[llvm] [AMDGPU] Simplify and improve codegen for llvm.amdgcn.set.inactive (PR #107889)

[Carl Ritson via llvm-commits]

================
@@ -178,6 +177,9 @@ define amdgpu_cs void @cfg(ptr addrspace(8) inreg %tmp14, i32 %arg) {
 ; GFX9-O0-NEXT:    v_mov_b32_e32 v0, v4
 ; GFX9-O0-NEXT:    s_or_saveexec_b64 s[0:1], -1
 ; GFX9-O0-NEXT:    v_mov_b32_e32 v1, 0
+; GFX9-O0-NEXT:    s_mov_b64 exec, s[0:1]
+; GFX9-O0-NEXT:    ; implicit-def: $sgpr0_sgpr1
+; GFX9-O0-NEXT:    s_or_saveexec_b64 s[0:1], -1
----------------
perlfu wrote:

Do we even peephole at O0? Not sure this matters.

https://github.com/llvm/llvm-project/pull/107889