[llvm] [CodeGen] Improve ExpandMemCmp for more efficient non-register aligned sizes handling (PR #69942)

[David Sherwood via llvm-commits]

================

----------------
david-arm wrote:

Given github doesn't do support stacked pull requests like Phabricator did, I think @igogo-x86 tried to solve this problem by pushing two commits into this branch. The first commit shows the pre-committed tests, and the second commit does show the effect the code change has on the tests.

https://github.com/llvm/llvm-project/pull/69942