United StatesChange Country, Oracle Worldwide Web Sites Communities I am a... I want to...
Bug ID: JDK-8013651 NMT: reserve/release sequence id's in incorrect order due to race
JDK-8013651 : NMT: reserve/release sequence id's in incorrect order due to race

Details
Type:
Bug
Submit Date:
2013-04-30
Status:
Closed
Updated Date:
2013-06-21
Project Name:
JDK
Resolved Date:
2013-06-13
Component:
hotspot
OS:
Sub-Component:
runtime
CPU:
Priority:
P2
Resolution:
Fixed
Affected Versions:
hs24,hs25
Fixed Versions:
hs24 (b49)

Related Reports
Backport:
Backport:
Backport:
Relates:

Sub Tasks

Description
When developing new tests for NMT I ended up with: 

#  Internal Error (..\..\src\share\vm\services\memSnapshot.cpp:623), pid=6180, tid=10848
#  assert(new_rec->is_allocation_record()) failed: Sanity check

Zhengyu looked at it and it seems like it's a race where we end up with the wrong sequence id when one thread is releasing memory and another thread is reserving and gets the same address but they end up with incorrectly ordered sequence numbers.
                                    

Comments
RULE bigapps/Kitchensink/stability Crash Internal Error ...memTracker.hpp...assert(base > ...) failed: wrong base address

one more failure: http://aurora.ru.oracle.com/functional/faces/RunDetails.xhtml?names=235016.JAVASE.PIT.VM-652


                                     
2013-06-11
updating priority to P2 because it causes bigapps crashes, release criterion. ILW = HMM
                                     
2013-06-11
URL:   http://hg.openjdk.java.net/hsx/hsx24/hotspot/rev/2d9b536bb027
User:  zgu
Date:  2013-06-13 14:41:18 +0000

                                     
2013-06-13
The regression test was not created for this issue (JDK-8016595 closed as wont-fix). It is hard to reproduce w/o regression test. Closing with noreg-hard label. 
                                     
2013-06-21



Hardware and Software, Engineered to Work Together