with recent fastdebug build 20050721085015.dcubed.service_hs_merge-debug from main baseline, Vtest threw assertion error Universe::heap()->is_in_or_null(*loc),"found non oop pointer".

java version:
Java HotSpot(TM) Server VM (build 20050721085015.dcubed.service_hs_merge-debug, mixed mode)
Flags used:
-XX:+UseConcMarkSweepGC -XX:+CMSClassUnloadingEnabled -XX:+CMSPermGenSweepingEnabled
Test machine: j2se-a.west
SunOS j2se-a 5.9 Generic_118558-06 sun4u sparc SUNW,Sun-Fire

The assertion errors happened twice within 4 days.

#head hs_err_pid10341.log ( "hs_err_pid10341.log" is attached )
# An unexpected error has been detected by HotSpot Virtual Machine:
#
# Internal Error (/net/prt-solsparc-q1-19/tmp/PrtBuildDir/workspace/src/share/vm/compiler/oopMap.cpp, 439 [ Patched ]), pid=10341, tid=13
#
# Java VM: Java HotSpot(TM) Server VM (20050721085015.dcubed.service_hs_merge-de
bug mixed mode)
#
# Error: assert(Universe::heap()->is_in_or_null(*loc),"found non oop pointer")

Stack trace:
=>[1] _lwp_kill(0x0, 0x6, 0x0, 0xff33c000, 0xfdff6d64, 0xf95634), at 0xff31f82c
  [2] raise(0x6, 0x0, 0xfa57ebf0, 0xfa57ec70, 0x2fd6c, 0x2fc00), at 0xff2d0a1c
  [3] abort(0xfa57ec90, 0xfec6aa2b, 0x15, 0x11, 0x0, 0x1), at 0xff2b6cd8
  [4] os::abort(0x30400, 0xfa57ec90, 0xfdff6e88, 0xff0175e0, 0xff0175e0, 0x1), at 0xfdfebd60
  [5] VMError::report_and_die(0x31c00, 0xff02df72, 0xfef0771f, 0xfef66401, 0xff02df72, 0x1), at 0xfe2d5024
  [6] report_assertion_failure(0xfef07367, 0x1b3, 0xfef073b9, 0x306f4, 0xfef88b90, 0x30400), at 0xfd851070
  [7] VMError::report(0xfa57f2a8, 0xfec58f76, 0x1, 0xfa57f1d8, 0xfef071d3, 0xfef88b90), at 0xfe2d3a60
  [8] VMError::report_and_die(0xfa57f2a8, 0x2f800, 0x2fad0, 0x2f800, 0x2a000, 0x30700), at 0xfe2d4a88
  [9] report_assertion_failure(0xfec58f23, 0x1b7, 0xfec58f76, 0x306f4, 0xfef88b90, 0x30400), at 0xfd851070
  [10] OopMapSet::all_do(0xfa57f4ec, 0xfb0f4cc8, 0xfa57f500, 0xfa57f9a4, 0xd887f8f4, 0xff014a8c), at 0xfdfd9c44
  [11] OopMapSet::oops_do(0xfa57f4ec, 0xfb0f4cc8, 0xfa57f500, 0xfa57f9a4, 0x31400, 0x21800), at 0xfdfd98fc
  [12] frame::oops_code_blob_do(0xfa57f4ec, 0xfa57f9a4, 0xfa57f500, 0x1, 0xfef88b90, 0xb800), at 0xfd8cdb3c
  [13] Threads::possibly_parallel_oops_do(0xfa57f9a4, 0x0, 0x1, 0xfef88b90, 0xff03ce3c, 0xfa57f4d8), at 0xfe1f5c34
  [14] SharedHeap::process_strong_roots(0x54168, 0x0, 0x1, 0xfa57f9a4, 0xfa57f980, 0x12), at 0xfe11bac4
  [15] GenCollectedHeap::gen_process_strong_roots(0x54168, 0x0, 0x1, 0x0, 0x1, 0xfa57f980), at 0xfd8ef0b8
  [16] ParNewGenTask::work(0xf9d7f5d8, 0x6, 0x639c0, 0x31000, 0xf9d7f5f8, 0x639c0), at 0xfe00a0e8
  [17] GangWorker::run(0x5da00, 0x54458, 0x0, 0x13, 0xac, 0xfef4ca45), at 0xfe300da4
  [18] _start(0x5da00, 0x2, 0xff016b54, 0x21c00, 0xfef88b90, 0x5e610), at 0xfdfea294
How to reproduce the bug:
1. the bug is easier on a bigger machine. I saw the bug on both j2se-a.west and j2se-b.west, but not on smaller systems.
j2se-a/b.west is 24 way * 1200MHZ.
2. log into j2se-b.west, contact ###@###.### for root passwd
3. export JAVA_HOME=<your java home>
4. run /bs/runvtest.ksh -XX:+UseConcMarkSweepGC -XX:+CMSClassUnloadingEnabled -XX:+CMSPermGenSweepingEnabled
5. the crash happened in a few days.