On Sat, 28 Mar 2026 00:53:41 GMT, Chris Plummer <[email protected]> wrote:
>> Although SA has some FlatArrayKlass support, it is mostly incomplete. This >> is causing heap walking to abort when a flattened array of value objects is >> encountered. 4 SA tests are failing in non-obvious ways because of this. >> >> [JDK-8380769](https://bugs.openjdk.org/browse/JDK-8380769) [lworld] >> resourcehogs/serviceability/sa/TestHeapDumpForLargeArray.java fails with >> 'truncating to' missing from stdout/stderr with --enable-preview >> [JDK-8380851](https://bugs.openjdk.org/browse/JDK-8380851) [lworld] >> serviceability/sa/ClhsdbJstackWithConcurrentLock.java fails with "', (a >> java/util/concurrent/locks/ReentrantLock$NonfairSync)' not found in jstack >> output" when using --enable-preview >> [JDK-8380852](https://bugs.openjdk.org/browse/JDK-8380852) [lworld] >> serviceability/sa/CDSJMapClstats.java fails with "'BootClassLoader' missing >> from stdout/stderr" with "-XX:-UseCompressedOops --enable-preview" >> [JDK-8380853](https://bugs.openjdk.org/browse/JDK-8380853) [lworld] >> serviceability/sa/ClhsdbJhisto.java fails with "'java.nio.HeapByteBuffer' >> missing from stdout/stderr" with "-XX:-UseCompressedOops --enable-preview" >> >> The main shortcoming is that there is no FlatArray type in SA yet, and also >> the code to instantiate it is missing in the following: >> >> >> public Oop newOop(OopHandle handle) { >> ... >> Klass klass = Oop.getKlassForOopHandle(handle); >> if (klass != null) { >> if (klass instanceof TypeArrayKlass) return new TypeArray(handle, >> this); >> + if (klass instanceof FlatArrayKlass) return new FlatArray(handle, this); >> if (klass instanceof ObjArrayKlass) return new ObjArray(handle, this); >> if (klass instanceof InstanceKlass) return new Instance(handle, this); >> } >> throw new UnknownOopException(handle.toString()); >> } >> >> >> Without the added code, this method throws UnknownOopException when it >> encounters a flattened array. That in turn causes the heap walking to stop >> scanning the region it is in, thinking that it has reached the top of the >> used part of the region. See ObjectHeap.iterateLiveRegions(). >> >> Tested tier1, tier2 svc, and tier5 svc CI testing with and w/o >> --enable-preview. Did tier5, tier6, and tier7 with --enable-preview on >> linux-x64-debug only. > > Chris Plummer has updated the pull request incrementally with one additional > commit since the last revision: > > minor comment updates Marked as reviewed by amenkov (Committer). ------------- PR Review: https://git.openjdk.org/valhalla/pull/2271#pullrequestreview-4032324567
