On Tue, 30 Jun 2026 09:14:35 GMT, Manuel Hässig <[email protected]> wrote:

> After the integration of #2565 three subtests of 
> `compiler/valhalla/inlinetypes/TestValueConstruction.java` started failing, 
> but only when `-XX:+StressUnstableIfTraps` was passed. This PR fixes this by 
> introducing new subtasks with a higher memlimit for the affected subtests 
> when `-XX:+StressUnstableIfTraps` is passed.
> 
> Alternatively, I could have simply raised the memlimit, but this would 
> diminish the capability of these tests to detect regressions in the 
> scalarization strategy.
> 
> Testing:
>  - [x] 50 repetitions of 
> `compiler/valhalla/inlinetypes/TestValueConstruction.java`
>  - [x] 50 repetitions of 
> `compiler/valhalla/inlinetypes/TestValueConstruction.java` with 
> `-XX:+StressUnstableIfTraps`
> 
> 
> ---------
> - [x] I confirm that I make this contribution in accordance with the [OpenJDK 
> Interim AI Policy](https://openjdk.org/legal/ai).

I also see failures without `-XX:+StressUnstableIfTraps` in our testing. I 
think we just need to increase the mem limit unconditionally again. At least 
temporarily.

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

PR Review: 
https://git.openjdk.org/valhalla/pull/2599#pullrequestreview-4601278622

Reply via email to