[PATCH] D106573: Enabling the copy-constant-to-alloca optimization in more instances

[Mohammad Fawaz via Phabricator via llvm-commits]

mfawaz added a comment.

In D106573#2909860 <https://reviews.llvm.org/D106573#2909860>, @jeroen.dobbelaere wrote:

> This patch triggers a (latent) infinite loop problem in instcombine when the memcpy cannot be removed.
> It can be seen when replacing the 'i1 false' of the llvm.memcpy in test11 with a 'i1 true', making the memcpy volatile.

New differential with the fix here: https://reviews.llvm.org/D106950


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D106573/new/

https://reviews.llvm.org/D106573