Recently I observe jdi_regression/nsk.jvmti failures ONLY on machine minh.sfbay

http://vmsqe.sfbay/nightly/mantis/DTWS/results/06-06-05/ServerVM/solx86/mixed/Main_Baseline/nsk.jvmti-NIGHTLY-Main_Baseline-ServerVM-mixed-solx86-2005-06-07-07-41-13/analysis.html
http://vmsqe.sfbay/nightly/mantis/DTWS/results/06-05-05/ServerVM/solx86/mixed/Main_Baseline/JDI_REGRESSION-NIGHTLY-Main_Baseline-ServerVM-mixed-solx86-2005-06-06-06-54-02/analysis.html
http://vmsqe.sfbay/nightly/mantis/DTWS/results/06-04-05/ServerVM/solx86/mixed/Main_Baseline/JDI_REGRESSION-NIGHTLY-Main_Baseline-ServerVM-mixed-solx86-2005-06-05-03-07-40/analysis.html

Here is the Error ID

    494E5445525052455445520E435050014F

Decodes to:

    interpreter.cpp, 335


minh.sfbay is our only SunFire V60x machine with Xeon CPU.
http://vesta.sfbay/cgi-bin/mach.cgi?machine=minh

Here are some details:
* The same test failed only on minh, passed on all other machines.
  I suspect it only failed on Xeon CPU running solaris?!
* The same test passed with fastdebug build, only failed with product
  build.(that's why this failure did not show up in nightly xxx_baseline which
  is running fastdebug build)
* It's only server VM problem, "-client" works fine.

Here is command line to reproduce it on machine minh.sfbay

/net/jano.sfbay/export/disk20/serviceability/tools/bin/runnsk \
  -server -jdk /import/java/jdk1.6.0/solaris-i586 \
  nsk/jvmti/ClearFieldAccessWatch/clrfldw001

###@###.### 2005-06-08 00:48:33 GMT