[PATCH] D143220: [RISCV] Remove side effects from vsetvli intrinsics.

[Philip Reames via Phabricator via llvm-commits]

reames added a comment.

No objection here.  I don't have a strong opinion on how this should work through the C intrinsic interface.

If we do land this, we should revisit the implementation of the InsertVSETVLI pass.  We are deliberately very conservative there around user VSETVLIs, and this patch would remove that restriction.  This would seem to unblock some optimizations, and we can probably simplify the implementation a bit too.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D143220/new/

https://reviews.llvm.org/D143220