8296709: WARNING: JNI call made without checking exceptions #11083

RealLucy · 2022-11-10T09:02:02Z

This small change fixes a warning that may pop up during runtime. May I please request reviews? Thank you!

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-8296709: WARNING: JNI call made without checking exceptions

Reviewers

Kevin Walls (@kevinjwalls - Committer) ⚠️ Review applies to 18aec934
Serguei Spitsyn (@sspitsyn - Reviewer) ⚠️ Review applies to cea8cf87
Christoph Langer (@RealCLanger - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk pull/11083/head:pull/11083
$ git checkout pull/11083

Update a local copy of the PR:
$ git checkout pull/11083
$ git pull https://git.openjdk.org/jdk pull/11083/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 11083

View PR using the GUI difftool:
$ git pr show -t 11083

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/11083.diff

bridgekeeper · 2022-11-10T09:02:55Z

👋 Welcome back lucy! 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 · 2022-11-10T09:06:10Z

@RealLucy The following label will be automatically applied to this pull request:

serviceability

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

mlbridge · 2022-11-10T09:09:14Z

Webrevs

kevinjwalls

Looks good I think. Got to check for those exceptions!

The example invocation also shows the me "exceeds capacity" warning, which must relate to the above WITH_LOCAL_REFS and call to createLocalRefSpace. Another time maybe. 8-)

RealLucy · 2022-11-10T11:09:50Z

Thanks for the review, Kevin!
I'm looking into the "exceeds capacity" warning separately. Need to find out who is holding so many local references, just for a java -version. Once I have enough knowledge, I'll file a separate bug.

sspitsyn

Looks good.
Thank you for taking care about this.
Serguei

openjdk · 2022-11-10T11:15:43Z

@RealLucy 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:

8296709: WARNING: JNI call made without checking exceptions

Reviewed-by: kevinw, sspitsyn, clanger

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 92 new commits pushed to the master branch:

c3b285a: 8296916: RISC-V: Move some small macro-assembler functions to header file
720c3d0: 8296932: G1: Remove unused G1RootClosures::weak_oops
196d021: 8297020: Rename GrowableArray::on_stack
5f51dff: 8296776: Stop using mtNone as marker for CHeap allocations in GrowableArray
499406c: 8285488: Improve DocFinder
97ab2c3: 8296953: Fix a typo in exception documentation
5e08b3f: 8296931: NMT tests slowed down considerably by JDK-8242181
65f7de2: 8022403: sun/java2d/DirectX/OnScreenRenderingResizeTest/OnScreenRenderingResizeTest.java fails
3f2f128: 8295952: Problemlist existing compiler/rtm tests also on x86
6ead2b0: 8296548: Improve MD5 intrinsic for x86_64
... and 82 more: https://git.openjdk.org/jdk/compare/d4376f8b55391485365797d1c4d0dbbc6ed2ad92...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.

dholmes-ora · 2022-11-10T12:20:58Z

Can the test be augmented to catch this case too?

RealLucy · 2022-11-10T13:29:57Z

Can the test be augmented to catch this case too?

David,
the test already catches both warnings. The warnings are only issued under special circumstances. That's why everything looks fine in most cases. SAP's commercial vm has a feature called "debugging on demand" which is active by default. Therefore we see the warnings in our tests.

RealCLanger · 2022-11-11T06:26:19Z

Maybe you could also call "bin/java -Xcheck:jni -agentlib:jdwp=transport=dt_socket,server=y,suspend=n -version" which would show the error? Or add this as another test for serviceability?

dholmes-ora · 2022-11-11T07:16:57Z

I was suggesting what @RealCLanger suggests in regards to the test - add a new case.

RealLucy · 2022-11-11T08:02:40Z

I was suggesting what @RealCLanger suggests in regards to the test - add a new case.

Ahh, now I understand. I will add the suggested test modification to this PR.

RealLucy · 2022-11-11T16:12:43Z

I've added a separate test which does the same but with an active jdwp agent. The test will fail until the warning
WARNING: JNI local refs: 921, exceeds capacity: 32
is fixed. I'm not sure how to do this. Just increasing the capacity doesn't seem to be the right thing.

RealCLanger

Looks good. I commented on two nits.

If you don't have a quick idea on how to fix WARNING: JNI local refs: 921, exceeds capacity: 32, you could open a bug for it (if there isn't one already) and right away exclude the test with reference to the new bug.
The alternative would be to filter out the local refs warning but then I guess this will be forgotten.
After all, either way is not perfect...

test/jdk/tools/launcher/TestXcheckJdwpJNIWarnings.java

dholmes-ora · 2022-11-13T23:51:28Z

Couldn't you simply add a second doExec statement in the existing test rather than copy the entire test file?

RealLucy · 2022-11-14T11:07:52Z

Couldn't you simply add a second doExec statement in the existing test rather than copy the entire test file?

Actually, I first did it that way. And I did not like it. It's not just the doExec, you need to duplicate the checks as well. So far, it's just personal preference.

The test case with active jdwp agent produces two warnings, but I have a fix for only one of the warnings. @RealCLanger suggested to create a bug for the unresolved warning and problem-list the test case until the bug is fixed. If this is the way to go, I need a separate test file.

RealCLanger · 2022-11-14T11:36:21Z

Couldn't you simply add a second doExec statement in the existing test rather than copy the entire test file?

Actually, I first did it that way. And I did not like it. It's not just the doExec, you need to duplicate the checks as well. So far, it's just personal preference.

The test case with active jdwp agent produces two warnings, but I have a fix for only one of the warnings. @RealCLanger suggested to create a bug for the unresolved warning and problem-list the test case until the bug is fixed. If this is the way to go, I need a separate test file.

Hm, I also don't like the code duplication...

You could use 2 @run tags. Via the run tags you can create two runs of the test, one like the current "default" and antoher one passing the debugging agent option as additional parameter which you have to hand to the the java child process then, too. Generally you'd expect the same results, e.g. good return code, the hello world and no warnings.

With that approach you can then exclude the failing run via exclude list.

RealLucy · 2022-11-14T14:41:30Z

Hm, I also don't like the code duplication...

Code duplication is gone. Just one test file with two @run tags.

kevinjwalls · 2022-11-14T14:53:22Z

To see where the JNI local refs warning comes from, I set a breakpoint in JavaThread::print_stack_on as that is called when the warning happens, although it doesn't print anything as there is no Java frame.
...
(gdb) cont
Continuing.
WARNING: JNI local refs: 442, exceeds capacity: 32

Breakpoint 1, 0x00007ffff6efb8c0 in JavaThread::print_stack_on(outputStream*) ()
from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
(gdb) bt
#0 0x00007ffff6efb8c0 in JavaThread::print_stack_on(outputStream*) ()
from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
#1 0x00007ffff6a0f38c in checked_jni_GetJavaVM () from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
#2 0x00007ffff7e8dfa8 in transport_startTransport () from /scratch/kwalls/jdk/jdk-17.0.3/lib/libjdwp.so
#3 0x00007ffff7e76f77 in startTransport () from /scratch/kwalls/jdk/jdk-17.0.3/lib/libjdwp.so
#4 0x00007ffff7e75069 in bagEnumerateOver () from /scratch/kwalls/jdk/jdk-17.0.3/lib/libjdwp.so
#5 0x00007ffff7e77948 in initialize () from /scratch/kwalls/jdk/jdk-17.0.3/lib/libjdwp.so
#6 0x00007ffff7e783f0 in cbEarlyVMInit () from /scratch/kwalls/jdk/jdk-17.0.3/lib/libjdwp.so
#7 0x00007ffff6b5e756 in JvmtiExport::post_vm_initialized() ()
from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
#8 0x00007ffff6f00f54 in Threads::create_vm(JavaVMInitArgs*, bool*) ()
from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
#9 0x00007ffff6a007d1 in JNI_CreateJavaVM () from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
...

Specifically breaking in JNIHandles::make_local I see a lot of activity like:

Breakpoint 2, 0x00007ffff6a240c4 in JNIHandles::make_local(oopDesc*) ()
from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
#0 0x00007ffff6a240c4 in JNIHandles::make_local(oopDesc*) () from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
#1 0x00007ffff6b6d1ed in LoadedClassesClosure::do_klass(Klass*) ()
from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
#2 0x00007ffff66c6016 in ClassLoaderData::loaded_classes_do(KlassClosure*) ()
from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
#3 0x00007ffff66c9dd8 in ClassLoaderDataGraph::loaded_classes_do(KlassClosure*) ()
from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
#4 0x00007ffff6b6ce6f in JvmtiGetLoadedClasses::getLoadedClasses(JvmtiEnv*, int*, _jclass***) ()
from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
#5 0x00007ffff6afa426 in jvmti_GetLoadedClasses () from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
#6 0x00007ffff7e7561b in classTrack_initialize () from /scratch/kwalls/jdk/jdk-17.0.3/lib/libjdwp.so
#7 0x00007ffff7e778f1 in initialize () from /scratch/kwalls/jdk/jdk-17.0.3/lib/libjdwp.so
#8 0x00007ffff7e783f0 in cbEarlyVMInit () from /scratch/kwalls/jdk/jdk-17.0.3/lib/libjdwp.so
#9 0x00007ffff6b5e756 in JvmtiExport::post_vm_initialized() ()
from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
#10 0x00007ffff6f00f54 in Threads::create_vm(JavaVMInitArgs*, bool*) ()
from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
#11 0x00007ffff6a007d1 in JNI_CreateJavaVM () from /scratch/kwalls/jdk/jdk-17.0.3/lib/server/libjvm.so
...

LoadedClassesClosure::do_klass does indeed call make_local many times.
If this is the cause of the warning for >400 references, not sure how the code should know how many references to expect (the number of loaded classes..).

If I do make classTrack_initialize state it wants a large number of local references, that warning disappears:

$ git diff src/jdk.jdwp.agent/share/native/libjdwp/classTrack.c
diff --git a/src/jdk.jdwp.agent/share/native/libjdwp/classTrack.c b/src/jdk.jdwp.agent/share/native/libjdwp/classTrack.c
index e3b30282daa..7dfb9ec01c4 100644
--- a/src/jdk.jdwp.agent/share/native/libjdwp/classTrack.c
+++ b/src/jdk.jdwp.agent/share/native/libjdwp/classTrack.c
@@ -171,6 +171,7 @@ setupEvents()
 void
 classTrack_initialize(JNIEnv *env)
 {
+    WITH_LOCAL_REFS(env, 500) {
     trackingEnv = getSpecialJvmti();
     if (trackingEnv == NULL) {
         EXIT_ERROR(AGENT_ERROR_INTERNAL, "Failed to allocate tag-tracking jvmtiEnv");
@@ -201,4 +202,5 @@ classTrack_initialize(JNIEnv *env)
     } else {
         EXIT_ERROR(error,"loaded classes array");
     }
+    } END_WITH_LOCAL_REFS(env)
 }

plummercj · 2022-11-14T18:17:47Z

This is the problematic code:

/* Get all loaded classes (must be inside a WITH_LOCAL_REFS) */
jvmtiError
allLoadedClasses(jclass **ppclasses, jint *pcount)
{
    jvmtiError error;

    *pcount = 0;
    *ppclasses = NULL;
    error = JVMTI_FUNC_PTR(gdata->jvmti,GetLoadedClasses)
                (gdata->jvmti, pcount, ppclasses);
    return error;
}

It seems there are a few places where allLoadedClasses() is called with either no WITH_LOCAL_REFS, or with an inadequate one. However, given that the number of jni localrefs to be allocated is unknown a potentially very large, I'm not sure how callers of JVMTI GetLoadedClasses() are suppose to defend against an excess be allocated.

RealCLanger

I think we're getting there. A few further nits and now you also need to update the copyright year.

test/jdk/tools/launcher/TestXcheckJNIWarnings.java

RealCLanger · 2022-11-15T13:39:09Z

test/jdk/tools/launcher/TestXcheckJNIWarnings.java

+ * @summary Launching app with or without jdwp agent shouldn't produce any jni warnings.
 * @modules jdk.compiler
 *          jdk.zipfs
 * @compile TestXcheckJNIWarnings.java


I think the @compile directive is not necessary

test/jdk/tools/launcher/TestXcheckJNIWarnings.java

RealCLanger · 2022-11-15T13:39:46Z

test/jdk/tools/launcher/TestXcheckJNIWarnings.java

+ * @summary Launching app with or without jdwp agent shouldn't produce any jni warnings.
+ * @modules jdk.compiler
+ *          jdk.zipfs
+ * @compile TestXcheckJNIWarnings.java


@compile can be removed.

RealCLanger

Looks great now 😄

RealLucy · 2022-11-16T14:14:54Z

Thank you all for your valuable feedback and for the reviews. Once the "still in progress" tests have completed successfully, I'll integrate. The failing linux-x86 tier1 test does not look related. It is a c2 compiler test.

kevinjwalls · 2022-11-16T14:28:39Z

Sounds good. Comment above showing jvmti_GetLoadedClasses as the cause of the references was just to make sure we knew the cause of that other warning, but I see there is JDK-8296936 to follow up that issue.

RealLucy · 2022-11-16T14:58:11Z

/integrate

openjdk · 2022-11-16T15:00:30Z

Going to push as commit eac26f4.
Since your change was applied there have been 97 commits pushed to the master branch:

4946737: 8297047: IGV: graphContent not set when opening a new tab
8b1ff9e: 8297132: BACKOUT JDK-8296889: Race condition when cancelling a request
e72b0ac: 8296429: Remove os::supports_sse
813b223: 8296926: Sort include lines of files in the include/ directory
8cdcec4: 8296774: Removed default MEMFLAGS value from CHeapBitMap
c3b285a: 8296916: RISC-V: Move some small macro-assembler functions to header file
720c3d0: 8296932: G1: Remove unused G1RootClosures::weak_oops
196d021: 8297020: Rename GrowableArray::on_stack
5f51dff: 8296776: Stop using mtNone as marker for CHeap allocations in GrowableArray
499406c: 8285488: Improve DocFinder
... and 87 more: https://git.openjdk.org/jdk/compare/d4376f8b55391485365797d1c4d0dbbc6ed2ad92...master

Your commit was automatically rebased without conflicts.

openjdk · 2022-11-16T15:00:59Z

@RealLucy Pushed as commit eac26f4.

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

8296709: WARNING: JNI call made without checking exceptions

18aec93

openjdk bot added the rfr Pull request is ready for review label Nov 10, 2022

openjdk bot added the serviceability serviceability-dev@openjdk.org label Nov 10, 2022

kevinjwalls approved these changes Nov 10, 2022

View reviewed changes

sspitsyn approved these changes Nov 10, 2022

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Nov 10, 2022

8296709: add testcase with active jdwp agent

a71384b

8296709: fix testcase naming

a3f65ed

RealCLanger suggested changes Nov 11, 2022

View reviewed changes

test/jdk/tools/launcher/TestXcheckJdwpJNIWarnings.java Outdated Show resolved Hide resolved

test/jdk/tools/launcher/TestXcheckJdwpJNIWarnings.java Outdated Show resolved Hide resolved

RealLucy added 2 commits November 14, 2022 12:16

8296709: Copyright year and more specific summary text

8f4caed

8296709: typo

3e2e44e

8296709: use multiple @run tags instead of multiple files

409f695

8296709: use multiple test ids

cea8cf8

RealCLanger reviewed Nov 15, 2022

View reviewed changes

sspitsyn approved these changes Nov 16, 2022

View reviewed changes

RealLucy added 2 commits November 16, 2022 12:23

8296709: Fix test run spec, exclude jdwp agent case

b9fc068

8296709: Copyright, descriptive text, @compile directive removed

8c73f96

RealCLanger approved these changes Nov 16, 2022

View reviewed changes

openjdk bot added the integrated Pull request has been integrated label Nov 16, 2022

openjdk bot closed this Nov 16, 2022

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Nov 16, 2022

8296709: WARNING: JNI call made without checking exceptions #11083

8296709: WARNING: JNI call made without checking exceptions #11083

Uh oh!

Conversation

RealLucy commented Nov 10, 2022 • 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 Nov 10, 2022

Uh oh!

openjdk bot commented Nov 10, 2022

Uh oh!

mlbridge bot commented Nov 10, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

kevinjwalls left a comment

Choose a reason for hiding this comment

Uh oh!

RealLucy commented Nov 10, 2022

Uh oh!

sspitsyn left a comment

Choose a reason for hiding this comment

Uh oh!

openjdk bot commented Nov 10, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dholmes-ora commented Nov 10, 2022

Uh oh!

RealLucy commented Nov 10, 2022

Uh oh!

RealCLanger commented Nov 11, 2022

Uh oh!

dholmes-ora commented Nov 11, 2022

Uh oh!

RealLucy commented Nov 11, 2022

Uh oh!

RealLucy commented Nov 11, 2022

Uh oh!

RealCLanger left a comment • 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!

Uh oh!

Uh oh!

dholmes-ora commented Nov 13, 2022

Uh oh!

RealLucy commented Nov 14, 2022

Uh oh!

RealCLanger commented Nov 14, 2022

Uh oh!

RealLucy commented Nov 14, 2022

Uh oh!

kevinjwalls commented Nov 14, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

plummercj commented Nov 14, 2022

Uh oh!

RealCLanger left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

RealCLanger Nov 15, 2022

Choose a reason for hiding this comment

Uh oh!

Uh oh!

RealCLanger Nov 15, 2022

Choose a reason for hiding this comment

Uh oh!

RealCLanger left a comment

Choose a reason for hiding this comment

Uh oh!

RealLucy commented Nov 16, 2022

Uh oh!

kevinjwalls commented Nov 16, 2022

Uh oh!

RealLucy commented Nov 16, 2022

RealLucy commented Nov 10, 2022 •

edited by openjdk bot

Loading

mlbridge bot commented Nov 10, 2022 •

edited

Loading

openjdk bot commented Nov 10, 2022 •

edited

Loading

RealCLanger left a comment •

edited

Loading

kevinjwalls commented Nov 14, 2022 •

edited

Loading