[llvm] Rework i1->i32 zext/anyext translation (PR #114721)

[Matt Arsenault via llvm-commits]

================
@@ -23,8 +23,8 @@ define amdgpu_kernel void @sextload_i1_to_i32_trunc_cmp_eq_0(ptr addrspace(1) %o
 ; SI: v_and_b32_e32 [[TMP:v[0-9]+]], 1, [[LOAD]]
 ; SI: v_cmp_eq_u32_e32 vcc, 1, [[TMP]]{{$}}
 ; SI-NEXT: s_xor_b64 [[NEG:s\[[0-9]+:[0-9]+\]]], vcc, -1
-; SI-NEXT: v_cndmask_b32_e64 [[RESULT:v[0-9]+]], 0, 1, [[NEG]]
-; SI: buffer_store_byte [[RESULT]]
+; SI-NEXT: s_and_b64 {{s\[[0-9]+:[0-9]+\]}}, [[NEG]], exec
+; SI: buffer_store_byte
----------------
arsenm wrote:

This check is missing something, you can't go directly from this exec mask and to the store 

https://github.com/llvm/llvm-project/pull/114721