[mrushton14]

I do embedded Linux development and I see bugs related to this stuff frequently. What can make things more tricky is different architectures have different ordering guarantees. Anyone know of any cpu architectures that intentionally sacrifice performance for a simpler less bug prone model? I'm thinking of markets like aviation where bugs like this seem particularly scary.

[duskwuff]

There's no need for a separate architecture to avoid concurrency issues; a simpler solution is to simply use a uniprocessor system. (And be careful with interrupts.)