Note this should be an improvement for ThunderX (though maybe it is not modeled correctly). Not ThunderX2T99 though (only because fusion happens even for add/cbz cases too).

In D34574#789539, @ahatanak wrote:

Is there any cases where we can turn the linker error into a compile time error?

Fine once namespace is renamed

This is fine once the properties are documented.

Can you also change IteratedDominanceFrontier to use this level info instead of computing it on it's own?

Commit complex-return.ll test separately.

Fixed formatting issue.

Added a unittest for the embedded manifest case.

Please rebase after r306193 and post the patch with full context.

This looks sensible to me. I don't know if there are any scenarios in which you'd be using the Microsoft CRT without having _MSC_VER defined, however. Added some people who may have a better idea of that (@compnerd, @majnemer and @rnk)

Comment in the test

Hmm ... last version did not compile. This should be good now.

LGTM

Please add a comment to the test case with the PR number at least.

Minor change in OrderedInstruction invalidation.

LGTM.