On Thu, 2 Apr 2026 14:36:01 GMT, Casper Norrbin <[email protected]> wrote:
> Hi everyone, > > The test `InlineTypeDensity.java` builds on incorrect assumptions that a null > restricted non-atomic value class is identical to a primitive type, which is > not necessarily the case. For example, the first element of an array has to > be aligned to 64 bits, while in a primitive type it does not. With COH > enabled, this causes the test to fail, as the value class becomes bigger than > its primitive equivalent. > > I have reworked this test to instead check that everything is flattened as > expected, and that the alignment of the first element is correctly 64 bits. > With these changes, I have also removed `vm.flagless` which makes us able to > run this test in more configurations. > > Testing: > - Tiers 1-3 > - Manual runs of the test with different flag combinations. > > --------- > - [x] I confirm that I make this contribution in accordance with the [OpenJDK > Interim AI Policy](https://openjdk.org/legal/ai). This pull request has now been integrated. Changeset: c8bdf176 Author: Casper Norrbin <[email protected]> Committer: Frederic Parain <[email protected]> URL: https://git.openjdk.org/valhalla/commit/c8bdf1767a40eb85172a040728968f775e0d5bbe Stats: 99 lines in 1 file changed: 23 ins; 47 del; 29 mod 8379614: [lworld] runtime/valhalla/inlinetypes/InlineTypeDensity.java fails with UCOH Reviewed-by: stefank, fparain ------------- PR: https://git.openjdk.org/valhalla/pull/2295
