RFR: 8263380: Unintended use of Objects.nonNull in VarHandles

[Andrey Turbanov]

On Wed, 10 Mar 2021 16:40:56 GMT, Claes Redestad <redestad at openjdk.org> wrote:

>> Nice catch!
>> 
>> We have a test that is supposed to test this: https://github.com/openjdk/jdk/blob/master/test/jdk/java/foreign/TestNulls.java But it is only checking if an NPE is thrown, and not checking for a message, since `Objects::requireNonNull` does not set an exception message. I guess that test was still passing because NPEs are thrown at some other point during the call.
>
>> Nice catch!
>> 
>> We have a test that is supposed to test this: https://github.com/openjdk/jdk/blob/master/test/jdk/java/foreign/TestNulls.java But it is only checking if an NPE is thrown, and not checking for a message, since `Objects::requireNonNull` does not set an exception message. I guess that test was still passing because NPEs are thrown at some other point during the call.
> 
> Right, checked that passes (had only run the java/lang/invoke tests locally). An alternative approach here is to verify all methods already implicitly check null and remove all these, but being explicit is nice and reduces possibility of surprise.

I found few more similar places in JFR code, where `Objects.nonNull` is used incorrectly.
jdk.jfr.internal.consumer.AbstractEventStream#setStartTime
jdk.jfr.consumer.EventStream#openRepository(java.nio.file.Path)
jdk.jfr.Recording#setFlushInterval

-------------

PR: https://git.openjdk.java.net/jdk/pull/2914