JDK-8140585 : PLAB statistics are flushed too late
  • Type: Bug
  • Component: hotspot
  • Sub-Component: gc
  • Affected Version: 9
  • Priority: P4
  • Status: Resolved
  • Resolution: Fixed
  • Submitted: 2015-10-27
  • Updated: 2015-11-29
  • Resolved: 2015-11-10
The Version table provides details related to the release that this issue/RFE will be addressed.

Unresolved : Release in which this issue/RFE will be addressed.
Resolved: Release in which this issue/RFE has been resolved.
Fixed : Release in which this issue/RFE has been fixed. The release containing this fix may be available for download as an Early Access Release or a General Availability Release.

To download the current JDK release, click here.
JDK 9
9 b94Fixed
Related Reports
Blocks :  
JDK-8139149 - Split G1 evacuate_collection_set into multiple steps
Relates :  
JDK-8030849 - Investigate high fragmentation/waste in some situations during allocation during GC in G1
Description
PLAB statistics are flushed after we recalculate the new desired plab sizes.

This leads to PLAB size adjustment appearing to be for the next GC.
Comments
I think the best option is to move the calls to adjust_desired_plab_sz() in G1DefaultAllocator::release_gc_alloc_regions() after the call to per_thread_states->flush() in G1CollectedHeap::evacuate_collection_set(). Adding a "blocker" reference for 8139149 since that code also touches the same area, and the overhead of creating merge errors is not worth waiting a few days.
27-10-2015
Found by M. Chernov
27-10-2015