FULL PRODUCT VERSION :
Windows:
JRE 1.8.0_60-b27 / JVM 64-Bit Server 25.60-b23, mixed mode
JRE 1.7.0_80-b15 / JVM 64-Bit Server 24.80-b11, mixed mode

Debian:
JRE 1.7.0_72-b14 / JVM 64-Bit Server VM 24.72-b04, mixed mode

FULL OS VERSION :
Microsoft Windows [Version 6.1.7601]
Debian 3.2.60-1+deb7u3 x86_64

A DESCRIPTION OF THE PROBLEM :
I've got exactly the same problem and behaviour as already described in https://bugs.openjdk.java.net/browse/JDK-7191630 (ReentrantReadWriteLock in inconsistent state), but ...
- I can't comment there
- It is reproducible in current/latest versions of JRE (Java7 & Java8)
- It is easily reproducible on my product - dead lock appears in around 2-3 minutes after starting a load test.

Visual VM generates 2-3 thread dumps like this one
   java.lang.Thread.State: WAITING (parking)
    at sun.misc.Unsafe.park(Native Method)
    - parking to wait for <0x000000066f3336d8> (a java.util.concurrent.locks.ReentrantReadWriteLock$FairSync)
...
    at java.util.concurrent.locks.ReentrantReadWriteLock$ReadLock.lock(ReentrantReadWriteLock.java:731)

Nothing is holding lock on 0x000000066f3336d8


THE PROBLEM WAS REPRODUCIBLE WITH -Xint FLAG: Did not try

THE PROBLEM WAS REPRODUCIBLE WITH -server FLAG: Yes

STEPS TO FOLLOW TO REPRODUCE THE PROBLEM :
I do not know how to reproduce it outside of my product

EXPECTED VERSUS ACTUAL BEHAVIOR :
I expect ReentrantReadWriteLock to enter into 'read' locks when there are no active 'write' lock. In my condition, when process is paused in debugger I see that "ReentrantReadWriteLock.sync.exclusiveOwnerThread" is null, but nothing can enter into that lock (neither read nor write). Also, thread dump does not show holds on problemous lock.
REPRODUCIBILITY :
This bug can be reproduced often.

CUSTOMER SUBMITTED WORKAROUND :
I've tried:
- latest versions of JDK
- using -XX:+UseMembar recommended on some forums
- Fair/Unfair ReentrantReadWriteLocks

Nothing helps.