[test] [11u] RFR: 8258414: OldObjectSample events too expensive

[Florian David]

Hi,
I'm resending this as a standalone email since some people do not seem to
receive it through the original mail thread.

-------------
Hi all,

Please find an update to the backport. Comments related to
https://mail.openjdk.java.net/pipermail/jdk-updates-dev/2021-April/005880.html
where the previous patch has broken the 11u build have been taken into
account:

- check_java_thread_in_native is replaced by check_java_thread_in_vm in
ObjectSampleCheckpoint::on_rotation
- ThreadInVMfromNative transition(thread) is removed. That check is part of
the original PR for JDK17 but since
some code in JFR was moved from JNI (in 11) to native in 16, this check is
not valid here.

Build and Tests:
- Linux release and fastdebug builds. Tested with make run-test-tier1 &&
make run-test TEST="jtreg:jdk/jfr/".
- Windows release and fastdebug builds. Tested with make run-test-tier1 &&
make run-test TEST="jtreg:jdk/jfr/". The
run-test-tier1 test suite did not complete entirely because of some issues
with paths being too long and
"vswhere.exe" not able to be run by the test on my machine. I would feel
more confident (especially Windows was broken
by the previous backport) if a maintainer could run the test suite on
Windows. The "jtreg:jdk/jfr/" test suite ran successfully.


Bug: https://bugs.openjdk.java.net/browse/JDK-8258414
webrev: http://cr.openjdk.java.net/~fdavid/8258414/webrev.03/
Original PR: https://github.com/openjdk/jdk/pull/2780

Thanks,
Florian DAVID