8369019: Improve timed-park mechanism in ObjectMonitor for virtual thread support #27597

pchilano · 2025-10-01T22:13:58Z

Please review the following fix. When blocking in ObjectMonitor::enter_internal we currently use timed-park for pinned virtual threads. This is done to alleviate some potential deadlocks cases where the successor is an unmounted virtual thread that cannot run. In particular this could happen during class loading/initialization if all other carriers are blocked waiting for the same class to be loaded/initialized.
This mechanism should be extended to cover ObjectMonitor::reenter_internal used in Object.wait (notification case). Also, the criteria to decide whether to do a timed-park should be based on whether there are unmounted vthreads already in the _entry_list, and not just if this is a pinned virtual thread. This covers mixed usages of the same ObjectMonitor between virtual threads and platform threads. This will become more relevant once we bring the changes currently in the fibers branch to preempt virtual threads during klass initialization.

These changes have been running in the loom pipeline for a couple of months already. I also added a new test case to test/jdk/java/lang/Thread/virtual/MonitorWaitNotify.java which deadlocks without these changes.

Thanks,
Patricio

Progress

Change must be properly reviewed (1 review required, with at least 1 Reviewer)
Change must not contain extraneous whitespace
Commit message must refer to an issue

Issue

JDK-8369019: Improve timed-park mechanism in ObjectMonitor for virtual thread support (Enhancement - P4)

Reviewers

David Holmes (@dholmes-ora - Reviewer)
Alan Bateman (@AlanBateman - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/27597/head:pull/27597
$ git checkout pull/27597

Update a local copy of the PR:
$ git checkout pull/27597
$ git pull https://git.openjdk.org/jdk.git pull/27597/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 27597

View PR using the GUI difftool:
$ git pr show -t 27597

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/27597.diff

Using Webrev

Link to Webrev Comment

bridgekeeper · 2025-10-01T22:15:53Z

👋 Welcome back pchilanomate! A progress list of the required criteria for merging this PR into master will be added to the body of your pull request. There are additional pull request commands available for use with this pull request.

openjdk · 2025-10-01T22:16:41Z

@pchilano This change now passes all automated pre-integration checks.

ℹ️ This project also has non-automated pre-integration requirements. Please see the file CONTRIBUTING.md for details.

After integration, the commit message for the final commit will be:

8369019: Improve timed-park mechanism in ObjectMonitor for virtual thread support

Reviewed-by: dholmes, alanb

You can use pull request commands such as /summary, /contributor and /issue to adjust it as needed.

At the time when this comment was updated there had been 167 new commits pushed to the master branch:

9b99bc8: 8369440: Remove RootResolverMarkScope and RootSetClosureMarkScope
2dfe458: 8369490: Remove unused Runinfo parameters in compiler/c2/gvn/TestBitCompressValueTransform.java
7e29d29: 8369097: Parallel: Support allocation in from space when heap is too full
... and 164 more: https://git.openjdk.org/jdk/compare/89af6e13f2354d6e32872791d157144cd478a88f...master

As there are no conflicts, your changes will automatically be rebased on top of these commits when integrating. If you prefer to avoid this automatic rebasing, please check the documentation for the /integrate command for further details.

➡️ To integrate this PR with the above commit message to the master branch, type /integrate in a new comment.

openjdk · 2025-10-01T22:17:21Z

@pchilano The following labels will be automatically applied to this pull request:

core-libs
hotspot-runtime

When this pull request is ready to be reviewed, an "RFR" email will be sent to the corresponding mailing lists. If you would like to change these labels, use the /label pull request command.

pchilano · 2025-10-01T22:43:25Z

/label remove core-libs

openjdk · 2025-10-01T22:44:08Z

@pchilano
The core-libs label was successfully removed.

mlbridge · 2025-10-01T23:38:23Z

Webrevs

dholmes-ora · 2025-10-02T05:32:08Z

src/hotspot/share/runtime/objectMonitor.cpp

-  // all other carriers have a vthread pinned to it waiting for said class
-  // to be loaded/initialized.
+  // If there are unmounted virtual threads ahead in the _entry_list we want
+  // to do a timed-park instead to alleviate some deadlocks cases where one


Suggested change

// to do a timed-park instead to alleviate some deadlocks cases where one

// to do a timed-park instead to alleviate some deadlock cases where one

dholmes-ora · 2025-10-02T05:33:27Z

src/hotspot/share/runtime/objectMonitor.cpp

  assert_mark_word_consistency();

+  // If there are unmounted virtual threads ahead in the _entry_list we want
+  // to do a timed-park instead to alleviate some deadlocks cases where one


Suggested change

// to do a timed-park instead to alleviate some deadlocks cases where one

// to do a timed-park instead to alleviate some deadlock cases where one

dholmes-ora · 2025-10-02T05:36:14Z

src/hotspot/share/runtime/objectMonitor.cpp

+  // _entry_list uses Atomic::cmpxchg() which already provides a fence that
+  // prevents this load from floating up previous store.
+  // Note that we can have false positives where timed-park is not necessary.
+  bool do_timed_parked = has_unmounted_vthreads();


Don't we still only need the timed-park if the current thread is a pinned vthread?

Yes, except if the monitor is also used in the context of a carrier thread. Currently there are only very few such cases and we disable preemption for them (e.g. interruptLock), so it’s likely not needed. With the upcoming changes to preempt on klass initialization, we could also have this situation if a class can be initialized both in the context of a carrier and a vthread. Since code executed in the context of the carriers is limited to library code there will also be very few cases of this, but I’ve seen at least one such case with LockSupport.

So you are saying the current code is insufficient and could still deadlock?

If there’s a monitor that is used both in the context of a virtual thread and the carriers then potentially yes. But this should almost never happen, and for the very few special cases we identified we currently disable preemption. So this is mostly to cover for the upcoming changes in case there is a class that can be initialized in the context of both a virtual thread and the carriers. Again this should also be a rare case but I’ve seen at least one case.

coleenp · 2025-10-03T19:02:23Z

src/hotspot/share/runtime/objectMonitor.cpp

+  // the notifier in notify_internal.
+  // Note that we can have false positives where timed-park is not necessary.
+  bool do_timed_parked = has_unmounted_vthreads();
+  static int MAX_RECHECK_INTERVAL = 1000;


Is this a constant? This is the same as the enter case, should there be only one?

Yes, I moved it to a global static.

AlanBateman · 2025-10-06T16:46:57Z

test/jdk/java/lang/Thread/virtual/MonitorWaitNotify.java

+     * paths and notification is done using notifyAll.
+     */
+    @Test
+    void testMixedPinnedUnmounted() throws Exception {


What would you think of testing timed-wait too? Some of the other tests are paramerized with a value source and some time values to test both untimed and timed waits.

AlanBateman · 2025-10-06T16:48:35Z

test/jdk/java/lang/Thread/virtual/MonitorWaitNotify.java


+    /**
+     * Test no deadlock happens when Object.wait is called from a mix of pinned and non-pinned
+     * paths and notification is done using notifyAll.


At some point then maybe we should combine this with RetryMonitorEnterWhenPinned. I'm not suggesting we do this now but some of the expanded description might be useful to include here as a passing reader might not immediately know what this test is doing.

I wasn’t sure where to put the extra test and missed RetryMonitorEnterWhenPinned.java. I agree it makes more sense to have it there. Moved now, let me know what you think.

Thanks for moving to RetryMonitorEnterWhenPinned as it is similar to the test we had for enter. The update, and conversion to JUnit look good.

openjdk · 2025-10-06T22:31:13Z

@pchilano core-libs has been added to this pull request based on files touched in new commit(s).

dholmes-ora · 2025-10-07T04:03:40Z

src/hotspot/share/runtime/objectMonitor.cpp

+  // _entry_list uses Atomic::cmpxchg() which already provides a fence that
+  // prevents this load from floating up previous store.
+  // Note that we can have false positives where timed-park is not necessary.
+  bool do_timed_parked = has_unmounted_vthreads();


So you are saying the current code is insufficient and could still deadlock?

dholmes-ora · 2025-10-07T04:08:50Z

src/hotspot/share/runtime/objectMonitor.inline.hpp

 }

+inline void ObjectMonitor::inc_unmounted_vthreads() {
+  assert(_unmounted_vthreads >= 0, "");


Suggested change

assert(_unmounted_vthreads >= 0, "");

assert(_unmounted_vthreads >= 0, "invariant");

Here and below - thanks.

dholmes-ora

Okay - thanks for clarifying.

pchilano · 2025-10-13T14:06:55Z

Thanks for the reviews and comments @dholmes-ora, @AlanBateman and @coleenp!
/integrate

openjdk · 2025-10-13T14:07:46Z

Going to push as commit 9feb8f2.
Since your change was applied there have been 178 commits pushed to the master branch:

0496806: 8359412: Template-Framework Library: Operations and Expressions
d47e6b7: 8369469: Rdtsc: Remove potential races in Rdtsc::initialize
d278043: 8367772: Refactor createUI in PassFailJFrame
... and 175 more: https://git.openjdk.org/jdk/compare/89af6e13f2354d6e32872791d157144cd478a88f...master

Your commit was automatically rebased without conflicts.

openjdk · 2025-10-13T14:07:57Z

@pchilano Pushed as commit 9feb8f2.

💡 You may see a message that your pull request was closed with unmerged commits. This can be safely ignored.

v1

0168c88

openjdk bot added hotspot-runtime [email protected] core-libs [email protected] labels Oct 1, 2025

openjdk bot removed the core-libs [email protected] label Oct 1, 2025

pchilano marked this pull request as ready for review October 1, 2025 23:33

openjdk bot added the rfr Pull request is ready for review label Oct 1, 2025

dholmes-ora reviewed Oct 2, 2025

View reviewed changes

Fix typos

398712a

coleenp reviewed Oct 3, 2025

View reviewed changes

make MAX_RECHECK_INTERVAL global static

e60c375

AlanBateman reviewed Oct 6, 2025

View reviewed changes

Move test to RetryMonitorEnterWhenPinned.java and add timed-variants

bf3613e

openjdk bot added the core-libs [email protected] label Oct 6, 2025

dholmes-ora reviewed Oct 7, 2025

View reviewed changes

Add string in asserts

9e435e0

dholmes-ora approved these changes Oct 8, 2025

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Oct 8, 2025

AlanBateman approved these changes Oct 12, 2025

View reviewed changes

openjdk bot added the integrated Pull request has been integrated label Oct 13, 2025

openjdk bot closed this Oct 13, 2025

openjdk bot removed the ready Pull request is ready to be integrated label Oct 13, 2025

openjdk bot removed the rfr Pull request is ready for review label Oct 13, 2025

	// to do a timed-park instead to alleviate some deadlocks cases where one
	// to do a timed-park instead to alleviate some deadlock cases where one

	assert(_unmounted_vthreads >= 0, "");
	assert(_unmounted_vthreads >= 0, "invariant");

8369019: Improve timed-park mechanism in ObjectMonitor for virtual thread support #27597

8369019: Improve timed-park mechanism in ObjectMonitor for virtual thread support #27597

Conversation

pchilano commented Oct 1, 2025 • edited by openjdk bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Progress

Issue

Reviewers

Reviewing

Uh oh!

bridgekeeper bot commented Oct 1, 2025

Uh oh!

openjdk bot commented Oct 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

openjdk bot commented Oct 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pchilano commented Oct 1, 2025

Uh oh!

openjdk bot commented Oct 1, 2025

Uh oh!

mlbridge bot commented Oct 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

pchilano Oct 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

openjdk bot commented Oct 6, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dholmes-ora left a comment

Choose a reason for hiding this comment

Uh oh!

pchilano commented Oct 13, 2025

Uh oh!

openjdk bot commented Oct 13, 2025

Uh oh!

openjdk bot commented Oct 13, 2025

Uh oh!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

4 participants

pchilano commented Oct 1, 2025 •

edited by openjdk bot

Loading

openjdk bot commented Oct 1, 2025 •

edited

Loading

openjdk bot commented Oct 1, 2025 •

edited

Loading

mlbridge bot commented Oct 1, 2025 •

edited

Loading

pchilano Oct 6, 2025 •

edited

Loading