RFR: 8315795: runtime/Safepoint/TestAbortVMOnSafepointTimeout.java fails after JDK-8305507

[Daniel D. Daugherty]

On Thu, 7 Sep 2023 16:02:55 GMT, Patricio Chilano Mateo <pchilanomate at openjdk.org> wrote:

> Please review this small fix to test TestAbortVMOnSafepointTimeout.java. With the current value of AbortVMOnSafepointTimeoutDelay we get cases where the spawned child VM exits with a safepoint timeout but without the main thread being able to print the PRE_STALL_TEXT text. I added a more detailed analysis in the bug comments.
> To fix it I increased the value of AbortVMOnSafepointTimeoutDelay to be about twice as much as the longest time it could take to print the PRE_STALL_TEXT based on the measurements I have done in the failing Windows machines.
> Without the fix I can reproduce the issue in about 10% of the runs running tier1_runtime. With the fix I run ~400 runs of tier1_runtime without failures.
> 
> Thanks,
> Patricio

Thumbs up. Minor comments below:

I'm looking forward to seeing if this fix quiets things down.
Also, this is a trivial fix so you don't need to wait 24 hours.

test/hotspot/jtreg/runtime/Safepoint/TestAbortVMOnSafepointTimeout.java line 71:

> 69:                 "-XX:+SafepointALot",
> 70:                 "-XX:+AbortVMOnSafepointTimeout",
> 71:                 "-XX:AbortVMOnSafepointTimeoutDelay=10000",

You should update:

L33: * @bug 8219584 8227528

to include this bug ID: 8315795.

Also, you should add a comment explaining why the delay is now 10 seconds
instead of 2.5 seconds. Something like:

                "-XX:AbortVMOnSafepointTimeoutDelay=10000",  // Using 10 seconds instead of 2.5 seconds for windows-debug

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

Marked as reviewed by dcubed (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/15622#pullrequestreview-1616045565
PR Comment: https://git.openjdk.org/jdk/pull/15622#issuecomment-1710620479
PR Review Comment: https://git.openjdk.org/jdk/pull/15622#discussion_r1318993148