JDK-8316691: Heap dump: separate stack traces for mounted virtual threads #15869

alexmenkov · 2023-09-21T20:06:13Z

This is subtask of JDK-8299426: Heap dump does not contain virtual Thread stack references
The change:

reorganize thread-related code/prepare it to use for unmounted vthreads:
- new ThreadDumper class caches stack frames, thread serial num, frame serial number (trace serial number is calculated from thread serial);
  ThreadDumper objects for all platform/carrier and mounted virtual threads are cached instead of ThreadStackTrace objects (they are created during HPROF_FRAME/HPROF_TRACE dumping, used lated for writing HPROF_GC_ROOT_THREAD_OBJ/HPROF_GC_ROOT_JAVA_FRAME/HPROF_GC_ROOT_JNI_LOCAL subrecords);
- new helper class JavaStackRefDumper to dump references from threadf stack;
separate track traces for mounted virtual threads:
- separate HPROF_FRAME/HPROF_TRACE records for mounted vthreads and carrier threads;
- separate HPROF_GC_ROOT_THREAD_OBJ/HPROF_GC_ROOT_JAVA_FRAME/HPROF_GC_ROOT_JNI_LOCAL subrecords;
updated hprof parser test lib to collect data about threads (HPROF_GC_ROOT_THREAD_OBJ subrecords) and corresponding stack traces and stack references.

Testing - tier1-tier3, new test

Output of the test for VtreadInHeapDumpTarg$VthreadMounted thread
without the fix:
`thread 0x8101be90, 16 frames

[0] VtreadInHeapDumpTarg$VthreadMounted.run()V (VtreadInHeapDump.java:127)
Java Local Reference: VtreadInHeapDumpTarg$VthreadMounted
Java Local Reference: VtreadInHeapDumpTarg$VThreadMountedReferenced
[1] java.lang.Thread.runWith(Ljava/lang/Object;Ljava/lang/Runnable;)V (Thread.java:1583)
Java Local Reference: java.lang.VirtualThread
Java Local Reference: java.lang.Class
Java Local Reference: VtreadInHeapDumpTarg$VthreadMounted
[2] java.lang.VirtualThread.run(Ljava/lang/Runnable;)V (VirtualThread.java:309)
Java Local Reference: java.lang.VirtualThread
Java Local Reference: VtreadInHeapDumpTarg$VthreadMounted
Java Local Reference: java.lang.Class
[3] java.lang.VirtualThread$VThreadContinuation$1.run()V (VirtualThread.java:190)
Java Local Reference: java.lang.VirtualThread$VThreadContinuation$1
[4] jdk.internal.vm.Continuation.enter0()V (Continuation.java:320)
Java Local Reference: java.lang.VirtualThread$VThreadContinuation
[5] jdk.internal.vm.Continuation.enter(Ljdk/internal/vm/Continuation;Z)V (Continuation.java:312)
Java Local Reference: java.lang.VirtualThread$VThreadContinuation
[6] jdk.internal.vm.Continuation.enterSpecial(Ljdk/internal/vm/Continuation;ZZ)V (Continuation.java:(native method))
[7] jdk.internal.vm.Continuation.run()V (Continuation.java:248)
Java Local Reference: java.lang.VirtualThread$VThreadContinuation
Java Local Reference: jdk.internal.misc.CarrierThread
[8] java.lang.VirtualThread.runContinuation()V (VirtualThread.java:221)
Java Local Reference: java.lang.VirtualThread
[9] java.lang.VirtualThread$$Lambda+0x0000023001057a58.run()V (:(unknown))
Java Local Reference: java.lang.VirtualThread$$Lambda+0x0000023001057a58
[10] java.util.concurrent.ForkJoinTask$RunnableExecuteAction.exec()Z (ForkJoinTask.java:1423)
Java Local Reference: java.util.concurrent.ForkJoinTask$RunnableExecuteAction
[11] java.util.concurrent.ForkJoinTask.doExec()I (ForkJoinTask.java:387)
Java Local Reference: java.util.concurrent.ForkJoinTask$RunnableExecuteAction
[12] java.util.concurrent.ForkJoinPool$WorkQueue.topLevelExec(Ljava/util/concurrent/ForkJoinTask;Ljava/util/concurrent/ForkJoinPool$WorkQueue;)V (ForkJoinPool.java:1312)
Java Local Reference: java.util.concurrent.ForkJoinPool$WorkQueue
Java Local Reference: java.util.concurrent.ForkJoinTask$RunnableExecuteAction
Java Local Reference: java.util.concurrent.ForkJoinPool$WorkQueue
[13] java.util.concurrent.ForkJoinPool.scan(Ljava/util/concurrent/ForkJoinPool$WorkQueue;II)I (ForkJoinPool.java:1843)
Java Local Reference: java.util.concurrent.ForkJoinPool
Java Local Reference: java.util.concurrent.ForkJoinPool$WorkQueue
Java Local Reference: [Ljava.util.concurrent.ForkJoinPool$WorkQueue;
Java Local Reference: java.util.concurrent.ForkJoinPool$WorkQueue
Java Local Reference: [Ljava.util.concurrent.ForkJoinTask;
Java Local Reference: java.util.concurrent.ForkJoinTask$RunnableExecuteAction
[14] java.util.concurrent.ForkJoinPool.runWorker(Ljava/util/concurrent/ForkJoinPool$WorkQueue;)V (ForkJoinPool.java:1808)
Java Local Reference: java.util.concurrent.ForkJoinPool
Java Local Reference: java.util.concurrent.ForkJoinPool$WorkQueue
[15] java.util.concurrent.ForkJoinWorkerThread.run()V (ForkJoinWorkerThread.java:188)
Java Local Reference: jdk.internal.misc.CarrierThread
Java Local Reference: java.util.concurrent.ForkJoinPool
Java Local Reference: java.util.concurrent.ForkJoinPool$WorkQueue`

with the fix:
`thread 0x8101c360, 6 frames

[0] VtreadInHeapDumpTarg$VthreadMounted.run()V (VtreadInHeapDump.java:127)
Java Local Reference: VtreadInHeapDumpTarg$VthreadMounted
Java Local Reference: VtreadInHeapDumpTarg$VThreadMountedReferenced
[1] java.lang.Thread.runWith(Ljava/lang/Object;Ljava/lang/Runnable;)V (Thread.java:1583)
Java Local Reference: java.lang.VirtualThread
Java Local Reference: java.lang.Class
Java Local Reference: VtreadInHeapDumpTarg$VthreadMounted
[2] java.lang.VirtualThread.run(Ljava/lang/Runnable;)V (VirtualThread.java:309)
Java Local Reference: java.lang.VirtualThread
Java Local Reference: VtreadInHeapDumpTarg$VthreadMounted
Java Local Reference: java.lang.Class
[3] java.lang.VirtualThread$VThreadContinuation$1.run()V (VirtualThread.java:190)
Java Local Reference: java.lang.VirtualThread$VThreadContinuation$1
[4] jdk.internal.vm.Continuation.enter0()V (Continuation.java:320)
Java Local Reference: java.lang.VirtualThread$VThreadContinuation
[5] jdk.internal.vm.Continuation.enter(Ljdk/internal/vm/Continuation;Z)V (Continuation.java:312)
Java Local Reference: java.lang.VirtualThread$VThreadContinuation

`thread 0x8101be90, 9 frames

[0] jdk.internal.vm.Continuation.run()V (Continuation.java:248)
Java Local Reference: java.lang.VirtualThread$VThreadContinuation
Java Local Reference: jdk.internal.misc.CarrierThread
[1] java.lang.VirtualThread.runContinuation()V (VirtualThread.java:221)
Java Local Reference: java.lang.VirtualThread
[2] java.lang.VirtualThread$$Lambda+0x000001fa96057a58.run()V (:(unknown))
Java Local Reference: java.lang.VirtualThread$$Lambda+0x000001fa96057a58
[3] java.util.concurrent.ForkJoinTask$RunnableExecuteAction.exec()Z (ForkJoinTask.java:1423)
Java Local Reference: java.util.concurrent.ForkJoinTask$RunnableExecuteAction
[4] java.util.concurrent.ForkJoinTask.doExec()I (ForkJoinTask.java:387)
Java Local Reference: java.util.concurrent.ForkJoinTask$RunnableExecuteAction
[5] java.util.concurrent.ForkJoinPool$WorkQueue.topLevelExec(Ljava/util/concurrent/ForkJoinTask;Ljava/util/concurrent/ForkJoinPool$WorkQueue;)V (ForkJoinPool.java:1312)
Java Local Reference: java.util.concurrent.ForkJoinPool$WorkQueue
Java Local Reference: java.util.concurrent.ForkJoinTask$RunnableExecuteAction
Java Local Reference: java.util.concurrent.ForkJoinPool$WorkQueue
[6] java.util.concurrent.ForkJoinPool.scan(Ljava/util/concurrent/ForkJoinPool$WorkQueue;II)I (ForkJoinPool.java:1843)
Java Local Reference: java.util.concurrent.ForkJoinPool
Java Local Reference: java.util.concurrent.ForkJoinPool$WorkQueue
Java Local Reference: [Ljava.util.concurrent.ForkJoinPool$WorkQueue;
Java Local Reference: java.util.concurrent.ForkJoinPool$WorkQueue
Java Local Reference: [Ljava.util.concurrent.ForkJoinTask;
Java Local Reference: java.util.concurrent.ForkJoinTask$RunnableExecuteAction
[7] java.util.concurrent.ForkJoinPool.runWorker(Ljava/util/concurrent/ForkJoinPool$WorkQueue;)V (ForkJoinPool.java:1808)
Java Local Reference: java.util.concurrent.ForkJoinPool
Java Local Reference: java.util.concurrent.ForkJoinPool$WorkQueue
[8] java.util.concurrent.ForkJoinWorkerThread.run()V (ForkJoinWorkerThread.java:188)
Java Local Reference: jdk.internal.misc.CarrierThread
Java Local Reference: java.util.concurrent.ForkJoinPool
Java Local Reference: java.util.concurrent.ForkJoinPool$WorkQueue`

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-8316691: Heap dump: separate stack traces for mounted virtual threads (Sub-task - P4)

Reviewers

Leonid Mesnik (@lmesnik - Reviewer)
Serguei Spitsyn (@sspitsyn - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/15869/head:pull/15869
$ git checkout pull/15869

Update a local copy of the PR:
$ git checkout pull/15869
$ git pull https://git.openjdk.org/jdk.git pull/15869/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 15869

View PR using the GUI difftool:
$ git pr show -t 15869

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/15869.diff

Webrev

Link to Webrev Comment

bridgekeeper · 2023-09-21T20:07:56Z

👋 Welcome back amenkov! 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 · 2023-09-21T20:10:43Z

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

hotspot-runtime
serviceability

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.

mlbridge · 2023-09-21T20:14:47Z

Webrevs

test/hotspot/jtreg/serviceability/jvmti/vthread/HeapDump/VtreadInHeapDump.java

test/lib/jdk/test/lib/hprof/model/Root.java

test/lib/jdk/test/lib/hprof/model/ThreadObject.java

test/lib/jdk/test/lib/hprof/parser/HprofReader.java

dholmes-ora · 2023-09-27T07:18:27Z

Is there anything to tell you which carrier thread is associated with which mounted VThread?

AlanBateman · 2023-09-27T07:25:50Z

Is there anything to tell you which carrier thread is associated with which mounted VThread?

The carrier and virtual threads are distinct and the HPROF format doesn't have record types to support associations like this. However, your question does make me wonder how it behaves when running with +ShowCarrierFrames.

alexmenkov · 2023-09-27T22:44:36Z

Is there anything to tell you which carrier thread is associated with which mounted VThread?
No (at least for now).
I'm not sure if we need this information in heap dump (and if we need a way to differentiate platform(carrier)/virtual threads).
If we need it, we can introduce new record type ((outside of HPROF_HEAP_DUMP record) for this later (new record type should not break existing hprof readers/parsers)

alexmenkov · 2023-09-28T02:51:00Z

Is there anything to tell you which carrier thread is associated with which mounted VThread?

The carrier and virtual threads are distinct and the HPROF format doesn't have record types to support associations like this. However, your question does make me wonder how it behaves when running with +ShowCarrierFrames.

IFAIU +ShowCarrierFrames does not makes sense for "general" heap dump (like generated by "jcmd GC.heap_dump").
We can handle it for HeapDumpOnOutOfMemoryError case (when we add fake frame with OOME ctor), but I don't think it'd add a value for diagnostics.

sspitsyn · 2023-10-03T23:15:52Z

I wonder why the enterSpecial frame is not present in mounted virtual thread stack trace case with the fix:
[6] jdk.internal.vm.Continuation.enterSpecial(Ljdk/internal/vm/Continuation;ZZ)V (Continuation.java:(native method))

src/hotspot/share/services/heapDumper.cpp

alexmenkov · 2023-10-04T01:21:02Z

I wonder why the enterSpecial frame is not present in mounted virtual thread stack trace case with the fix: [6] jdk.internal.vm.Continuation.enterSpecial(Ljdk/internal/vm/Continuation;ZZ)V (Continuation.java:(native method))

I used the same logic to split stack traces as ThreadStackTrace::dump_stack_at_safepoint (it was added with VT integration, it's used, for example by VM_ThreadDump).
My understanding that enterSpecial is considered as a special frame which is not belongs to VT stack trace
@AlanBateman could you please comment if my understanding is correct.

src/hotspot/share/services/heapDumper.cpp

sspitsyn · 2023-10-05T14:40:29Z

used the same logic to split stack traces as ThreadStackTrace::dump_stack_at_safepoint
(it was added with VT integration, it's used, for example by VM_ThreadDump).
My understanding that enterSpecial is considered as a special frame which is not belongs to VT stack trace.

That's right.
I forgot it is not returned by the vframe::java_sender(), so JVMTI GetStackTrace() does not return it as well.

lmesnik

The fix looks good. (I looked at the test changes mostly.) The only question, if it makes sense to update the test to verify that heapdump works correctly without continuations.

openjdk · 2023-10-06T19:16:08Z

@alexmenkov 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:

8316691: Heap dump: separate stack traces for mounted virtual threads

Reviewed-by: lmesnik, sspitsyn

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 232 new commits pushed to the master branch:

33591a3: 8317795: Add an ImmutableBitSetPredicate variant for bitsets <= 128 elements
fb4098f: 8317730: Change byte_size to return size_t
c2abf12: 8261894: Remove support for UseSHM
ad7a8e8: 8315505: CompileTask timestamp printed can overflow
fcff222: 8317706: Exclude java/awt/Graphics2D/DrawString/RotTransText.java on linux
ef41aa0: 8317711: Exclude gtest/GTestWrapper.java on AIX
ad6dce3: 8317705: ProblemList sun/tools/jstat/jstatLineCountsX.sh on linux-ppc64le and aix due to JDK-8248691
e882718: 8314199: Initial size PBEKeyFactory#validTypes is not up-to-date
d313915: 8316233: VirtualThreadStart events should not be thread-filtered
053f456: 8316447: 8 sun/management/jmxremote tests ignore VM flags
... and 222 more: https://git.openjdk.org/jdk/compare/c04c9ea3615aad875ccb7e103e2d885b34cda7f6...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.

sspitsyn

The fix looks pretty good to me.
I've posted several nits though.

src/hotspot/share/services/heapDumper.cpp

test/hotspot/jtreg/serviceability/jvmti/vthread/HeapDump/VThreadInHeapDump.java

alexmenkov · 2023-10-07T00:46:33Z

The fix looks good. (I looked at the test changes mostly.) The only question, if it makes sense to update the test to verify that heapdump works correctly without continuations.

good point.
Added testing with -VMContinuations
Also realized that the test itself does not need extra VM options (they are needed only for target process) and the test does not require othervm mode

sspitsyn

Thank you for the update!
Looks good.
Thanks,
Serguei

alexmenkov · 2023-10-10T20:11:31Z

/integrate

openjdk · 2023-10-10T20:13:31Z

Going to push as commit 2b8276a.
Since your change was applied there have been 232 commits pushed to the master branch:

33591a3: 8317795: Add an ImmutableBitSetPredicate variant for bitsets <= 128 elements
fb4098f: 8317730: Change byte_size to return size_t
c2abf12: 8261894: Remove support for UseSHM
ad7a8e8: 8315505: CompileTask timestamp printed can overflow
fcff222: 8317706: Exclude java/awt/Graphics2D/DrawString/RotTransText.java on linux
ef41aa0: 8317711: Exclude gtest/GTestWrapper.java on AIX
ad6dce3: 8317705: ProblemList sun/tools/jstat/jstatLineCountsX.sh on linux-ppc64le and aix due to JDK-8248691
e882718: 8314199: Initial size PBEKeyFactory#validTypes is not up-to-date
d313915: 8316233: VirtualThreadStart events should not be thread-filtered
053f456: 8316447: 8 sun/management/jmxremote tests ignore VM flags
... and 222 more: https://git.openjdk.org/jdk/compare/c04c9ea3615aad875ccb7e103e2d885b34cda7f6...master

Your commit was automatically rebased without conflicts.

openjdk · 2023-10-10T20:13:50Z

@alexmenkov Pushed as commit 2b8276a.

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

alexmenkov added 2 commits September 21, 2023 12:45

separate stack traces for mounted VT

d96426c

copyright in hprof test lib

f2d3c37

openjdk bot added the rfr Pull request is ready for review label Sep 21, 2023

openjdk bot added serviceability serviceability-dev@openjdk.org hotspot-runtime hotspot-runtime-dev@openjdk.org labels Sep 21, 2023