RFR: JDK-8305416: runtime/Thread/TestAlwaysPreTouchStacks.java failed with "Did not find expected NMT output"
Thomas Stuefe
stuefe at openjdk.org
Tue Apr 4 06:17:05 UTC 2023
On Tue, 4 Apr 2023 02:52:00 GMT, David Holmes <dholmes at openjdk.org> wrote:
> I don't understand the issue of daemon versus non-daemon as that only affects when the VM can exit, not how long individual threads "stick around" ???
You are right, that's badly phrased. The old version of this test fired up threads that printed something, then finished. The JVM waited for them to exit, then exited itself, printing the NMT report in the process. That report did not contain the thread stacks for the already exited test threads.
The new test makes sure the threads never exit. They have to be daemonized in order to not hold up the JVM exit. Now, the NMT reported stack sized contain the test thread stack sizes.
-------------
PR Comment: https://git.openjdk.org/jdk/pull/13295#issuecomment-1495407143
More information about the hotspot-runtime-dev
mailing list