Clear breakpts in class redef and fix HCR warning

[tajila]

The current implementation emits a warning message when a JVMTI operation suceeds.

Also, clear breakpoints unconditionally when a class is redefined.

[tajila]

jenkins test sanity.openjdk zlinux jdk22

[tajila]

jenkins test sanity,extended,sanity.openjdk zlinux jdk17

[tajila]

Failures:

FAILED test targets:
17:12:35  	jdk_lang_VarHandleTest_j9_0 - Test results: passed: 32; failed: 6 
17:12:35  		Failed test cases: 
17:12:35  			TEST: java/lang/invoke/VarHandles/VarHandleTestByteArrayAsDouble.java
17:12:35          TEST: java/lang/invoke/VarHandles/VarHandleTestByteArrayAsChar.java
17:12:35          TEST: java/lang/invoke/VarHandles/VarHandleTestByteArrayAsFloat.java
17:12:35          TEST: java/lang/invoke/VarHandles/VarHandleTestByteArrayAsLong.java
17:12:35          TEST: java/lang/invoke/VarHandles/VarHandleTestByteArrayAsInt.java
17:12:35          TEST: java/lang/invoke/VarHandles/VarHandleTestByteArrayAsShort.java

[tajila]

Another sanity.openjdk run with -XX:-EnableExtendedHCR https://openj9-jenkins.osuosl.org/job/Grinder/3668/

[pshipton]

The VarHandles failures are this known issue #19693

[tajila]

@gacholio please review these changes

[pshipton]

Pls fix the commit title to reflect the change to clear the breakpoints. The current title is misleading.

[tajila]

test links:
https://openj9-jenkins.osuosl.org/job/Test_openjdk17_j9_extended.functional_s390x_linux_Personal/44/
https://openj9-jenkins.osuosl.org/job/Test_openjdk17_j9_sanity.functional_s390x_linux_Personal/482/
https://openj9-jenkins.osuosl.org/job/Test_openjdk22_j9_sanity.openjdk_s390x_linux_Personal/2/

[pshipton]

I expected a title that would mention both changes.

[pshipton]

With this fix, can the -XX:+EnableExtendedHCR option be removed from any of the tests where it was added by #19554 ?

[gacholio]

If the playlists are shared between open and IBM builds, then we must retain the option as the default differs between the two builds.

[pshipton]

If the playlists are shared between open and IBM builds, then we must retain the option as the default differs between the two builds.

Not sure about this logic. The IBM builds don't need any test change because -XX:+EnableExtendedHCR is enabled by default. The OpenJ9 builds may need to have this option added to certain tests, however the tests that need it were determined before the bug fix in hshelp.c, perhaps the option can be removed from some of the tests with this change.

  if (extensionsEnabled == 0) {
		if (JVMTI_ERROR_NONE != rc) {      <----- bug fix
			emitExtendedHCRWarning(currentThread);
		}
		return rc;
	}

[tajila]

With this fix, can the -XX:+EnableExtendedHCR option be removed from any of the tests where it was added by #19554 ?

Yes

[gacholio]

jenkins test sanity zlinux jdk21

[pshipton]

It seems another update to the tests is expected, setting as draft until it's ready.

[gacholio]

I don't understand what you're waiting for - by definition, only our internal tests will ever need this option, and every test which uses extended must have the option because the default is off.

[tajila]

jenkins test sanity,extended zlinux jdk21

[tajila]

jenkins test sanity,extended zlinux jdk21

[tajila]

@pshipton this is ready to go

[pshipton]

Pls create the PR for 0.46.