RFR: 8326121: vmTestbase/gc/g1/unloading/tests/unloading_keepRef_rootClass_inMemoryCompilation_keep_cl failed with Full gc happened. Test was useless.

Thomas Schatzl tschatzl at openjdk.org
Wed May 29 06:56:03 UTC 2024


On Tue, 28 May 2024 15:51:54 GMT, Ivan Walulya <iwalulya at openjdk.org> wrote:

> Hi all,
> 
> Please review this change to trigger SkippedException instead of RunTimeExceptions when a test is deemed useless because of full-GCs. The test triggers frequent `System.gc()` calls which in some cases may aggressively shrink the heap, and subsequent allocations lead to full-GC. This may affect the usefulness of the test, but it is not a test failure. So we added  SkippedException to indicate this situation.
> 
> Testing: Local testing with very small heaps.
>               Tier 1.

Looks good, please fix the spacing problem.

Also, please update the CR with the evaluation in the PR description.

test/hotspot/jtreg/vmTestbase/gc/g1/unloading/UnloadingTest.java line 162:

> 160:             }
> 161:         }
> 162: //        if (WhiteBox.getWhiteBox().g1GetTotalFullCollections() != 0 || (oldGenBean != null && oldGenBean.getCollectionCount() != 0)) {

Maybe remove these System.out.println() here and above - your call.

test/hotspot/jtreg/vmTestbase/nsk/share/test/Tests.java line 87:

> 85:                                 if (o instanceof TestExitCode)
> 86:                                         exitCode = ((TestExitCode) o).getExitCode();
> 87:                         }catch (SkippedException se) {

Spacing.

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

Marked as reviewed by tschatzl (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/19432#pullrequestreview-2084453130
PR Review Comment: https://git.openjdk.org/jdk/pull/19432#discussion_r1618300332
PR Review Comment: https://git.openjdk.org/jdk/pull/19432#discussion_r1618299352


More information about the hotspot-gc-dev mailing list