JDK19 serviceability_jvmti_j9_1_FAILED serviceability/jvmti/vthread/SelfSuspendDisablerTest/SelfSuspendDisablerTest.java Test FAILED: Unexpected thread state

[JasonFengJ9]

Failure link

From an internal build(rhel7s390x-4-6):

openjdk version "19.0.1" 2022-10-18
IBM Semeru Runtime Open Edition 19.0.1+10 (build 19.0.1+10)
Eclipse OpenJ9 VM 19.0.1+10 (build master-f6c7bee4b, JRE 19 Linux s390x-64-Bit Compressed References 20221018_71 (JIT enabled, AOT enabled)
OpenJ9   - f6c7bee4b
OMR      - 357df2f80
JCL      - 110e1e7808 based on jdk-19.0.1+10)

Rerun in Grinder - Change TARGET to run only the failed test targets.

Optional info

Failure output (captured from console output)

[2022-11-05T17:27:26.328Z] variation: Mode650
[2022-11-05T17:27:26.328Z] JVM_OPTIONS:  -XX:-UseCompressedOops 

[2022-11-05T17:46:43.917Z] TEST: serviceability/jvmti/vthread/SelfSuspendDisablerTest/SelfSuspendDisablerTest.java

[2022-11-05T17:46:43.918Z] STDOUT:
[2022-11-05T17:46:43.918Z] Agent init started
[2022-11-05T17:46:43.918Z] Agent init finished
[2022-11-05T17:46:43.918Z] Expected platform thread state: 000000 got: 000000
[2022-11-05T17:46:43.918Z] Expected virtual  thread state: 000000 got: 000000
[2022-11-05T17:46:43.918Z] Expected platform thread state: 000005 got: 000005
[2022-11-05T17:46:43.918Z] Expected virtual  thread state: 000005 got: 000005
[2022-11-05T17:46:43.918Z] Expected platform thread state: 100005 got: 500005
[2022-11-05T17:46:43.918Z] Timeout refired 960 times
[2022-11-05T17:46:43.918Z] STDERR:
[2022-11-05T17:46:43.918Z] Exception in thread "" java.lang.RuntimeException: Test FAILED: Unexpected thread state
[2022-11-05T17:46:43.918Z] 	at SelfSuspendDisablerTest.testJvmtiThreadState(SelfSuspendDisablerTest.java:67)
[2022-11-05T17:46:43.918Z] 	at SelfSuspendDisablerTest.lambda$main$1(SelfSuspendDisablerTest.java:83)
[2022-11-05T17:46:43.918Z] 	at java.base/java.lang.VirtualThread.run(VirtualThread.java:292)
[2022-11-05T17:46:43.918Z] 	at java.base/java.lang.VirtualThread$VThreadContinuation.lambda$new$0(VirtualThread.java:179)
[2022-11-05T17:46:43.918Z] 	at java.base/jdk.internal.vm.Continuation.execute(Continuation.java:171)

[2022-11-05T17:46:43.918Z] Test results: passed: 112; error: 1
[2022-11-05T17:46:43.918Z] Report written to /home/jenkins/workspace/Test_openjdk19_j9_extended.openjdk_s390x_linux/jvmtest/openjdk/report/html/report.html
[2022-11-05T17:46:43.918Z] Results written to /home/jenkins/workspace/Test_openjdk19_j9_extended.openjdk_s390x_linux/aqa-tests/TKG/output_16676679384301/serviceability_jvmti_j9_1/work
[2022-11-05T17:46:43.918Z] Error: Some tests failed or other problems occurred.
[2022-11-05T17:46:43.918Z] 
[2022-11-05T17:46:43.918Z] serviceability_jvmti_j9_1_FAILED

50x internal grinder - all failed

[babsingh]

The test fails for a platform thread. This is not related to Project Loom, and does not need to be addressed for the Java 19 release.

The thread is correctly suspended. But, the JVMTI_THREAD_STATE_IN_NATIVE flag is set in the thread-state, which causes the test to fail. The failure only occurs on Linux s390. As per the rules in JVMTI GetThreadState's spec, JVMTI_THREAD_STATE_IN_NATIVE is allowed to be set along with JVMTI_THREAD_STATE_SUSPENDED.

So, the test needs to perform a check similar to OMR_ARE_ALL_BITS_SET instead of an equality check (state != expectedState) to resolve this failure.

[2022-11-05T17:46:43.918Z] Expected platform thread state: 100005 got: 500005

Expected:
- 0x000001: JVMTI_THREAD_STATE_ALIVE
- 0x000004: JVMTI_THREAD_STATE_RUNNABLE
- 0x100000: JVMTI_THREAD_STATE_SUSPENDED

Got:
- 0x000001: JVMTI_THREAD_STATE_ALIVE
- 0x000004: JVMTI_THREAD_STATE_RUNNABLE
- 0x100000: JVMTI_THREAD_STATE_SUSPENDED
- 0x400000: JVMTI_THREAD_STATE_IN_NATIVE

Macro Definitions

openj9/runtime/include/jvmti.h.m4

Lines 72 to 87 in f21e296

	#define JVMTI_THREAD_STATE_ALIVE 0x00000001
	#define JVMTI_THREAD_STATE_TERMINATED 0x00000002
	#define JVMTI_THREAD_STATE_RUNNABLE 0x00000004
	#define JVMTI_THREAD_STATE_WAITING_INDEFINITELY 0x00000010
	#define JVMTI_THREAD_STATE_WAITING_WITH_TIMEOUT 0x00000020
	#define JVMTI_THREAD_STATE_SLEEPING 0x00000040
	#define JVMTI_THREAD_STATE_WAITING 0x00000080
	#define JVMTI_THREAD_STATE_IN_OBJECT_WAIT 0x00000100
	#define JVMTI_THREAD_STATE_PARKED 0x00000200
	#define JVMTI_THREAD_STATE_BLOCKED_ON_MONITOR_ENTER 0x00000400
	#define JVMTI_THREAD_STATE_SUSPENDED 0x00100000
	#define JVMTI_THREAD_STATE_INTERRUPTED 0x00200000
	#define JVMTI_THREAD_STATE_IN_NATIVE 0x00400000
	#define JVMTI_THREAD_STATE_VENDOR_1 0x10000000
	#define JVMTI_THREAD_STATE_VENDOR_2 0x20000000
	#define JVMTI_THREAD_STATE_VENDOR_3 0x40000000

[babsingh]

Fix

• Fix SelfSuspendDisablerTest ibmruntimes/openj9-openjdk-jdk19#50; and
• Fix SelfSuspendDisablerTest ibmruntimes/openj9-openjdk-jdk#524.

Running a personal build: https://hyc-runtimes-jenkins.swg-devops.com/view/OpenJ9%20-%20Personal/job/Pipeline-Build-Test-Personal/14771/

[keithc-ca]

I don't agree with the solution to this. The quoted document says:

More than one of
JVMTI_THREAD_STATE_SUSPENDED
JVMTI_THREAD_STATE_INTERRUPTED
JVMTI_THREAD_STATE_IN_NATIVE
can be set, but if any is set, JVMTI_THREAD_STATE_ALIVE is set.

This says to me that JVMTI_THREAD_STATE_IN_NATIVE is not permitted in combination with JVMTI_THREAD_STATE_SUSPENDED.

[tajila]

This says to me that JVMTI_THREAD_STATE_IN_NATIVE is not permitted in combination with JVMTI_THREAD_STATE_SUSPENDED.

Can you please elaborate? "More than one of..." implies that both JVMTI_THREAD_STATE_IN_NATIVE and JVMTI_THREAD_STATE_SUSPENDED can be set simultaneously.

[keithc-ca]

Perhaps I read too quickly (or misread that document). Either I assumed that ALIVE was incompatible with SUSPENDED or I misread 'more than one of' as 'not more than one of' (like earlier sections). I no longer see a problem with the combination of NATIVE and SUSPENDED.

On the other hand, the test was excessively relaxed, by ignoring all bits other than the expected ones: now situations where more bits are set than legally should be will not (but should) fail.

[tajila]

@babsingh can you address Keiths comment

[babsingh]

Yes, planning to create a fix by EOD.

[babsingh]

FIX for the issue described in #16276 (comment):

• Check for illegal bits in SelfSuspendDisablerTest ibmruntimes/openj9-openjdk-jdk19#52; and
• Check for illegal bits in SelfSuspendDisablerTest ibmruntimes/openj9-openjdk-jdk#529.

fyi @tajila @keithc-ca