RFR: 8287596: Reorg jdk.test.lib.util.ForceGC [v2]

[Xue-Lei Andrew Fan]

On Wed, 1 Jun 2022 21:07:16 GMT, Xue-Lei Andrew Fan <xuelei at openjdk.org> wrote:

>> This is a follow up update per comments in [JDK-8287384 PR](https://github.com/openjdk/jdk/pull/8907).  The tier1 and tier2 test in open part looks good to me.  Please help to run Mach5 just case the closed test cases are impacted.
>
> Xue-Lei Andrew Fan has updated the pull request incrementally with one additional commit since the last revision:
> 
>   back to wait 1 second

My macOS (M1) and Linux (CentOS) may be too powerful to reproduce the failure.  I tried to set the JTREG time out to 60 seconds, and the timeout issue could be reproduced.  In `test/jdk/java/lang/invoke/defineHiddenClass/UnloadingTest.java`, there are 6 calls to the method `assertFalse(unloader.tryUnload())`.  Each call to tryUnload() takes at least 10 seconds.  So the 6 calls takes 60 seconds at least.   If I set the regression timeout value to 70 seconds, the test still can pass.   It implies the rest part other than tryUnload() is pretty fast.

It looks like a regression introduced with the update for [JDK-8287384](https://bugs.openjdk.java.net/browse/JDK-8287384).  I did not fine the cause yet.  I will have more checking tomorrow.

-------------

PR: https://git.openjdk.java.net/jdk/pull/8979