[PATCH] D30751: [MachineCopyForwarding] Add new pass to do register COPY forwarding at end of register allocation.

[Quentin Colombet via Phabricator via llvm-commits]

qcolombet added a comment.

Hi Geoff,

Two questions:

1. Why is this useful?
2. If useful, why do we end up with this pattern?

For #1, I can think of giving more freedom to the post RA scheduler or eliminating the copy. If it is solely for post RA scheduling, I believe the scheduler should be able to do that. If it is to remove the copy, why do we have it in the first place?

Cheers,
-Quentin


https://reviews.llvm.org/D30751