RFR: 8371380: [LOOM] The debug agent should avoid enabling VIRTUAL_THREAD_START/END events when possible

[Chris Plummer]

On Tue, 25 Nov 2025 09:09:50 GMT, Serguei Spitsyn <sspitsyn at openjdk.org> wrote:

>> Don't enabled VIRTUAL_THREAD_START/END events unless absolutely necessary. Solves performance issues when trying to debug apps that create a lot of virtual threads. Details in first comment.
>> 
>> With these changes the Skynet benchmark no longer shows any slowdown when launching with debugging enabled or when attaching the debugger.
>> 
>> Tested with all tier2, tier3, tier5, and tier6 CI testing (with filters to only run svc tests).
>
> src/jdk.jdwp.agent/share/native/libjdwp/eventFilter.c line 1444:
> 
>> 1442:                 } else {
>> 1443:                     error = threadControl_setEventMode(JVMTI_DISABLE, EI_VIRTUAL_THREAD_END, thread);
>> 1444:                 }
> 
> Q: Why is the `THREAD_START` event disabled globally (for all threads) but the `THREAD_END` event is disabled for specific thread?

THREAD_START can't have a thread filter. THREAD_END can. "thread" == requestThread(node) == the thread filter if there is one, so when disabling THREAD_START it could be an actual thread or it could just be NULL. For THREAD_START "thread" should always be NULL, so I suppose I could have just asserted that and used "thread" as the argument instead of "NULL".

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

PR Review Comment: https://git.openjdk.org/jdk/pull/28485#discussion_r2560664560