[llvm] [AMDGPU] promote i1 arg type for admgpu_cs (PR #82971)

[Matt Arsenault via llvm-commits]

================
@@ -2078,12 +2078,219 @@ entry:
   ret void
 }
 
-; test that i1 argument for amdgpu_cs is working
-define amdgpu_cs void @amdgpu_cs_i1_arg(i1 %arg0) {
-; GCN-LABEL: amdgpu_cs_i1_arg:
-; GCN:       ; %bb.0: ; %bb
-; GCN-NEXT:    s_endpgm
-bb:
+define amdgpu_cs void @amdgpu_cs_i1(i1 %arg0) {
+; SI-LABEL: amdgpu_cs_i1:
+; SI:       ; %bb.0:
+; SI-NEXT:    s_mov_b32 s3, 0xf000
+; SI-NEXT:    s_mov_b32 s2, -1
+; SI-NEXT:    v_mov_b32_e32 v0, 0
+; SI-NEXT:    buffer_store_byte v0, off, s[0:3], 0
+; SI-NEXT:    s_endpgm
+;
+; VI-LABEL: amdgpu_cs_i1:
+; VI:       ; %bb.0:
+; VI-NEXT:    v_mov_b32_e32 v0, 0
----------------
arsenm wrote:

Brokenly storing a constant 0 

https://github.com/llvm/llvm-project/pull/82971