On Mon, 13 Jun 2022 22:37:52 GMT, Mandy Chung <mch...@openjdk.org> wrote:
>> Xue-Lei Andrew Fan has updated the pull request with a new target base due
>> to a merge or a rebase. The pull request now contains ten commits:
>>
>> - Merge
>> - Merge master
>> - Merge
>> - add timeout parameter
>> - rollback is not in this branch
>> - rollback JDK-8287384
>> - back to wait 1 second
>> - Remove trailing white space
>> - 8287596: Reorg jdk.test.lib.util.ForceGC
>
> test/jdk/java/lang/invoke/defineHiddenClass/UnloadingTest.java line 224:
>
>> 222:
>> 223: boolean tryUnload() {
>> 224: return ForceGC.wait(() -> weakRef.get() == null, 1000L);
>
> This verifies that the reference will never become weakly unreachable, i.e.
> it remains strongly reachable.
>
> I suggest to add a method default to 200ms (?) timeout scaled with the
> timeout factor and expect the specified condition will never met. Maybe
> call it `ForceGC.neverTrue(BooleanSupplier)`? Need to tune the base waiting
> time; for example if timeout factor is 4, then total wait would be 800 ms.
I see your points.
When I changed the waiting time to 1 second scaled with timeout factor, it
looks like safe to use the ForceGC.wait() method for either cleanup and
non-cleanup checking. The default JTREG timeout for a test case is 120
seconds. 1 seconds may be small enough so that we could use it for both cases.
-------------
PR: https://git.openjdk.org/jdk/pull/8979
