CRR (S/M): 6929868 G1: introduce min / max young gen size bounds

Bengt Rutisson bengt.rutisson at oracle.com
Wed Sep 7 14:23:47 PDT 2011


Hi again,

I still need one more review for these changes. Anyone feel up to it? It 
is not such a big change, but it is probably a good idea to first review 
Tony's changes "CRR (M): 7084509: G1: fix inconsistencies and mistakes 
in the young list target length calculations", which, by the way, also 
need another review as far as I know...

Bengt

On 2011-09-01 15:01, Bengt Rutisson wrote:
>
> Hi all,
>
> I'd like one more review for this change (Tony has already reviewed it).
>
> http://cr.openjdk.java.net/~brutisso/6929868/webrev.01/
>
> The change is based on Tony's change that he has out for review in the 
> thread "CRR (M): 7084509: G1: fix inconsistencies and mistakes in the 
> young list target length calculations". Since the feature complete 
> date for hs22 is creeping up on us I would like to get this reviewed 
> already now rather than waiting for Tony's changes to go in.
>
> The change tries to make G1 more intuitive in the way it uses the 
> command line flags for setting the young gen size.
>
> Here is a short summary of what the changes contain:
>
> * UseAdaptiveSizePolicy is ignored by G1
> * Setting -Xmn is equivalent to setting -XX:NewSize == -XX:MaxNewSize 
> and will keep the young gen size at the specified value
> * If NewSize < MaxNewSize the young gen size will adjust to a value 
> based on the pause target. But it will not go above MaxNewSize or 
> below NewSize.
> * If NewRatio is set the young gen size will only change when the heap 
> size changes.
> * NewSize and MaxNewSize overrides NewRatio if all of them are specified.
>
> CR:
> 6929868 G1: introduce min / max young gen size bounds
> http://monaco.us.oracle.com/detail.jsf?cr=6929868
>
> Thanks,
> Bengt
>
>



More information about the hotspot-gc-dev mailing list