RFR: 8155524: HotCardCache shouldn't be part of ConcurrentG1Refine

Kim Barrett kim.barrett at oracle.com
Fri Apr 29 06:29:11 UTC 2016


Please review this cleanup that moves ownership of the hot card cache
from the ConcurrentG1Refine class (which doesn't directly use it at
all) to the G1CollectedHeap, where it is also made accessible to the
G1RemSet, as both classes make direct use of it.

This allows the G1RemSet to largely disconnect from the refinement
objects.  (There's still one lingering reference, to
ConcurrentG1Refine::thread_num(), which requires additional work to
clean up.  The G1RemSetSummary also conditionally refers to refinement
objects for logging; see JDK-815525.)

As a result of this change, ConcurrentG1Refine no longer needs the
heap object at construction time.

Also cleaned up includes and forward declarations in
concurrentG1Refine.hpp.

CR:
https://bugs.openjdk.java.net/browse/JDK-8155524

Webrev:
http://cr.openjdk.java.net/~kbarrett/8155524/webrev.00/

Testing:
Ad-hoc nightly test run using G1.




More information about the hotspot-gc-dev mailing list