RFR: 8152962: CMSCollector::shouldConcurrentCollect incorrectly logs against the debug stream

Stefan Karlsson stefan.karlsson at oracle.com
Thu Apr 7 17:35:19 UTC 2016


Thanks, Stefan,

StefanK

On 2016-04-06 11:07, Stefan Johansson wrote:
> Hi StefanK,
>
> On 2016-04-06 09:36, Stefan Karlsson wrote:
>> Hi all,
>>
>> Please review this patch to fix an incorrect log level in 
>> CMSCollector::shouldConcurrentCollect.
>>
>> http://cr.openjdk.java.net/~stefank/8152962/webrev.01
>> https://bugs.openjdk.java.net/browse/JDK-8152962
> Looks good,
> StefanJ
>>
>> The patch sets up one LogTarget instance and uses it throughout the 
>> function. That prevents us from accidentally logging at the wrong log 
>> level.
>>
>> The patch also fixes a pre-existing bug, that we hit the following 
>> assert in debug builds:
>>
>> ~LogStreamNoResourceMark() {
>>   guarantee(_current_line.size() == 0, "Buffer not flushed. Missing 
>> call to print_cr()?");
>> }
>>
>> This was fixed by adding a newline to the CMSStats::print_on call.
>>
>> Thanks,
>> StefanK
>>
>



More information about the hotspot-gc-dev mailing list