JDK16 Changes

[babsingh]

1. Add String.getBytes for JDK16.

2. Add Lookup.ORIGINAL and update defineHiddenClassWithClassData for JDK16.

3. Add implementation of JavaLangAccess.addExports for JDK16.

Closes: #11312

Signed-off-by: Babneet Singh sbabneet@ca.ibm.com

[pshipton]

jenkins test sanity zlinux jdknext depends ibmruntimes/openj9-openjdk-jdk#openj9-staging

[babsingh]

@pshipton Made a minor correction. The PR builds would need to be relaunched.

[tajila]

jenkins test sanity zlinux jdknext depends ibmruntimes/openj9-openjdk-jdk#openj9-staging

[tajila]

what change was made?

[babsingh]

what change was made?

fixed typo/grammar in String.getBytes description.

[pshipton]

FYI new builds aren't necessarily required for changes to comments. The original builds continue on even if you push more changes and they are no longer shown in the PR.

test_insert12 and test_insert3 are failing in JCL_Test, which means we won't automatically promote jdknext even if these changes are merged.

[pshipton]

@tajila fyi if you look at the forced-push link in babsingh force-pushed the babsingh:jdk16_comp branch from 277aa79 to a7b6eaf 2 hours ago above, you can see the changes
https://github.com/eclipse/openj9/compare/277aa796c83429afca607c280c1d68aa89051394..a7b6eaf0b7d83ff02fb1644519940e42d15c438e

[keithc-ca]

A recent addition to JavaLangAccess (void addExports(Module module, String packageName)) must be implemented.

[babsingh]

Update:

1. I will temporarily disable test_insert12 and test_insert3 to enable jdknext builds. Investigating and fixing these failures will take some time.
2. I will also add JavaLangAccess.addExports to this PR.

[keithc-ca]

Perhaps the failure in test_insert12 was fixed by openjdk?

• 8257511: JDK-8254082 brings regression to AbstractStringBuilder.insert(int dstOffset, CharSequence s, int start, int end)

See https://github.com/ibmruntimes/openj9-openjdk-jdk/blame/11ff3afa7597ba8742399ddf387ab5e17a9929f8/src/java.base/share/classes/java/lang/AbstractStringBuilder.java#L1720

[babsingh]

Perhaps the failure in test_insert12 was fixed by openjdk?

That looks like the fix. Rebuilding with the latest source to confirm.

[babsingh]

The PR has been updated to include JavaLangAccess.addExports.

The OpenJDK commit mentioned in #11322 (comment) fixes test_insert12 and test_insert3.

[pshipton]

jenkins test sanity zlinux jdknext depends ibmruntimes/openj9-openjdk-jdk#openj9-staging

[pshipton]

https://ci.eclipse.org/openj9/job/PullRequest-OpenJ9/4984/

[pshipton]

PR testing passed. The only change was to a comment so it's still valid.

[pshipton]

@babsingh Seems we should have tested jdk15, as it no longer compiles with this change. The old definition of defineHiddenClassWithClassData() didn't get removed and now there are two.

22:44:42  /home/jenkins/workspace/Build_JDK15_ppc64_aix_Nightly/build/aix-ppc64-server-release/support/j9jcl/java.base/share/classes/java/lang/invoke/MethodHandles.java:2075: error: method defineHiddenClassWithClassData(byte[],Object,ClassOption...) is already defined in class Lookup
22:44:42  		Lookup defineHiddenClassWithClassData(byte[] bytes, Object classData, ClassOption... classOptions) throws IllegalAccessException {