[llvm] [AMDGPU] Prevent hang in SIFoldOperands (PR #82099)

[Matt Arsenault via llvm-commits]

================
@@ -0,0 +1,16 @@
+# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -run-pass=si-fold-operands -verify-machineinstrs -o - %s
+
+---
+name:            fold_reg_sequence
+body:             |
+  bb.0:
+    liveins: $vgpr0_vgpr1, $vgpr2
+    %33:sreg_32 = S_MOV_B32 0
----------------
arsenm wrote:

would be good to compact the register numbers with -run-pass=none 

https://github.com/llvm/llvm-project/pull/82099