[PATCH] D42451: [GlobalOpt] Fix exponential compile-time with selects.

[Eli Friedman via Phabricator via llvm-commits]

efriedma added a comment.

> Though I wonder if it's possible to trigger a similar exponential behavior with GEP instructions

I don't think so; a GEP has exactly one pointer operand.

> does this happen on real code

I ran into this problem compiling a proprietary codebase.


Repository:
  rL LLVM

https://reviews.llvm.org/D42451