[crac] RFR: 8321660: [CRaC] Trim native heap before checkpoint to decrease image size

[Volker Simonis]

On Mon, 11 Dec 2023 11:48:40 GMT, Radim Vansa <rvansa at openjdk.org> wrote:

> Just a question: have you considered pros and cons trimming the heap inside the safepoint (VM_Crac)?

What would be the benefit of doing the trim inside the safepoint? Currently, native heap trimming is deliberately not being done at a safpoint (see [PR for JDK-8293114: JVM should trim the native heap](https://github.com/openjdk/jdk/pull/14781)). Now, I agree that `VM_Crac` is a special safoint in the sense that it is not returning anyway and thus the time it takes is irrelevant.  But still, I don't think we should move tasks into a safepoint unless they functionally require to be executed at a safepoint. What do you think?

Apart from that, I don't think most test failures (e.g. tier1 runtime/serviceablitly) are related to this change because I couldn't reproduce them locally.

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

PR Comment: https://git.openjdk.org/crac/pull/148#issuecomment-1850258653