Minor tweaks per comments.

In D58548#1407164, @dexonsmith wrote:

I like this idea, and I don’t think the textual IR central is too important. A few things:

• Changes to the IR should always be discussed on llvm-dev. Did this already happen?

Add some wording to LangRef and clang-format.

In D57601#1397317, @reames wrote:

Question for Reviewers: Going through uses of isVolatile in backend code, I see suspicious bits in Lania, ARM, and Hexagon. What's the best way to handle this? I can easily apply the workaround I mention for out of tree backends, or I could make them equally conservative for isAtomic. The other in tree backends appear to be fine, though I'd welcome review from folks knowledgeable of non-x86 backends.

I still don't like it...It's different, unusual, and IMO surprising to have such a wildcard ignore.

In D58260#1399596, @efriedma wrote:

I'm concerned that "nocapture" is not sufficient to describe the necessary property here.

I think this warning (-Wbuiltin-requires-header) doesn't really make sense as its own warning.

OK, I'm happy with all of that.

Looks reasonable to me.

It'd be great to see this somewhat more widely publicized, outside of just the clang community. If libc implementors are aware of the gains and are willing to provide an actually-faster bcmp implementation, it'd be a lot better, than having this optimization that doesn't really optimize anything without users providing their own bcmp implementation.

Address comment.

I've no idea whether buildbot auto-updates or if it requires someone to go manually poke it, but this at least looks reasonable.

Mmm, I think this adding a default /* ignore is setting us up for problems. I'd be okay with a default ignore for '/build*', though.