When threads transition state to/from native code they use the memory_serialization_page as a means of performing memory synchronization. Due to the use of a single 4K page, and the requirement for 128byte alignment this is limited to 32 distinct slots. Combined with a poor hashing function we can get excessive contention on these slots and encounter a performance bottleneck.
- relates to
-
JDK-8152292 Consider using proper OS APIs for os::serialize_thread_states
-
- Closed
-
-
-
- Resolved
-
-
-
- Resolved
-
There are no Sub-Tasks for this issue.