We were seeing very occasional crashes that looked like corrupted ThreadPlans. This happens when we go to destroy a process which is running, and the attempt to interrupt the process fails so we proceed to tear down the process anyway, but in the middle of doing that the stop event comes in and we crash trying to handle it. We fixed that by protecting the ThreadPlanStack with a mutex so that either the tear down would win, and the event would get discarded, or the event would get handled, and then we would continue with the teardown.
After that change was promulgated we didn't see any more of these crashes but last week I got one crash report that can only be explained by corruption of the ThreadPlanStackMap - which I hadn't added a mutex to. It looks like we need one there as well.