Refactor J9JVMTI_HEAP_EVENT_STACK data parsing

[fengxue-IS]

Added new structure to hold JNI and Stack local referrer data. Update JNI locals found via jniLocalReferences table to be marked as Stack root.

Closes: #18416

Signed-off-by: Jack Lu Jack.S.Lu@ibm.com

[fengxue-IS]

#19486 reverted some of the changes in #18180 which properly handle reference chain walking on virtual thread stack.
This change corrects the behaviour by caching the virtual thread object in ReferenceChainWalker and fill the value into J9MM_StackSlotDescriptor->vmThread before executing the callback function.

[fengxue-IS]

@babsingh @LinHu2016 can you please take a look at this
local testing passed, personal build passed
Note VThreadStackRefTest.java is still excluded atm

[babsingh]

Consistent formatting

There are places where there is space before and after *. Others only have space before *; there is no space after *. The latter is the preferred approach. If we're making small changes, then we should follow the existing formatting used in the file. If we're changing majority of the function, then should update the function to follow the preferred approach.

processStackRoot(J9JVMTIObjectIteratorData * data, J9JVMTIObjectTag * entry, jlong size, jlong classTag, jvmtiHeapRootKind kind, J9MM_StackSlotDescriptor *stackSlotDescriptor)

Similarly, if we're adding a cast, then the preferred approach is to not have space after the closing bracket ). Use the above logic to either employ existing formatting or transition to the preferred approach.

slot = (jint) walkState->slotIndex;

[babsingh]

How have these changes been tested? Were any personal builds run? If so, which test suites were selected?

[fengxue-IS]

How have these changes been tested? Were any personal builds run? If so, which test suites were selected?

I've include link to personal build, see above #18864 (comment)

[babsingh]

local testing passed, personal build passed

Are any of the failures and unstable builds related to this PR?

Note VThreadStackRefTest.java is still excluded atm

Does VThreadStackRefTest.java pass on all platforms?

[fengxue-IS]

local testing passed, personal build passed

Are any of the failures and unstable builds related to this PR?

No

Note VThreadStackRefTest.java is still excluded atm

Does VThreadStackRefTest.java pass on all platforms?

I've only tested on xlinux and amac for JDK 21, but since the test are all testing JVM internals, there shouldn't be any platform related differences.
I've opened adoptium/aqa-tests#5567 to re-enable the test

[LinHu2016]

LGTM, need to confirm ReferenceChainWalker is not caller during concurrent scavenger(currently it is only used during STW GC, read threadObject via Java API during concurrent scavenger might cause an infinity loop).

[fengxue-IS]

@LinHu2016 added assertions to ensure API only used under exclusive access

[babsingh]

jenkins test sanity,extended.openjdk zlinux jdk21

[babsingh]

jenkins test sanity alinux jdk11

[babsingh]

There are compilation errors on JDK11. vthread code is missing ifdefs.

11:38:59  /home/jenkins/workspace/Build_JDK11_aarch64_linux_Personal/openj9/runtime/gc_base/ReferenceChainWalker.cpp: In function 'void j9gc_ext_reachable_objects_do(J9VMThread*, jvmtiIterationControl (*)(J9Object**, J9Object*, void*, IDATA, IDATA, IDATA), void*, uintptr_t)':
11:38:59  /home/jenkins/workspace/Build_JDK11_aarch64_linux_Personal/openj9/runtime/gc_base/ReferenceChainWalker.cpp:95:24: error: 'class MM_ReferenceChainWalker' has no member named 'includeVThreadObject'
11:38:59     95 |   referenceChainWalker.includeVThreadObject();
11:38:59        |                        ^~~~~~~~~~~~~~~~~~~~
11:38:59  /home/jenkins/workspace/Build_JDK11_aarch64_linux_Personal/openj9/runtime/gc_base/ReferenceChainWalker.cpp: In function 'void j9gc_ext_reachable_from_object_do(J9VMThread*, J9Object*, jvmtiIterationControl (*)(J9Object**, J9Object*, void*, IDATA, IDATA, IDATA), void*, uintptr_t)':
11:38:59  /home/jenkins/workspace/Build_JDK11_aarch64_linux_Personal/openj9/runtime/gc_base/ReferenceChainWalker.cpp:129:24: error: 'class MM_ReferenceChainWalker' has no member named 'includeVThreadObject'
11:38:59    129 |   referenceChainWalker.includeVThreadObject();
11:38:59        |                        ^~~~~~~~~~~~~~~~~~~~
11:38:59  runtime/gc_base/CMakeFiles/j9gcbase.dir/build.make:335: recipe for target 'runtime/gc_base/CMakeFiles/j9gcbase.dir/ReferenceChainWalker.cpp.o' 

[fengxue-IS]

Added ifdef to fix JDK11 compile error, the JDK 21 PR build also failed due to jenkins flow error, @babsingh can you restart the builds please

[babsingh]

jenkins test sanity,extended.openjdk zlinux jdk21

[babsingh]

jenkins test sanity win,alinux jdk11

[babsingh]

jenkins test sanity win jdk21

[babsingh]

LGTM. Test_openjdk21_j9_extended.openjdk PR build has unrelated security/SSL failures.