[PATCH] D130537: AMDGPU: Take care of "tied" operand when removeOperand

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added a comment.

I'd also like a mir test



================
Comment at: llvm/test/CodeGen/AMDGPU/frame-index-elimination-tied-operand.ll:2
+; RUN: llc -march=amdgcn -mcpu=gfx1100 < %s | FileCheck --check-prefix=GCN %s
+
+define amdgpu_kernel void @tied_operand_test() local_unnamed_addr {
----------------
Can you merge this test with one of the existing frame elimination tests?


================
Comment at: llvm/test/CodeGen/AMDGPU/frame-index-elimination-tied-operand.ll:12-13
+  %scratch = alloca i16, align 4, addrspace(5)
+  %cond = select i1 undef, i16 addrspace(5)* %scratch, i16 addrspace(5)* undef
+  %spec.select = select i1 undef, i16 addrspace(5)* %cond, i16 addrspace(5)* %scratch
+  %dead.load = load i16, i16 addrspace(5)* %spec.select, align 2
----------------
Avoid using undefs in this test


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D130537/new/

https://reviews.llvm.org/D130537