RFR(xs): 8153201: TestOptionsWithRanges fails with -XX:OldPLABSize=2147483648

Bengt Rutisson bengt.rutisson at oracle.com
Fri Apr 1 05:21:44 UTC 2016


Hi Sangheon,

On 2016-04-01 02:05, sangheon wrote:
> Hi all,
>
> Could I have a couple of reviews for this tiny change for OldPLABSize 
> flag?
>
> We would face an assert complaining too large array size based on 
> OldPLABSize.
> And we already have a constraint for this flag checking same 
> value(ThreadLocalAllocBuffer::max_size) but the constraint function is 
> checking for CMS GC and G1 GC.
> This change is just adding parallel gc on that check routine to limit 
> both min and max.
>
> CR: https://bugs.openjdk.java.net/browse/JDK-8153201
> Webrev: http://cr.openjdk.java.net/~sangheki/8153201/webrev.00

Do we need to check for the GC at all? SerialGC does not use PLABs, but 
that just means that those who use SerialGC should not have OldPALBSize 
on their command lines. But if they do, why should it be ok to set it to 
an arbitrary value? So, I would suggest just removing the GC check all 
together.

Thanks,
Bengt

> Testing: JPRT, runtime/commandline JTREG tests for all platforms.
>
> Thanks,
> Sangheon



More information about the hotspot-gc-dev mailing list