[PATCH] D127858: [RISCV] Turn on SeparateConstOffsetFromGEPPass for RISC-V target

[Alex Bradbury via Phabricator via llvm-commits]

asb added a comment.

My previous experience looking at whether it's worth enabling SeparateConstOffsetFromGEPPass was that there seemed to be cases where codegen was made worse and it was hard to resolve this (it was disabled for AArch64 by default in rGcd2334e86e018757b5a51eb16e5a814a6d95bded <https://reviews.llvm.org/rGcd2334e86e018757b5a51eb16e5a814a6d95bded> but D113284 <https://reviews.llvm.org/D113284> started to look at changing this).

Are you seeing any instances of worse code in any of the tests / benchmarks you're looking at? Presumably the changes in D127727 <https://reviews.llvm.org/D127727> were motivated by such cases?

Sidenote: You need to add `Scalar` to the RISCVCodeGen target in llvm/lib/Target/RISCV/CMakeLists.txt so the build still works in the case BUILD_SHARED_LIBS=True.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D127858/new/

https://reviews.llvm.org/D127858