1
>
> But could you please remove old "pipe.println("docontinue")" statement 
> instead of commenting it out (no new webrev is required)
>
> --alex
>
> On 02/20/2019 09:58, Chris Plummer wrote:
>> Looks good.
>>
>> Chris
>>
>> On 2/20/19 7:57 AM, Gary Adams wrote:
>>> The issuspended002 has been on the ProblemList since the tests were 
>>> moved to
>>> the open repos. The proposed changeset applies the same fix that was 
>>> used in issuspended001.
>>>
>>> The current test fails when the debuggee replies with the 
>>> "docontinue" response,
>>> while it is still holding a lock in a syncrhonized block. When the 
>>> debugger then
>>> suspends the vm the debuggee test thread fails to proceed to the 
>>> expected breakpoint.
>>> Testing in progress.
>>>
>>>   Issue: https://bugs.openjdk.java.net/browse/JDK-4903717
>>>   Webrev: http://cr.openjdk.java.net/~gadams/4903717/webrev.00/
>>
>>
>>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.java.net/pipermail/serviceability-dev/attachments/20190221/dd2af67b/attachment-0001.html>