RFR 8214124: [TESTBUG] Bugs in runtime/NMT/MallocStressTest.java

Aleksey Shipilev shade at redhat.com
Tue Nov 20 17:03:52 UTC 2018


On 11/20/18 5:48 PM, Zhengyu Gu wrote:
> This test assumes Math.abs(random.nextInt()) >= 0, but it is *not* always true, as
> Math.abs(Integer.MIN_VALUE) == Integer.MIN_VALUE, which is a negative value [1].
> 
> The first instance could result miscounting of allocation size and the second one could result index
> OOB.
> 
> Bug: https://bugs.openjdk.java.net/browse/JDK-8214124
> Webrev: http://cr.openjdk.java.net/~zgu/8214124/webrev.00/

Looks good.

Fun trivia: we have been ripping our hairs out recently chasing a similar testbug in Shenandoah
tests: http://mail.openjdk.java.net/pipermail/shenandoah-dev/2018-October/007916.html

-Aleksey



More information about the hotspot-runtime-dev mailing list