[PATCH] D141748: [WoA] Use fences for sequentially consistent stores/writes

[Eli Friedman via Phabricator via llvm-commits]

efriedma added a subscriber: BillyONeal.
efriedma added a comment.

The explanation makes sense.

Can we make this specifically apply to _Interlocked* functions, instead of all atomics on Windows targets?  I'd prefer not to impose this performance penalty on other users of atomics if we can avoid it.

It looks like for atomic rmw ops, MSVC generates one barrier, not two; can we do the same?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D141748/new/

https://reviews.llvm.org/D141748