RFR: 8310489: New test runtime/ClassInitErrors/TestStackOverflowDuringInit.java failed

[Coleen Phillimore]

On Mon, 26 Jun 2023 06:38:52 GMT, David Holmes <dholmes at openjdk.org> wrote:

> I haven't been able to reproduce the failure outside of the occasional CI failures we have, but I have done some further analysis.
> 
> I've extracted the key aspects of the test and put them in a simplified "non fuzzed" form so that it is much easier to see what the test does and what it expects to happen. I've also tweak the running the test so that it uses the minimum amount of stack (so we get stackoverflow sooner) and runs in interpreted mode (so we eliminate any potential JIT oddities with inlining etc.). Hopefully this will be more stable - otherwise I will just delete the test.
> 
> Testing:
>  - 50x on each x64 platform in our CI
> 
> Thanks.

New test makes sense.  Old test code makes no sense.  Adding -Xint gives you a better chance of getting SOE.  Not guaranteed though but maybe this'll be more stable.

Since this is a test failure, it's ok to check this in with one review or soon anyway.

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

Marked as reviewed by coleenp (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/14648#pullrequestreview-1499672413
PR Comment: https://git.openjdk.org/jdk/pull/14648#issuecomment-1608437377