code review for more CompiledMethodLoad event fixes (7024234, 7024970)

[Daniel D. Daugherty]

Greetings,

I have a couple more fixes for bugs that have shown up in nightly
testing. Both bugs are race conditions:

   7024234 2/3 jvmti tests fail assert(!_oops_are_stale) failed:
               oops are stale on Win-AMD64
   7024970 2/3 assert(ServiceThread::is_service_thread(Thread::current()))
               failed: Service thread must post enqueue

So far 7024234 has only been observed in nightly on Win-AMD64 configs.
However, I have been able to reproduce 7024234 on my WinXP VMware client.
So far 7024970 has only been observed in nightly on Linux configs.

Here is the URL for the combined webrev:

    http://cr.openjdk.java.net/~dcubed/7024234_7024970-webrev/2/

These fixes have been through several internal rounds of code review
by multiple people so this is more of a heads up that these changes
are coming to OpenJDK7/HSX-21. I'm planning to push these changes
to RT_Baseline (http://hg.openjdk.java.net/jdk7/hotspot-rt/hotspot)
tomorrow (2011.03.15) in order to make this week's push from RT_Baseline
to Main_Baseline (http://hg.openjdk.java.net/jdk7/hotspot/hotspot).

Thanks, in advance, any additional comments or observations.

Dan