RFR(s): 8198268: Add time argument to ConcurrentGCTimer::register_gc_pause_start/_end
per.liden at oracle.com
Mon Feb 19 07:46:44 UTC 2018
On 02/17/2018 09:32 PM, Kim Barrett wrote:
>> On Feb 16, 2018, at 8:40 AM, Per Liden <per.liden at oracle.com> wrote:
>> Patch to allow a GC to register a time when calling ConcurrentGCTimer::register_gc_pause_start/end. This makes the ConcurrentGCTimer API more in line with all other register_* functions in GCTimer and ConcurrentGCTimer, which all take an optional time argument.
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8198268
>> Webrev: http://cr.openjdk.java.net/~pliden/8198268/webrev.0/
> The change to add the optional arguments looks good.
Thanks for reviewing.
> However, why do
> we have GCTimer::register_gc_pause_start/end? Couldn't
> ConcurrentGCTimer::register_gc_pause_start/end just call (public)
> register_gc_phase_start/end directly?
You mean call the function on _time_partitions instead of the super
class? I guess it's just a matter or taste/design. The STWGCTimer is
kind of using the same pattern. I didn't want top mess with the actual
implementation in this patch, just expose the new argument.
More information about the hotspot-gc-dev