[jdk11u-dev] RFR: 8207200: Committed > max memory usage when getting MemoryUsage

[Goetz Lindenmaier]

On Mon, 20 Nov 2023 15:23:10 GMT, Severin Gehwolf <sgehwolf at openjdk.org> wrote:

>> I backport this for parity with 11.0.22-oracle.
>
> This seems a fairly risky backport for a somewhat minor issue that it tries to solve. Are you sure this one is appropriate for  OpenJDK 11u in its current lifecyce? Also, I agree with @RealCLanger on [his comment ](https://bugs.openjdk.org/browse/JDK-8207200?focusedId=14390322&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-14390322) on the bug and what he said, hasn't changed IMO.

Hi @jerboaa 
There are a few points to this. 
First, I don't think the issue is that minor. As I understand it, the MemoryUsage reported is always bogus since the values can be updated concurrently. It only became apparent when committed exceeded max.
Further, two things have changed since 2020:
Oracle obviously considers the risk acceptable.
OpenJDK falls back behind Oracle JDK and reports wrong values. 

Finally, others asked for this, too: Man Cao in https://bugs.openjdk.org/browse/JDK-8209061

If you agree, I would ask Andrew to backport the four changes involved individually and ask Richard, our GC expert, to review them.
What do you think?
(https://bugs.openjdk.org/browse/JDK-8208498, https://bugs.openjdk.org/browse/JDK-8209061, https://bugs.openjdk.org/browse/JDK-8209062, https://bugs.openjdk.org/browse/JDK-8207200)

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

PR Comment: https://git.openjdk.org/jdk11u-dev/pull/2281#issuecomment-1833816799