RFR(XXS): 8022883: Assertion failed: sweptCount >= flushedCount + markedCount + zombifiedCount in closed/com/oracle/jfr/compiler/CodeSweeperSweepFlushTest.java

Albert Noll albert.noll at oracle.com
Tue Sep 17 22:15:27 PDT 2013


Vladimir, Igor, thanks for the reviews.

Yes, yet another variable. I think it's good for now since we plan to 
introduce
a separate sweeper thread. With the separate sweeper thread, we do not need
'NmethodSweepFraction' anymore, so we can simplify things a lot.

Best,
Albert



On 17.09.2013 20:25, Igor Veresov wrote:
> Looks good. Although it's the third counter in this loop that is 
> incremented by one.
>
> igor
>
> On Sep 17, 2013, at 1:19 AM, Albert Noll <albert.noll at oracle.com 
> <mailto:albert.noll at oracle.com>> wrote:
>
>> Hi,
>>
>> please review this small patch:
>>
>> http://cr.openjdk.java.net/~anoll/8022883/ 
>> <http://cr.openjdk.java.net/%7Eanoll/8022883/>
>>
>> https://bugs.openjdk.java.net/browse/JDK-8022883
>>
>> Problem: The number of nmethods that are visited by the sweeper is 
>> handed over incorrectly to JFR. More specifically, the 'todo' is only 
>> an estimate of the visited nmethods, not the actual number (see 
>> existing comment).
>>
>> Solution: Provide correct number of visited nmethods.
>>
>> Testing: Failing test case passes.
>>
>> Many thanks in advance for reviewing this patch.
>>
>> Best,
>> Albert
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/hotspot-compiler-dev/attachments/20130918/3c457447/attachment.html 


More information about the hotspot-compiler-dev mailing list