Bug: humongous arrays allocations OOMEs

[Aleksey Shipilev]

Regression/acceptance test here:
 http://cr.openjdk.java.net/~shade/shenandoah/acceptance-alloc.patch

-Aleksey

On 11/29/2016 11:33 AM, Roman Kennke wrote:
> Seems to be caused by a bug in current heuristics. The heuristics
> refactoring +fixes I'm working on seems to solve it. Stay tuned...
> 
> 
> Roman
> 
> 
> Am Dienstag, den 29.11.2016, 11:03 +0100 schrieb Aleksey Shipilev:
>> Hi,
>>
>> A very simple test:
>>
>> public class Alloc {
>> 	static final int SIZE = Integer.getInteger("size", 1_000_000);
>> 	static Object sink;
>>
>> 	public static void main(String... args) throws Exception {
>> 		for (int c = 0; c < 1000000; c++) {
>> 			sink = new int[SIZE];
>> 		}
>> 	}
>> }
>>
>> OOMEs after some critical array size:
>>
>> $ java -XX:+UseShenandoahGC -Dsize=1  Alloc
>> <OK>
>> $ java -XX:+UseShenandoahGC -Dsize=1000  Alloc
>> <OK>
>> $ java -XX:+UseShenandoahGC -Dsize=1000000  Alloc
>> Exception in thread "main" java.lang.OutOfMemoryError: Java heap
>> space
>> 	at Alloc.main(Alloc.java:12)
>>
>> Thanks,
>> -Aleksey
>>