[llvm] AMDGPU coverity fixes (PR #182013)

[Matt Arsenault via llvm-commits]

================
@@ -18253,7 +18253,7 @@ SITargetLowering::getRegForInlineAsmConstraint(const TargetRegisterInfo *TRI_,
       // Check for lossy scalar/vector conversions.
       if (VT.isVector() && VT.getSizeInBits() != 32)
         return std::pair(0U, nullptr);
-      if (Idx < RC->getNumRegs())
+      if (RC && Idx < RC->getNumRegs())
----------------
arsenm wrote:

probably should test this 

https://github.com/llvm/llvm-project/pull/182013