Re: MPS prstack

[Mattias Engdegård]

27 maj 2024 kl. 16.15 skrev Helmut Eller <eller.helmut@gmail.com>:

>  prstack.stack[prstack.sp] = e;
>  __atomic_store_n (&prstack.sp, prstack.sp + 1, __ATOMIC_SEQ_CST);

Surely that is a severe overkill when all we need is a compiler barrier?
Or have I misunderstood the context the scanners are called in?
Shouldn't they have the same view of memory as the mutator?