RFR(xs): 8220151: SafepointTracing::end_of_last_safepoint_ms should return ms since epoch.
robbin.ehn at oracle.com
Tue Mar 5 12:52:54 UTC 2019
On 2019-03-05 13:10, David Holmes wrote:
> Hi Robbin,
> On 5/03/2019 9:58 pm, Robbin Ehn wrote:
>> Hi all, please review.
>> os::javaTimeNanos and os::javaTimeMillis may use different starting points.
> They are completely unrelated except in the most degenerate of cases on a system
> with no high-res clock.
>> nanos cannot be converted to millies when comparing to absolute ms time.
>> SafepointTracing::end_of_last_safepoint_ms must use os::javaTimeMillis.
> If it's intended to be a timestamp then yes.
> Looks fine to me.
>> Passes t1.
>> Thanks, Robbin
More information about the hotspot-runtime-dev