On Thu, 1 Oct 2026 23:08:42 GMT, Serguei Spitsyn <[email protected]> wrote:
> When the JVMTI `IterateThroughHeap` has a class filter then the flattened > value objects contained in the fields of filtered class objects are not > reported. The fix is: > - If the object has flattened fields, visit and report them as heap objects > even though this object was not reported. > > The update adds new test with a needed coverage: > > `test/hotspot/jtreg/serviceability/jvmti/valhalla/Heapwalking/FlatClassFilterTest.java` > > Testing: > - check new test is failing without fix and is passing with the fix > - TBD: run mach5 tiers 1-6 > > --------- > - [x] I confirm that I make this contribution in accordance with the [OpenJDK > Interim AI Policy](https://openjdk.org/legal/ai). test/hotspot/jtreg/serviceability/jvmti/valhalla/Heapwalking/FlatClassFilterTest.java line 79: > 77: Value[] arr = > (Value[])ValueClass.newNullableAtomicArray(Value.class, 5); > 78: for (int i = 0; i < arr.length; i++) { > 79: arr[i] = i == 2 ? null : new Value(seed + 10 + i); Suggestion: arr[i] = (i == 2 ? null : new Value(seed + 10 + i)); test/hotspot/jtreg/serviceability/jvmti/valhalla/Heapwalking/libFlatClassFilterTest.cpp line 54: > 52: callbacks.heap_iteration_callback = heap_iteration; > 53: > 54: err = jvmti->IterateThroughHeap(0, filter, &callbacks, nullptr); Have you considered increasing test coverage by also testing with JVMTI_HEAP_FILTER_CLASS_UNTAGGED in a separate test case? ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/33172#discussion_r4200198178 PR Review Comment: https://git.openjdk.org/jdk/pull/33172#discussion_r4200982479
