[llvm] [AMDGPU] Fix for 131386 by reducing implicit definitions on register restoration (PR #133986)

[Matt Arsenault via llvm-commits]

================
@@ -0,0 +1,43 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 5
+# RUN: llc -mtriple=amdgcn--amdhsa -mcpu=gfx942 -verify-machineinstrs -run-pass prologepilog,machine-cp -o - %s | FileCheck -check-prefix=GFX942 %s
+
+--- |
+  define amdgpu_kernel void @agpr_spill_copy() #0 { ret void }
+
+  attributes #0 = { "amdgpu-num-vgpr"="32" }
----------------
arsenm wrote:

Avoid using amdgpu-num-vgpr. Use amdgpu-waves-per-eu (does this even need a budget restriction? There's no actual allocation) 

https://github.com/llvm/llvm-project/pull/133986