On Tue, 27 Apr 2021 23:15:35 GMT, Chris Plummer <cjplum...@openjdk.org> wrote:
> I don't see any mention in the CR of the test failing due to this issue. How
> was it discovered? How is the fix being verified?
I discovered it researching JCK failure on a platform with UsageTracker enabled.
JCK test does exactly the same as this test, but I was not able to reproduce
the failure for this test (but this is a race, so that's normal)
I ensured the fix works during development when created otherTread before
getReady and had printdump on.
It produced a lot of ClassPrepare event for Thread and lambda-related classes.
> test/hotspot/jtreg/vmTestbase/nsk/jvmti/ClassPrepare/classprep001/classprep001.cpp
> line 292:
>
>> 290:
>> 291: JNIEXPORT void JNICALL
>> 292: Java_nsk_jvmti_ClassPrepare_classprep001_getReady(JNIEnv *env, jclass
>> cls, jthread thread) {
>
> It's not clear to me why you now pass in the thread instead of calling
> GetCurrentThread, when the thread passed in is still always the current
> thread. Same thing for the `check()` method.
I think it makes the code more flexible and easier to extend
-------------
PR: https://git.openjdk.java.net/jdk/pull/3732
