RFR: Backport of G1 sorted freelist

Jesper Wilhelmsson jesper.wilhelmsson at oracle.com
Mon Mar 24 21:23:46 UTC 2014


Hi,

Disregard this RFR. The change that caused the rejected hunk was backported 
earlier today so now the patch from 9 applies cleanly. Pushing to 8u now.
/Jesper


Jesper Wilhelmsson skrev 24/3/14 21:34:
> Hi,
>
> Would someone like to have a look at this backoprt of sorting the G1 freelist to
> 8u. The patch from 9 applied with one minor rejection. The rejected hunk was
>
> --- g1CollectedHeap.cpp
> +++ g1CollectedHeap.cpp
> @@ -5907,7 +5906,7 @@
>       _cg1r->hot_card_cache()->reset_card_counts(hr);
>     }
>     hr->hr_clear(par, true /* clear_space */, locked /* locked */);
> -  free_list->add_as_head(hr);
> +  free_list->add_ordered(hr);
>   }
>
>   void G1CollectedHeap::free_humongous_region(HeapRegion* hr,
>
>
> and the reason for it being rejected was that the line above the changed line
> now has one less argument:
>
> 5984     _cg1r->hot_card_cache()->reset_card_counts(hr);
> 5985   }
> 5986   hr->hr_clear(par, true /* clear_space */);
> 5987   free_list->add_as_head(hr);
> 5988 }
> 5989
> 5990 void G1CollectedHeap::free_humongous_region(HeapRegion* hr,
>
>
> So I changed the line above manually.
>
> Full 8u webrev: http://cr.openjdk.java.net/~jwilhelm/8036025/webrev.8u/
> Final 9 webrev: http://cr.openjdk.java.net/~jwilhelm/8036025/webrev.4/
>
> Thanks!
> /Jesper



More information about the hotspot-gc-dev mailing list