RFR: 8325872: Make GuaranteedSafepointInterval default 0 [v2]

[Daniel D. Daugherty]

On Wed, 17 Apr 2024 17:30:54 GMT, Coleen Phillimore <coleenp at openjdk.org> wrote:

>> This change keeps the GuaranteedSafepointInterval option for use with -XX:+HandshakeALot and -XX:+SafepointALot and removes it from working with the MonitorDeflationThread.  The MonitorDeflation thread has it's own timers and doesn't require safepoint to trigger anymore, so the GuaranteedSafepointInterval is logically unrelated except for historical reasons.
>> The option AsyncDeflationInterval default is 250 so was used anyway in this code for the monitor deflation thread timer.
>> 
>> Tested with tier1-8.  Also performance tested this which didn't show any improvements.
>
> Coleen Phillimore has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Fix argument conditionals, reran tests with SafepointALot and HandshakeALot.

Thumbs up.

I only had a question about some code that you might want to change.

src/hotspot/share/runtime/vmThread.cpp line 317:

> 315:     return false;
> 316:   }
> 317:   static jlong last_halot_ms = 0;

Below this point is this code:

  // If only HandshakeALot is set, but GuaranteedSafepointInterval is 0,
  // we emit a handshake if it's been more than a second since the last one.
  jlong interval = GuaranteedSafepointInterval != 0 ? GuaranteedSafepointInterval : 1000;


Do you need this now that you've made the change in src/hotspot/share/runtime/arguments.cpp?

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

Marked as reviewed by dcubed (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/18820#pullrequestreview-2007193112
PR Review Comment: https://git.openjdk.org/jdk/pull/18820#discussion_r1569560130