RFR (S): 8134631: G1DummyRegionsPerGC fires assert of assert(words <= filler_array_max_size()) failed: too big for a single object

Thomas Schatzl thomas.schatzl at oracle.com
Thu Nov 12 08:43:47 UTC 2015


Hi Sangheon,

On Wed, 2015-11-11 at 14:25 -0800, sangheon.kim wrote:
> Hi Tom,
> 
> On 11/11/2015 02:11 PM, Tom Benson wrote:
> > Hi Sangeon,
> > I do prefer this version, but I think it would be a good idea to 
> > preserve the original value and restore it like you did in your first 
> > version, rather than re-setting it specifically to 
> > _humongous_object_threshold_in_words.
> Okay, I changed it as you said.
> 
> http://cr.openjdk.java.net/~sangheki/8134631/webrev.02/

  looks good.

I agree with Tom that it is much better (less intrusive) to temporarily
patch this up in the debug code.

Instead of the manual re-assignments you could simply use

SizeTFlagSetting tf(&_filler_array_max_size, word_size);

That would make sure that after the scope is closed, the value will be
reset. It also saves two LOC.
If you make that change, I do not need a re-review.

Thanks,
  Thomas




More information about the hotspot-gc-dev mailing list