[llvm] [AMDGPU] Reset VGPR MSBs at the end of fallthrough basic block (PR #164901)

[Stanislav Mekhanoshin via llvm-commits]

================
@@ -480,7 +480,7 @@ body:             |
     ; GCN-NEXT: v_mov_b32_e32 v0 /*v256*/, v1
     $vgpr256 = V_MOV_B32_e32 undef $vgpr1, implicit $exec
 
-    ; No mode switch on fall through
+    ; No mode switch on fall through if this is the only predecessor
----------------
rampitec wrote:

I really think this small optimization is not worth it and I just have to remove it altogether and reset mode unconditionally. These blocks shall not occur in an optimized code.

https://github.com/llvm/llvm-project/pull/164901