RFR(S): 8220672: [TESTBUG] TestCPUSets should check that cpuset does not exceed available cores

Leonid Mesnik leonid.mesnik at oracle.com
Mon May 20 20:12:41 UTC 2019


+1

This patch is reasonable  and a good workaround. However new issue to find more robust/correct would be nice.

Leonid

> On May 19, 2019, at 9:01 PM, David Holmes <david.holmes at oracle.com> wrote:
> 
> Hi Misha,
> 
> On 18/05/2019 8:36 am, mikhailo.seledtsov at oracle.com wrote:
>> Please review this change that fixes the TestCPUSets test. In short, in some configurations the size of CPU set
>> exceeds the available amount of cores; see bug description and comments for details.
>> The fix (or some may say work-around) in this change is to limit the size of the set to the number of available CPUs.
> 
> I think the workaround is reasonable and should be pushed. But as per the bug report I think we still need to get to the bottom of where the misinformation is coming from and whose bug it really is.
> 
> Thanks,
> David
> -----
> 
>>    JBS: https://bugs.openjdk.java.net/browse/JDK-8220672
>>    Webrev: http://cr.openjdk.java.net/~mseledtsov/8220672.00/
>>    Testing:
>>      - ran the affected test on hosts where issue occurred
>>      - ran the affected test on a variety of hosts in build-and-test cluster
>>      The problem does not show up anymore.
>> Thank you,
>> Misha



More information about the hotspot-runtime-dev mailing list