Recommended pattern to use LLVM JIT in a multithreaded application

Since my last question was a bit long, I’ll keep this short:

What is the “best practice” way to use LLVM JIT in a multithreaded application? I’ve noticed there can only be one JIT ExecutionEngine present in the application, and putting the engine in a singleton instance doesn’t seem to work.

Thanks,
Eric Yew

I forgot the details. But I believe your application needs to manage the lock to the module provider.

Evan