JIT has much chrome UI as possible, and identify hot spots where we should be careful in the future. If possible we shoud create some test in these places to be sure we don't regress.
- Push a patch to allow using tracevis on a per function basis
- Identify the "hot" places
- Prevent the trace to abort into the essentials functions
- Try to optimize the "hot" places (places where tracing abort, but also places where monitoring slow us down)
- Correct all the trace abort