On Thu, 27 Aug 2026 18:30:26 GMT, Stefan Karlsson <[email protected]> wrote:
>> Shiv Shah has updated the pull request incrementally with one additional >> commit since the last revision: >> >> enable gc logging in the test > > I think we need "no keepalive" validation in the checked wrapper. Thanks for spotting the issue @stefank ! Well this is a real deja-vu moment. We introduced the no-keep-alive for IsSameObject back in JDK 12 precisely because this test would fail with OOME. https://bugs.openjdk.org/browse/JDK-8215451 I just updated the JBS issue with a comment. I'm not sure we should be changing the checked behaviour just because of this one test. Whilst I agree that the checked logic does not need to keep the reference alive, it is normally not considered harmful (one GC cycle). Changing the code to use no-keepalive would involve introducing a new API, or making a couple of layers of template functions so that we could pass the desired Access DECORATOR through to the actual access. That seems a very heavyweight change to fix one test that is not normally run under -Xcheck:jni in the first place. ------------- PR Comment: https://git.openjdk.org/jdk/pull/32229#issuecomment-5448826650
