[llvm] [AMDGPU] Allow folding of non-subregs through REG_SEQUENCE (PR #151033)

[Matt Arsenault via llvm-commits]

================
@@ -1465,6 +1477,33 @@ void SIFoldOperandsImpl::foldOperand(
       return;
   }
 
+  if (!FoldingImmLike && OpToFold.isReg() && ST->needsAlignedVGPRs()) {
----------------
arsenm wrote:

I don't think that will work actually. I was going to suggest checking if the operand is tied, but that hack is not actually tying the operand and nothing is really ensuring that the two operands will remain the same 

https://github.com/llvm/llvm-project/pull/151033