[llvm] [AMDGPU][True16][CodeGen] Do not fold hi16 SGPR copy (PR #220858)

[Matt Arsenault via llvm-commits]

================
@@ -2215,7 +2215,8 @@ bool SIFoldOperandsImpl::tryFoldFoldableCopy(
       OpToFold.getSubReg()) {
     if (DstRC == &AMDGPU::SReg_32RegClass &&
         DstRC == MRI->getRegClass(OpToFold.getReg())) {
-      assert(OpToFold.getSubReg() == AMDGPU::lo16);
+      if (OpToFold.getSubReg() != AMDGPU::lo16)
+        return false;
----------------
arsenm wrote:

Instead of special casing this, can you check getSubClassWithSubReg or getMatchingSuperRegClass

https://github.com/llvm/llvm-project/pull/220858