Application is hanging in this stack trace(see below),
and only happen 1 time in production.
kill -3, jstack, ... do not work.

It may be related to CR6872049 CompactibleFreeListSpace::block_size from ::block_start_unsafe() except that this issue is a hang
and not a crash.

----------------- lwp# 12 / thread# 12 --------------------
ffffffff7e3af144 __1cYCompactibleFreeListSpaceKblock_size6kMpknIHeapWord__L_ (4
c148, ffffffff7eb08ae8, 6a800, ffffffff7ea9e000, 6eef1c, 4c000) + 64
ffffffff7e35131c __1cbEBlockOffsetArrayNonContigSpaceSblock_start_unsafe6kMpkv_
pnIHeapWord__ (fffffffdb9036000, fffffffdb95a0000, 10026ec98, ffffffff73c00000,
fffffffdb9036000, ffffffff7ea9e000) + cc
ffffffff7e7006d8 __1cRCardTableModRefBSYprocess_chunk_boundaries6MpnFSpace_pnVD
irtyCardToOopClosure_nJMemRegion_5ppWLL_v_ (10012df10, 10026ec20, 100ab5200, fff
ffffdb95a0000, ffffffff7e3b6120, 102a4f900) + 50
ffffffff7e7003b8 __1cRCardTableModRefBSbFpar_non_clean_card_iterate_work6MpnFSp
ace_nJMemRegion_pnVDirtyCardToOopClosure_pnQMemRegionClosure_bi_v_ (10012df10, f
ffffffdb9580000, ffffffff75000ff0, 14, 0, 1400) + 190
ffffffff7e359c9c __1cRCardTableModRefBSWnon_clean_card_iterate6MpnFSpace_nJMemR
egion_pnVDirtyCardToOopClosure_pnQMemRegionClosure_b_v_ (10012df10, 10026ec20, f
fffffff79bffb10, 100ab5200, ffffffff79bffb20, fffffffd5d000000) + 6c
ffffffff7e35b3e4 __1cLCardTableRSbDyounger_refs_in_space_iterate6MpnFSpace_pnQO
opsInGenClosure__v_ (1001104b0, 10026ec20, ffffffff7e3ae7f8, ffffffff7eae15a0, 7
42cb4, ffffffff7ea9e000) + 9c
ffffffff7e3da168 __1cbDConcurrentMarkSweepGenerationUyounger_refs_iterate6MpnQO
opsInGenClosure__v_ (10011a980, 1001104b0, fffffffd5d000000, 6c3c00, ffffffff7e3
da118, 102f3ce20) + 50
ffffffff7e477800 __1cQGenCollectedHeapYgen_process_strong_roots6MibbnKSharedHea
pOScanningOption_pnQOopsInGenClosure_4_v_ (10011d8e0, 2, 1, 0, 10011d970, 102f3c
e20) + 120
ffffffff7e703db4 __1cNParNewGenTaskEwork6Mi_v_ (ffffffff75eff180, 5, 10011b430,
102f3cc90, 0, 100aba5b8) + 164
ffffffff7e81adc4 __1cKGangWorkerEloop6M_v_ (10012c800, 0, 10011c900, 1, 10011d9
f0, 1831c) + 84
ffffffff7e6f1680 java_start (10012c800, 751, ffffffff7eaffbdc, ffffffff7ea9e000

java options
-d64 -server -Xms512m -Xmx8080m -XX:MaxPermSize=512m -Xss2m -verbosegc
-XX:NewRatio=3 -XX:SurvivorRatio=4 -XX:+UseCMSCompactAtFullCollection
-XX:CMSFullGCsBeforeCompaction=0 -XX:+ParallelRefProcEnabled
-XX:+ExplicitGCInvokesConcurrent -XX:SoftRefLRUPolicyMSPerMB=0
-XX:ParallelGCThreads=10 -XX:+UseConcMarkSweepGC -XX:+UseParNewGC
-XX:+CMSClassUnloadingEnabled -Dsun.rmi.dgc.client.gcInterval=900000
-Dsun.rmi.dgc.server.gcInterval=900000

Filing this bug for further evaluation by engineering.