Memory leak in jdk1.7.0 (42 MB per day)

FULL PRODUCT VERSION :
java version "1.7.0_02"
Java(TM) SE Runtime Environment (build 1.7.0_02-b13)
Java HotSpot(TM) Server VM (build 22.0-b10, mixed mode)

FULL OS VERSION :
SunOS desna 5.10 Generic_141444-09 sun4u sparc SUNW,Sun-Fire-V490

A DESCRIPTION OF THE PROBLEM :
We've tried jdk 1.7.0_02. Ten days of running give 420 MB memory leak of the following objects:<BR>
- java.lang.management.MemoryUsage,<BR>
- [C (array of char),<BR>
- java.util.HashMap$Entry,<BR>
- [Ljava.util.HashMap$Entry (array of HashMap$Entry),<BR>
and some others.

This doesn't happen on jdk1.6.x.

First output of the "jmap -histo:live" command:

     num #instances #bytes class name
    ----------------------------------------------
       1: 229527 14926888 [C
       2: 289290 13885920 java.lang.management.MemoryUsage
       3: 321029 10272928 java.util.HashMap$Entry
       4: 69923 10262184 <constMethodKlass>
       5: 69923 9527672 <methodKlass>
       6: 7048 7787040 <constantPoolKlass>
       7: 241693 7734176 java.lang.String
       8: 2038 5898408 [Ljava.util.concurrent.ConcurrentHashMap$HashEntry;
       9: 7048 5479056 <instanceKlassKlass>
      10: 5954 4499552 <constantPoolCacheKlass>
      11: 67844 4091672 [Ljava.util.HashMap$Entry;
      12: 41250 3942848 [B
      13: 65649 3151152 java.util.HashMap
      14: 71891 2875640 java.util.TreeMap$Entry
    ...
    Total 2320965 138000120

The last output of the "jmap -histo:live" command done in 10 days after the first:

     num #instances #bytes class name
    ----------------------------------------------
       1: 3147110 151061280 java.lang.management.MemoryUsage
       2: 3178875 101724000 java.util.HashMap$Entry
       3: 1087332 53822632 [C
       4: 1099503 35184096 java.lang.String
       5: 639442 31529224 [Ljava.util.HashMap$Entry;
       6: 637247 30587856 java.util.HashMap
       7: 629422 25176880 [Ljava.lang.management.MemoryUsage;
       8: 314711 17623816 com.sun.management.GcInfo
       9: 70107 10292776 <constMethodKlass>
      10: 631864 10109824 java.util.HashMap$EntrySet
      11: 314711 10070752 sun.management.GcInfoCompositeData
      12: 70107 9552696 <methodKlass>
      13: 7075 7817080 <constantPoolKlass>
      14: 314713 7554128 [Ljava.lang.Integer;
      15: 2048 5898744 [Ljava.util.concurrent.ConcurrentHashMap$HashEntry;
      16: 7075 5497200 <instanceKlassKlass>
      17: 315792 5052672 java.lang.Integer
      18: 47680 4912352 [B
    ...
    Total 13206419 558217856

I also have 8 other histograms, made after each day of the tests. They show linear number of objects increasing. This is definitely not a noise. It is a stable leak 42 MB per day.

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

THE PROBLEM WAS REPRODUCIBLE WITH -server FLAG: Yes

REGRESSION. Last worked in version 6u29

STEPS TO FOLLOW TO REPRODUCE THE PROBLEM :
Unfortunately, we haven't extracted the issues into a sample code. So far this is the whole our big product which reproduces the problem.

REPRODUCIBILITY :
This bug can be reproduced always.

CUSTOMER SUBMITTED WORKAROUND :
Move to the jdk 1.6.x
NOTE: if this is a critical customer issue it must go through customer support.