On Wed, 2 Sep 2026 03:50:51 GMT, Shiv Shah <[email protected]> wrote:
> Gives objects that are only reachable from a parked virtual thread’s stack a
> root in the heap dump. the writer now treats an unmounted virtual thread the
> way the hotspot dumper does, a thread object root on the dummy stack trace
> plus a java frame root for each oop in the thread’s stack chunks, so tools
> stop reporting those objects as unreachable. the chunk walk uses the gc
> bitmap, so it covers chunks that have been through a gc, which is the long
> parked case, freshly parked chunks still need the frame walk. revptrs picks
> up the same walk, and the revptrs test now fails on any swallowed exception
>
> The new heap dump test fails on master with no root for the referenced object
> and passes with the change. serviceability and sun/tools groups pass on both
> thread factories.
>
>
> ---------
> - [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/sa/ClhsdbRevPtrsForVirtualThread.java line 29:
> 27: * @summary Test that clhsdb 'revptrs' finds references from a virtual
> thread's stack chunk
> 28: * @requires vm.hasSA
> 29: * @requires vm.gc.Serial
What are the issues with other GCs?
test/hotspot/jtreg/serviceability/sa/ClhsdbRevPtrsForVirtualThread.java line 66:
> 64: }
> 65:
> 66: String output = test.run(theApp.getPid(), List.of("revptrs "
> + addr), null, null);
Although it probably won't impact this test, we currently have a reverse
pointer bug https://bugs.openjdk.org/browse/JDK-8241235. Make sure you at least
run this test with -Xcomp.
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/32637#discussion_r3928218208
PR Review Comment: https://git.openjdk.org/jdk/pull/32637#discussion_r3928241364
