On Wed, 3 Sep 2025 01:41:15 GMT, pf0n <d...@openjdk.org> wrote: >> test/jdk/jdk/jfr/event/gc/objectcount/TestObjectCountAfterGCEventWithShenandoah.java >> line 11: >> >>> 9: * & vm.opt.ExplicitGCInvokesConcurrent != true >>> 10: * @library /test/lib /test/jdk >>> 11: * @run main/othervm -XX:+UnlockExperimentalVMOptions >>> -XX:-UseFastUnorderedTimeStamps -XX:+UseShenandoahGC >>> -XX:MarkSweepDeadRatio=0 -XX:-UseCompressedOops >>> -XX:-UseCompressedClassPointers -XX:+IgnoreUnrecognizedVMOptions >>> jdk.jfr.event.gc.objectcount.TestObjectCountAfterGCEventWithShenandoah >> >> Are all of these flags necessary to run this test? Can we pare any >> unnecessary options? > > All of the collectors that test for the ObjectCountAfterGC event uses these > flags. I thought it would be best to keep it consistent for the Shenandoah > test.
Best practice is to remove flags that are no-ops, otherwise we accrue junk here. I'd urge removal of such cruft, in particular: MarkSweepDeadRatio, UseCompressedOops & UseCompressedClassPointers (which you may have added for debugging ease), and IgnoreUnrecognizedVMOptions (which should, I think, not be there in any tests unless there is an explicitly stated reason to include it). I am not sure we even need UseFastUnorderedTimeStamps either since the test does not, to my knowledge, examine the order of timestamped events in the JFR recording. If further constraint checking is added to the tests that requires this, it should be added back at that time. For the other collectors, we can consider a separate ticket to clean any unnecessary flags up (I suspect only MarkSweepDeadRatio may be needed for the cases where one is looking for a collection to necessarily move objects-- which I don't think these tests test for -- so even that flag appears unnecessary at least for these tests.) I vote for unburdening the flag load for the new Shenandoah test at least, as suggested by William. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/26977#discussion_r2320627227
