[llvm] [AMDGPU] s_quadmask* implicitly defines SCC (PR #161582)

[via llvm-commits]

================
@@ -0,0 +1,57 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
----------------
LU-JOHN wrote:

If I move the tests to `llvm.amdgcn.quadmask.ll ` I will need to modify the RUN lines since `-global-isel=1` produces:
```
        s_quadmask_b64 s[0:1], s[0:1]
        s_and_b32 s4, s4, 1
        v_mov_b32_e32 v0, s0
        s_cmp_eq_u32 s4, 0
```
but `-global-isel=0` produces:
```
        s_and_b32 s4, s6, 1
        s_quadmask_b64 s[0:1], s[0:1]
        s_cmp_eq_u32 s4, 0
```
Is this okay?


https://github.com/llvm/llvm-project/pull/161582