On Mon, 2 May 2022 23:20:52 GMT, Alex Menkov <amen...@openjdk.org> wrote:
> The test counts all "system" threads before the execution and expects that
> this number remains the same during test execution.
> This makes the test fragile - JVM may start internal threads, some threads
> may end.
>
> The fix updates the test:
> - the test checks only test threads, and verify that the live threads are
> reported by GetAllThreads and terminated threads are not reported;
> - dropped "system" thread counting stuff;
> - added proper deallocation of GetThreadInfo results.
test/hotspot/jtreg/vmTestbase/nsk/jvmti/GetAllThreads/allthr001/allthr001.cpp
line 170:
> 168: }
> 169:
> 170: void release(JNIEnv *env, jvmtiThreadInfo *info) {
How about "releaseThreadInfo"?
test/hotspot/jtreg/vmTestbase/nsk/jvmti/GetAllThreads/allthr001/allthr001.cpp
line 238:
> 236: }
> 237: if (printdump == JNI_TRUE) {
> 238: printf(" >>> %s", inf.name);
Don't you want a newline here?
test/hotspot/jtreg/vmTestbase/nsk/jvmti/GetAllThreads/allthr001/allthr001.cpp
line 240:
> 238: printf(" >>> %s", inf.name);
> 239: }
> 240: for (j = 0, found = 0; j < thrInfo[ind].unexpected.cnt &&
> !found; j++) {
found = JNI_FALSE
-------------
PR: https://git.openjdk.java.net/jdk/pull/8512
