RFR (S): 8010818: NPG: Remove metaspace memory pools
erik.helin at oracle.com
Wed Mar 27 02:23:59 PDT 2013
On 03/27/2013 10:31 AM, Stefan Karlsson wrote:
> Looks good.
> On 03/26/2013 04:03 PM, Erik Helin wrote:
>> Hi all,
>> I pushed "8000754: NPG: Implement a MemoryPool MXBean for Metaspace"
>> last Friday (April 22). Unfortunately, the change triggered a couple
>> of bugs:
>> - http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8010719
>> 8010719 : NPG: The metaspace memory pools acquires Metaspace
>> allocation lock out of order with Service_lock
>> - http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8010734
>> 8010734 : NPG: The test MemoryTest.java needs to be updated to
>> support metaspace
>> - http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8010730
>> 8010730 : NPG: Bad assert in JVM_ENTRY(jobject,
>> jmm_GetMemoryUsage(JNIEnv* env, jboolean heap))
>> In order to not cause disturbance upstream (in hotspot-main), it is
>> better to revert the change 8000754, since these bugs are only
>> triggered due to the metaspace memory pools.
>> The reverse patch was created by running:
>> hg diff -r 4321 -r 4322 --reverse
>> in the hotspot-gc/hotspot repistory.
>> - All tests that failed during nightly testing are now passing
>> - JPRT
More information about the hotspot-gc-dev