On Sat, 13 Sep 2025 05:13:39 GMT, Jaikiran Pai <j...@openjdk.org> wrote:
> Can I please get a review of this test-only change which addresses the test > failure in > `sun/security/util/AlgorithmConstraints/InvalidCryptoDisabledAlgos.java`? > > As noted in https://bugs.openjdk.org/browse/JDK-8367583, this test fails when > run in our CI. The test intentionally sets up some Security property which is > expected to cause a `ExceptionInInitializerError` when a security API is > used. The test also uses the `jdk.test.lib.Utils` test library which in its > static block has a piece of code which intializes a `MessageDigest` > conditionally depending on whether or not the `build` part of the > `Runtime.version()` is greater than `0`. The initialization of > `MessageDigest` during the static initialization of the `Utils` class leads > to certain security APIs to be used too early and thus causing the > `ExceptionInInitializerError` much before the test expected. The test thus > ends up failing in that case. > > For local builds, build number is by default 0, so this failure wasn't > observed locally. In order to reproduce this issue, I built a local JDK with: > > > bash configure --with-version-build=42 > make clean images > > I then ran this test against this built JDK and it was able to reproduce this > failure. > > The change in this PR updates the test to no longer use the test library's > `Utils` class and thus avoids this issue. The test continues to test the same > APIs and expectations that it was previously doing, except that it does it > without the use of this test utility class. I have run this test change > locally against a JDK with build number 0 and against a JDK build number > greater than 0 and it passes against both. > > I'll launched a CI run just to make sure nothing else fails with this change. I've seen this before. https://bugs.openjdk.org/browse/JDK-8345578. ------------- Marked as reviewed by weijun (Reviewer). PR Review: https://git.openjdk.org/jdk/pull/27269#pullrequestreview-3220746033
