RFR(XS) 8076319: jstat verified class fix

[harold seigel]

Hi Coleen,

Thanks for the review!

I reworded the comment to say:

       // Timer includes any side effects of class verification (resolution,
       // etc), but not recursive calls to Verifier::verify().

because instead of the code preceding calls to verify_code(), the code 
is now in Verifier::verify().

Harold

On 5/29/2015 12:44 PM, Coleen Phillimore wrote:
>
> This looks good but I don't understand this comment:
>
> *+   // etc), but not recursive entry into verify_code().*
>
> Can you reword?
>
> Thanks,
> Coleen
>
> On 5/28/15 2:17 PM, harold seigel wrote:
>> Hi,
>>
>> Please review this small change so that the PerfData 
>> sun.cls.classVerifyTime counter only reports on the number of classes 
>> that are actually verified.  The changes were tested with JCK lang 
>> and VM, and JTREG and NSK Quick tests to check for regressions.
>>
>> To test the actual fix, I ran the Queens program and the number of 
>> verified classes changed from 335 (without the fix) to 1. When Queens 
>> was run with -verify, the number of verified classes went back to 335.
>>
>> Open webrev: http://cr.openjdk.java.net/~hseigel/bug_8076319/
>>
>> JBS Bug: https://bugs.openjdk.java.net/browse/JDK-8076319
>>
>> Thanks, Harold
>