Even with the fix for 6700114 in place, the getcpool001 test continues
to fail. It fails less frequently now, but it still fails. I've verified
that the fix is in place in the HSX-14-B12 bits and in RT_Baseline after
I pushed the fix for 6700114 for HSX-15-B03. It looks like the test
catches more than one problem on almost identical code paths and I didn't
fix the second code path.

Here is my nightly analysis entry:

New nsk.jvmti failures (from 2009.03.04)
    nsk/jvmti/unit/GetConstantPool/getcpool001
        This test failed the following assertion on Solaris SPARC
        Server VM -Xcomp (machine denab):

            Internal Error (src/share/vm/prims/jvmtiThreadState.cpp:203)
            Error: assert(_thread->get_interp_only_mode() == 1,
                "leaving interp only when mode not one")

        This failure mode is covered by the following bug:

            6700114 3/4 Assertion (_thread->get_interp_only_mode() == 1,
                        "leaving interp only when mode not one")

        This bug is in state "fix available" and the fix is included in
        the bits being tested by this nightly.
http://sqeweb.sfbay/nfs/tools/gtee/results/JDK7/NIGHTLY/VM/2010-07-30/G1_GC_Baseline/vm/solaris-i586/client/mixed/solaris-i586_vm_client_mixed_nsk.jvmti.unit.testlist/ResultDir/getcpool001//hs_err_pid15974.log

nsk/jvmti/unit/GetConstantPool/getcpool001
nsk.jvmti.unit.GetConstantPool.getcpool001