On Tue, 8 Apr 2025 21:19:06 GMT, Roger Riggs <rri...@openjdk.org> wrote:
>> Certain specific types of tests involving GC and reference processing need >> to account for the delay between a GC completing (during which the GC clears >> a Reference), and the Reference being added to its associated queue. At >> present, ad hoc mechanisms (with delays/timeout) are used, but can lead to >> intermittent test failures >> ([JDK-8298783](https://bugs.openjdk.org/browse/JDK-8298783) is a recent >> example). >> >> A better mechanism already exists in the private >> `Reference.waitForReferenceProcessing()` method. This PR makes >> `waitForReferenceProcessing()` available to tests via the `WhiteBox` and >> `ForceGC` test libraries. > > test/lib/jdk/test/whitebox/WhiteBox.java line 569: > >> 567: * This method should usually be called after a call to >> WhiteBox.fullGC(). >> 568: */ >> 569: public static void waitForReferenceProcessing() { > > Can the code duplication be avoided? That would be nice. A primary goal here is to not disrupt existing tests that use `WhiteBox` or `ForceGC`, (e.g. by necessitating additional tags, `@run` arguments, etc). I don't think there's a way for `WhiteBox.java` and `ForceGC.java` to both refer to code at some shared location without tests having to add a new tag - `@build`, I think. However I will look into it a bit further. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/24527#discussion_r2034139841
