RFR (s): 8143252: Clean up G1CollectedHeap interface

Derek White derek.white at oracle.com
Thu Nov 19 15:52:14 UTC 2015


Hi Mikael,

  On 11/19/15 3:41 AM, Mikael Gerdin wrote:
> Hi Derek,
>
> On 2015-11-19 06:01, Derek White wrote:
>> This is a small cleanup that:
>>
>>   * Deletes an undefined method declaration
>>     G1CollectedHeap::collected_locked()
>
> While you're at it, would you mind removing the following declaration 
> as well?
>
> 1355   // Update mod union table with the set of dirty cards.
> 1356   void updateModUnion();
> 1357
> 1358   // Set the mod union bits corresponding to the given 
> memRegion.  Note
> 1359   // that this is always a safe operation, since it doesn't clear 
> any
> 1360   // bits.
> 1361   void markModUnionRange(MemRegion mr);
> 1362
>
> G1 has never had a mod union table since its integration in the open 
> source jvm (AFAIK).

OK. I'm surprised with all of the source analysis tools we have that 
this kind of error is never detected!
>>   * Renames method to be more descriptive:
>>     G1CollectedHeap::do_collection() ---> do_full_collection()
>
> Nice to clean up the naming confusion here.

I was so confused I had to take a couple of hours and trace out all of 
the paths into GC, and noticed these issues along the way.
>
>>   * Removes unused parameter "word_size" from (newly renamed)
>>     do_full_collection() and resize_if_necessary_after_full_collection()
>>
>> CR: https://bugs.openjdk.java.net/browse/JDK-8143252
>> Webrev: http://cr.openjdk.java.net/~drwhite/8143252/webrev.01/
>
> Looks good to me, I don't need to see an updated webrev with the mod 
> union declarations removed.
>
> /Mikael
Thanks Mikael!

  - Derek


More information about the hotspot-gc-dev mailing list