RFR(S): 7016955: G1: remove the is_zeroed parameter from the HeapRegion constructor

Bengt Rutisson bengt.rutisson at oracle.com
Fri Sep 14 13:06:02 UTC 2012

Hi all,

This is really convoluted code but I think the change looks good. 
(Thanks Stefan for helping me untangle the control flow.)

There are two comments heapRegion.cpp that I think should be removed now:

  504   // Note that initialize() will set the start of the unmarked 
area of the
  505   // region.

  980   // false ==> we'll do the clearing if there's clearing to be done.


On 2012-09-13 20:39, Jon Masamitsu wrote:
> Basically, it's deleting the G1OffsetTableContigSpace
> version of initialize(), using the ContiguousSpace initialize()
> and doing the little bit of extra work the constructor.
> And, of course, dumping the is_zeroed parameter.
> Looks good.
> On 09/13/12 10:08, John Cuthbertson wrote:
>> Hi Everyone,
>> Can I have a volunteer to review the changes, which were contributed 
>> by Brandon Mitchell at Twitter, for this CR? The webrev can be found 
>> at: http://cr.openjdk.java.net/~johnc/7016955/webrev.0/
>> Summary:
>> When Tony removed the zero filling thread, the is_zeroed parameter in 
>> the HeapRegion constructor became unused. These changes remove that 
>> unused parameter and clean up the HeapRegion initialization code.
>> Testing:
>> SPECjvm98 and SPECjbb2005 on Linux (Brandon); GC test suite on 
>> solaris (x86 and sparc) and jprt (JohnC)
>> Thanks,
>> JohnC

More information about the hotspot-gc-dev mailing list