[PATCH] D15668: [attrs] Split off the forced attributes utility into its own pass that can be potentially run much earlier than FuncitonAttrs proper.

[Chandler Carruth via llvm-commits]

chandlerc added a comment.

I didn't introduce this functionality, so I'm not particularly invested in its current design, but I'd like to save that discussion for later. I'd suggest something on llvm-dev and CC-ing the author. The question is, does this refactoring make sense given that the functionality is already in the tree...


================
Comment at: lib/Transforms/IPO/ForceFunctionAttrs.cpp:51
@@ +50,3 @@
+      .Case("readonly", Attribute::ReadOnly)
+      .Case("argmemonly", Attribute::ArgMemOnly)
+      .Case("returns_twice", Attribute::ReturnsTwice)
----------------
probinson wrote:
> I see it's a cut-and-paste but this one is out of order.
Yea, I'd like to leave cleanups of this for follow-up. I really just want to move this code first and make sure folks are comfortable with that.


http://reviews.llvm.org/D15668