RFR: 8342580: GenShen: TestChurnNotifications fails executing in unintended test-id modes with ShenandoahGCMode=generational [v2]

[Y. Srinivas Ramakrishna]

On Wed, 23 Oct 2024 17:01:55 GMT, Y. Srinivas Ramakrishna <ysr at openjdk.org> wrote:

>> Generational mode should not be tested with explicit heuristics settings, which are silently ignored. The test failed in these settings because it needs to be told which memory pool to check for notifications, which needs to be specified differently for generational. Ideally, we'd have the test figure this out rather than needing this to be explicitly specified. For now I just tightened the `@requires` clause to run only in the intended modes.
>> 
>> Related cleanups will follow separately (such as emitting a warning when an incompatible heuristics option is specified, code which may have been lost in a previous refactor). See details in the ticket.
>> 
>> **Testing** : tested with no specific collector, with Shenandoah, withGenShen, and with other collectors with test explicitly specified. As intended, in the first and second case 6 test-ids are run, in the thirdd case 1 test-id is run, and in the remaining cases, 0 test-ids are run.
>
> Y. Srinivas Ramakrishna has updated the pull request with a new target base due to a merge or a rebase. The incremental webrev excludes the unrelated changes brought in by the merge/rebase. The pull request contains four additional commits since the last revision:
> 
>  - Merge branch 'master' into test_requires
>  - Merge branch 'master' into test_requires
>  - Don't test generational with heuristics it ignores. Separately, we'll
>    fix the code for generational such that the choice of a specific
>    heuristic elicits a warning.
>  - TestChurnNotifications fails when jtreg run with GenShen

Thanks for your reviews!

I'm landing this with the latest round of testing after the sync with master still in flight. (The previous round before that sync completed without failures.) I do expect this round of GHA tests to show up failures in TestReferenceRefersToShenandoah, which we are addressing in a sibling ticket (actually tracked in two different open tickets at the moment), and which is obviously unrelated to this change.

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

PR Comment: https://git.openjdk.org/shenandoah/pull/520#issuecomment-2432930481