[llvm] [RISCV] Prune dead LI in vsetvli coalescing with trivially dead vsetvli (PR #98647)

Philip Reames via llvm-commits llvm-commits at lists.llvm.org
Mon Jul 15 09:10:45 PDT 2024


preames wrote:

> I'll note that this is similar to #90629, but the latter is generalized to all dead instructions not just LIs. Do you think it's still worth picking up after this lands?

I think we need to factor out a generic isTriviallyDeadInstruction routine at MI, and then that change becomes near trivial.  

https://github.com/llvm/llvm-project/pull/98647


More information about the llvm-commits mailing list