RFR: 8311986: runtime/os/TestTracePageSizes.java fails for ShenandoahGC

Thomas Stuefe stuefe at openjdk.org
Fri Jul 14 09:37:15 UTC 2023


On Fri, 14 Jul 2023 09:07:47 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

> I think it has to do with https://bugs.openjdk.org/browse/JDK-8310110 -- the test passes before that commit, and fails after it.

Yes, before https://bugs.openjdk.org/browse/JDK-8310110 we did not output anything for Shenandoah, so this test was just is a noop since the Matcher does not match anything to check.

I now remember again. Underlying cause is that AUX bitmap is madviced incorrectly for Shenandoah. 

See https://github.com/openjdk/jdk/pull/14486#issuecomment-1598871634 . 

I have a line of patches upcoming, starting with https://github.com/openjdk/jdk/pull/14739, then https://github.com/openjdk/jdk/pull/14559, and the AUX bitmap issue must wait at least until 14739 is in. Then, finally, this test can be enabled properly for Shenandoah.

Its all sideline stuff though, since my main focus is Lilliput and native heap trimming. Its also Reviewer-bound - the THP detection fix alone is already open since > 2 weeks.

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

PR Comment: https://git.openjdk.org/jdk/pull/14858#issuecomment-1635582330


More information about the hotspot-runtime-dev mailing list