RFR(S): 8184271: Time related C1 intrinsics produce inconsistent results when floating around
Tobias Hartmann
tobias.hartmann at oracle.com
Thu Jul 13 14:34:10 UTC 2017
Thanks Vladimir!
Best regards,
Tobias
On 13.07.2017 16:33, Vladimir Ivanov wrote:
> Looks good.
>
> Best regards,
> Vladimir Ivanov
>
> On 7/13/17 5:24 PM, Tobias Hartmann wrote:
>> Hi,
>>
>> please review the following patch:
>> https://bugs.openjdk.java.net/browse/JDK-8184271
>> http://cr.openjdk.java.net/~thartmann/8184271/webrev.00/
>>
>> The problem is that the C1 intrinsics for System.nanoTime(), System.currentTimeMillis() and JVM.counterTime() are not pinned and may produce inconsistent results when floating around (see tests). The solution is to not unpin them.
>>
>> Thanks to Vladimir I. for the analysis!
>>
>> Tested with JPRT and RBT (running).
>>
>> Thanks,
>> Tobias
>>
More information about the hotspot-compiler-dev
mailing list