On Wed, 24 Apr 2024 16:04:30 GMT, Serguei Spitsyn <sspit...@openjdk.org> wrote:
> This is a fix of the following JVMTI scalability issue. A closed benchmark
> with millions of virtual threads shows 3X-4X overhead when a JVMTI agent has
> been loaded. For instance, this is observable when an app is executed under
> control of the Oracle Studio `collect` utility.
> For performance analysis, experiments and numbers, please, see the comment
> below this description.
>
> The fix is to replace the global counter `_VTMS_transition_count` with the
> mark bit `_VTMS_transition_mark` in each `JavaThread`'.
>
> Testing:
> - Tested with mach5 tiers 1-6
src/hotspot/share/prims/jvmtiEnvBase.cpp line 1638:
> 1636: // Iterates over all JavaThread's, counts VTMS transitions and
> restores
> 1637: // jt->jvmti_thread_state() and jt->jvmti_vthread() for VTMS
> transition protocol.
> 1638: void count_transitions_and_correct_jvmti_thread_states() {
The method doesn't count anything anymore.
Rename it to `correct_jvmti_thread_states()`?
Comment needs to be updated too.
src/hotspot/share/prims/jvmtiThreadState.cpp line 501:
> 499: oop vt = JNIHandles::resolve_external_guard(vthread);
> 500: java_lang_Thread::set_is_in_VTMS_transition(vt, false);
> 501: assert(thread->VTMS_transition_mark(), "sanity ed_heck");
Suggestion:
assert(thread->VTMS_transition_mark(), "sanity check");
src/hotspot/share/runtime/javaThread.hpp line 668:
> 666: void toggle_is_disable_suspend() { _is_disable_suspend =
> !_is_disable_suspend; };
> 667:
> 668: bool VTMS_transition_mark() { return
> Atomic::load(&_VTMS_transition_mark); }
Suggestion:
bool VTMS_transition_mark() const { return
Atomic::load(&_VTMS_transition_mark); }
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/18937#discussion_r1580096556
PR Review Comment: https://git.openjdk.org/jdk/pull/18937#discussion_r1580101609
PR Review Comment: https://git.openjdk.org/jdk/pull/18937#discussion_r1580108674
