RFR: 8153394: Add Unified Logging to make it easy to trace time taken in initPhase2

Rachel Protacio rachel.protacio at oracle.com
Wed Jun 15 14:13:57 UTC 2016


Thanks for the review, Robbin!

Basically, it is too early in the VM to directly do 
"log_info(<tag>)(<msg>)" for example, but the TraceTime class was made 
to circumvent that by relegating the logging/timer to a different part 
of the code. When I commented on JIRA, I had forgotten about the 
TraceTime workaround.

Rachel

On 6/15/2016 8:56 AM, Robbin Ehn wrote:
> Hi Rachel
>
> Looks good!
>
> Out of curiosity, what is difference since you wrote
> "it's too early for Unified Logging."
> in jira ?
>
> Thanks!
>
> /Robbin
>
> On 06/14/2016 10:50 PM, Rachel Protacio wrote:
>> Hello!
>>
>> Please review this small change, adding unified logging for Phase2 
>> initialization, using the TraceTime class. (A jigsaw enhancement, 
>> which is why it's going in after FC.)
>>
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8153394
>> Open webrev: http://cr.openjdk.java.net/~rprotacio/8153394.00/
>>
>> Output sample:
>>
>>    $ java -Xlog:startuptime+modules -version
>>    [0.410s][info][modules,startuptime] Phase2 initialization, 0.1302802
>>    secs
>>    java version "9-internal"
>>    Java(TM) SE Runtime Environment (fastdebug build
>>    9-internal+0-2016-06-09-092114.rprotaci.8153394)
>>    Java HotSpot(TM) 64-Bit Server VM (fastdebug build
>>    9-internal+0-2016-06-09-092114.rprotaci.8153394, mixed mode)
>>
>> Includes added conditions to logging/StartupTimeTest.java. Passes JPRT.
>>
>> Thanks,
>> Rachel



More information about the hotspot-runtime-dev mailing list