JDK-8016479 : UseParallelGC collector does not return space to the OS when the heap shrinks.
Type:Bug
Component:hotspot
Sub-Component:gc
Affected Version:hs24,hs25
Priority:P2
Status:Resolved
Resolution:Cannot Reproduce
Submitted:2013-06-12
Updated:2014-03-20
Resolved:2013-06-18
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.
When the heap shrinks the space is not uncommitted as is expected.
Comments
Using a small test program, it has been demonstrated that UseParallelGC does shrink
the committed space in the heap. This is both on hs24 and hs25. I've observed
some behavior with runThese -quick that may be another problem. Still investigating
that and will open a different CR if it is a real problem.
18-06-2013
Finally got a reply from above. This is now considered a P2 for hs24.
18-06-2013
Triaged for hs25.
17-06-2013
I was looking at the behavior on hotspot 25 and from that drew the conclusion that there was not a problem
but I'm looking at hotspot 24 now and there is different behavior (maybe). Needs more investigation.
14-06-2013
I've the same runThese test that I recall running when I first thought this was a bug. From what I see now the problem does not exist.
14-06-2013
Will be closing this as not a bug.
14-06-2013
Yes, this is for density. If possible it should be fixed in hs24 since it is blocking other projects that are shipping with hs24 later this year.
13-06-2013
Moving this to 8-pool. I think this is for density, which is not hs24 or hs25.
13-06-2013
I think the problem is in PSVirtualSpace code. I did an initial investigation when I first realized this was
a problem and that's as far as I got. Don't take that as a fact. More a recollection.