Code review comment for lp:~raof/mir/fix-threaded-dispatcher-death-test-race

I can't say for sure but think you're imagining atomic solves more racing problems than it does.

Atomics I think are implemented using spinlocks or similar. So mutual exclusion is guaranteed but they are not designed to solve the large scale ordering problems of thread synchronization (where those threads are not yet accessing memory so memory_order is irrelevant).

Roughly speaking I agree with Alan's assessment that either atomic is not required or the test is broken.