[llvm] [AMDGPU] Teach SILateBranchLowering pass to preserve MachineLoopInfo (PR #178276)

[Matt Arsenault via llvm-commits]

================
@@ -0,0 +1,37 @@
+# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -passes="si-late-branch-lowering,print<machine-loops>" -filetype=null %s 2>&1 | FileCheck %s
----------------
arsenm wrote:

This test doesn't show anything. The print would require the analysis to be recomputed anyway. This probably passes without this patch 

https://github.com/llvm/llvm-project/pull/178276