Re: RFR(s): 8243451: nsk.share.jdi.Debugee.isJFR_active() is incorrect and corresponsing logic seems to be broken

[serguei.spit...@oracle.com]

Hi Fairoz,

It looks okay in general.
But I'm not sure this check is going to work.
The problem is the HeapwalkingDebuggee.useStrictCheck method is invoked in the
context of the HeapwalkingDebugger process, not the HeapwalkingDebuggee process.
I'm not sure, what exactly you wanted to do here.
It can occasionally work for you as long as both processes are run with the same options.

Thanks,
Serguei


On 6/1/20 08:52, Fairoz Matte wrote:

Hi Erik,

Thanks for the review, below is the updated webrev.
http://cr.openjdk.java.net/~fmatte/8243451/webrev.02/

Thanks,
Fairoz

-----Original Message-----
From: Erik Gahlin
Sent: Monday, June 1, 2020 4:26 PM
To: Fairoz Matte <fairoz.ma...@oracle.com>
Cc: serviceability-dev@openjdk.java.net
Subject: Re: RFR(s): 8243451: nsk.share.jdi.Debugee.isJFR_active() is incorrect
and corresponsing logic seems to be broken

Hi Fairoz,

What I think you need to do is something like this:

         if (className.equals("java.lang.Thread")) {
             return !isJfrInitialized();
         }

...

     private static boolean isJfrInitialized() {
         try {
             Class<?> clazz = Class.forName("jdk.jfr.FlightRecorder");
             Method method = clazz.getDeclaredMethod("isInitialized",
new Class[0]);
             return (boolean) method.invoke(null, new Object[0]);
         } catch (Exception e) {
             return false;
         }
     }

Erik

On 2020-06-01 12:30, Fairoz Matte wrote:

Hi Erik,

Thanks for your quick response,
Below is the updated webrev to handle if jfr module is not present
http://cr.openjdk.java.net/~fmatte/8243451/webrev.01/

Thanks,
Fairoz

-----Original Message-----
From: Erik Gahlin
Sent: Monday, June 1, 2020 2:31 PM
To: Fairoz Matte <fairoz.ma...@oracle.com>
Cc: serviceability-dev@openjdk.java.net
Subject: Re: RFR(s): 8243451: nsk.share.jdi.Debugee.isJFR_active() is
incorrect and corresponsing logic seems to be broken

Hi Fairoz,

If the test needs to run with builds where the JFR module is not
present(?), you need to do the check using reflection.

If not, looks good.

Erik

On 1 Jun 2020, at 10:27, Fairoz Matte <fairoz.ma...@oracle.com> wrote:

Hi,

Please review this small test infra change to identify at runtime
the JFR is

active or not.

JBS - https://bugs.openjdk.java.net/browse/JDK-8243451
Webrev - http://cr.openjdk.java.net/~fmatte/8243451/webrev.00/

Thanks,
Fairoz