[llvm] [AMDGPU] Fix si-optimize-exec-masking stepping into debug values (PR #201947)

[Matt Arsenault via llvm-commits]

================
@@ -0,0 +1,121 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -mtriple=amdgcn--amdhsa -mcpu=gfx1030 -run-pass=si-optimize-exec-masking -o - %s | FileCheck %s
----------------
arsenm wrote:

Can you add a debugify line to an existing test for this pass?

https://github.com/llvm/llvm-project/pull/201947